Marketing Ops Community

Head of Training and Certification

Indeed.com

Director of Marketing Strategy and Operations

MarTech Alliance

I am the lead instructor for a brand new 8 week course on all things Marketing Operations! Visit Martech Alliance online for more details.

Course Instructor

Amazon Web Services (AWS)

Global Marketing Operations Leader

Hitwise

Leads the marketing and demand generation teams in the Americas and APAC. 

- Oversee the planning and direction of marketing programs
- Partner and support sales to execute strategy for winning clients and upselling existing clients
- Report on key business metrics, campaign results, and drive optimization
- Development of Marketing ROI reporting and analytics 
- Lead nurturing, drip campaigns, lead scoring, sales and marketing alignment
- Oversees sponsorships, paid advertising, and online media spend
- Development and execution of campaigns across paid search and social, online advertising, webinars, events and content marketing
- Marketing automation and Salesforce implementation

Director, Global Marketing

American Marketing Association

Leads content marketing and communications strategy across all channels. 

Oversees communications and social media teams.

Manages writers and content producers for blogs, eBooks, and advertisements.

Director Of Communications

Leaf Group

Responsible for marketing and demand generation programs including marketing automation, lead nurturing, email marketing, messaging, sales enablement, and alignment across paid, content marketing, and event marketing programs. 

Leads and manages team of demand generation professionals.

Passionately implementing unified Marketo, Salesforce, Facebook Advertising and content marketing B2B programs. Marketo Certified Expert.

- Database Hygiene 
- Database Migration 
- De-duplication 
- Data Standardization and Data Normalization

Senior Marketing Manager

Trumpia

Drove 500% growth of new marketing platform through partner marketing and relentless adoption programs for AT&T Mobility’s 1,500+ business sales force. 

Established webinar lead generation program accounting for over 80% of marketing qualified leads 
Instrumental in product marketing, pricing, packaging, business development and go-to-market strategies. 

Online marketing including Google analytics, website and landing page optimization, marketing automation using Pardot, email marketing, collateral development and copywriting. 

Led event marketing team to successfully execute AT&T new product roadshow and events in 15 major US markets.

Digital Marketing 
Marketing Automation 
Google Analytics and Adwords 
Product Marketing
Customer Marketing 
Channel Marketing

Marketing Manager

Independent

Small business marketing consulting including: 

Digital Marketing 
Email Marketing 
Marketing Automation 
Social Media Marketing 
Online Marketing 
Website Design 
SEO/SEM

Marketing Consultant

Director of Marketing Strategy & Operations | Martech Leader | Speaker

Author and speaker. Co-host of the Humans of Martech Podcast.

Subscribe to the Marketing Operations Leader newsletter for MOPs strategies and insights every week: https://darrellalfonso.substack.com/

Highlights: 
- Ranked as one of the Top Marketers in the US by Propolis 2022
- Named one of the "Top Martech Marketers to Follow" 2020 by Martech Alliance 
- Keynote speaker for Martech Conference and Get Stacked B2B Marketing Conference 2021
- Awarded Fearless Marketer 2018 
- Marketo Champion 2018 and 2019 
- VP of Communications American Marketing Association - 2017-2019
- Marketo User Group Leader 2017-2019
- Leading Program Manager for Connect@Amazon (35,000 member Amazon organization)
- Frequent Martech Speaker and Author 
- Marketo and Salesforce Certified 

Posts are my own and do not reflect the views of my current or former employers.

Darrell Alfonso

Sorry late notice everyone! 

We have an amazing marketing ops town hall where we'll dive into practical AI use cases for marketing ops. 

We'll have Lindsay Rothlisberger, Jomar Ebalida, Mike Rizzo and myself cover some awesome use cases, and we'll also open it up to the group to share what they've learned. 

See you all there, link in the comments to sign up to join us for free.

#marketing #martech #marketingoperations #ai

aztela

We bridge the gap between Business and Data to transform your data into actionable insights.  Here are some results how we helped companies decrease churn, increase profits just by using their data  ↓

Founder & CEO

Savvy Social

Alignment of GTM departments for a +20% uplift in profitability, and decrease in cost due to data-driven initiatives. EMEA hit the highest month in terms of MRR. We hit over +278K in MRR in Summer which should be the toughest period. 

Projects 

- Building Revenue Operations in the Sales team for regional teams in EMEA, North America, and Asia
- Accurate reporting  across every touchpoint through the funnel
-  Automate & Operationalize  lead lifecycle for frictionless processes
-  Dashboard for sales, executives, and the marketing team
-  Automation, workflows, and operation for sales teams and fulfillment sites across the organization
- Sales projection, forecasting, and data analytics** for EMEA, NA, and Asia regions
- Aligning departments (marketing, sales, and CS)
- Implementing CRMs and integrating with other GTM tech tool

Head of Analytics

HYPCCCYCL

Member
HYPCCCYCL is an invite-only, professional community where top B2B revenue leaders cross-train their GTM skills with VCs.

Brand Ambassador

RevGenius

Member

Astela

< Open the vault >
< put quality in your pipeline >
< Revenue engine that lasts forever NewEra of GTM >

We are like your Outsourced CRO aka Revenue engine. 

•Market penetration
•GoToMarket strategy 
•Fastest way to Impact 
•Fixing the old way of doing outbound with  TheVault (Outbound2.0.0) which is customer centric overall aligning bot revenue engines sales and marketing. 

We created that demand so outbound feels like inbound. 

Revenue focused > Meetings.

Co-Founder

Spendesk

Account Executive

- Started seeing results after 2-3 months.
- Been for few months as full cycle rep. 
- 980k ARR sourced
- Shift happened after seeing that volume was not gone help 
- Started to pursue quality in targeting and picking accounts, personas & ICP. 
- As it was early, the company also went in marketing to understand things that could help me. 
- Been involved a lot with the marketing and CS team all the time. Also working closely with the manager.

Senior Sales Representative

Etiket Agency

- Establishing their and runing their paid advertisement performance campaigns for B2B clients.
- Managing their campaign budgets for B2B clients. 
- Running SEO and FB campaigns mostly. 
- Running marketing team full of 14 people at the end started only with 2.

- Researching for clients campaigns. 
- Managing CRM. 
- Working closely with marketing and sales and CEO. 
- Admin work.

Sales And Marketing Specialist

Head of RevOps | Recently helped SaaS organisation 19% faster growth, +10-20% sales productivity, and 30% lower GTM costs with Revenue Operations

My Name is Ali and I love building things. Whether that be data solutions, my business, or helping other organizations at the end what I'm doing is building myself into the person I want to become. 

Our thesis is that data is the lifeblood of any organization and will only work if it's made custom to your goals, needs and priorities at any stage of your growth.

Unfortunately, many businesses don't have that as they 

-  Rely on off-the-shelf solutions that are not custom to their needs and don't adapt 

- Their data is all over the place resulting in not able to make any data driven decisions.

- Can't rely on their data to make rational logical decisions. 

- They track surface-level KPIs and not actionable insights that would give them a competitive advantage 

- They don't know how to bridge the gap between business and data expertise.

Well at aztela we are that bridge that connects business and data together to develop a moat inside of your business resulting in: 

- One of our clients generated over 20x ROI since starting working with us resulting in over $10m in revenue attributed 

- Increasing profits due to seeing everything at one place which would normally take you 100h a month

- Decreased churn and increased LTV able to see bottlenecks, and root cause as opposed then spinning in circles 

- Fixing inefficient operations due to removing data silos 

We do that by utilizing data you already have and turning it into actionable insights. 

We implement custom analytics dashboards (BI) and advanced solutions to achieve business results just by utilizing data you already have.  

If you want to create a moat, competitive edege in your market click on this link to speak to our team- > 
https://calendly.com/alizu/15-minute-meeting-aztela?month=2024-04I

Ali Z

If your new Head of Data spends their first 90 days building dashboards, you didn't hire a leader. You hired an expensive analyst. 

Most Heads of Data fail
They start fixing pipelines, rebuilding dashboards, and sketching out 6-month strategies.

A company finally hires a Head of Data.

You inherit a half-working warehouse, dashboards no one trusts, and inbox full of 
“Can you pull this"? 

You audit pipelines. You clean schemas. You plan migrations.

By next 45-days CEO asks: “Why we still in same mess?”

Because you fixed systems but not trust or adoption.
You added complexity but - not impact. 

Most Heads of Data or promoted CIO/CTO or IT are thrown in 'data choas'

But here the 90-Day Playbook 
This is what actually builds trust, adoption and sets your data as growth edge. 

Phase 1: Diagnose, Don’t Deliver

Your job isn’t to fix dashboards, buy Databricks and say "we data-driven".
It’s to understand where data actually dies inside the business.

Meet every key stakeholder 

Ask: 
"Biggest priorities, and challenges" 
“What data do you trust?”

Asses the current data enviroment and overall complexity. 

When first started, I met with everyone. 
Like everyone. It didn't matter if VP or manager in operations floor.

In few weeks I had more info then people who were there for 5-years. 
+Would find extreme champion "data superusers" in it. 

Phase 2: Align and Design

Priortization is everything at that stage. 

Identify quick wins you can deliver, based on the current landscape, prioritize accordingly so each use-case serves the next one. 

Assess based on 
- Impact 
- Complexity
- Reach  use by many departments) 
- Risk 

Define metric ownership. Start communicating weekly.

Phase 3: Deliver Proof, Not Promises

My last post I mentioned about endless AI and data science projects and how they fail. 

As head of data, your job is to make data work and produce impact, trust and adoption. 

But one key factor is getting to feedback as fast. 

CFO won't care if your ML model is 92% if it doesn't make money. 
Same as VP won't care if your data accuracy is 98%, if he don't trust metrics. 

Deliver the highest priority. 
Align on definitions. 
Deliver first win under 30-45 days
Overcommunicate wins and issues.

What to avoid 

Don’t buy new tools in your first month.
Don’t rebuild dashboards before fixing foundations.
Don't fall into an "ad-hoc" death cycle with endless no ROI projects. 
Don’t promise 'AI' next week while having 10 definitions of revenue. 

The first 90 days decide if your data gone be growth or reporting function. 
You’re not being judged on architecture, complexity or new tools. 

If your CFO still doesn’t believe the data by Day 90, it’s game over.

Fix trust first.
Deliver visible wins.
Communicate relentlessly

That’s how you stop being “the data person” and start being the person who made data work.

If you’re stepping into that role (or just did), I’ll share the exact 90-Day Playbook. 
Drop a “90” below

Ali Šifrar

CEO @ aztela | Fractional data leader helping scaleups+ mid-market, gain control, value from their data to accelerate their data, digital & ai journey.

You can have 99% data quality and still have zero trust. A 3-year-old spreadsheet is more trusted than your $1M data warehouse. Here is how to fix it.

Your executives don’t trust the data, dashboard, even if they agree on all definitions. 

You can see it in every meeting:
-The CFO cross-checks the numbers in Excel.
-The COO asks, “Where did this data come from?”
-Sales says, “Our CRM doesn’t show that.”
-The CEO quietly stops using it altogether.

And your data team? They swear the pipelines are fine.

But VPs don't care if it's 99% data quality if they don't trust the data. 
So if the data’s “right,” why doesn’t anyone believe it?

At one huge industrial giant, data could have easily become chaos.
But it didn’t. 

With their Head of Data & AI Ops, they built a self-service data and AI platform that serves thousands globally. 

And it wasn’t about fancy tools.
It was about consistency and ownership

They didn’t “fix dashboards” or "roll out data quality tool" 
They focused on adoption, trust, and feedback. 

Here are some fixes to increase trust and overall data quality

1. Trust ≠ Accuracy

You might agree on metrics, aligned to impact, and assign owner 
But 2-weeks past and here VP say "I don't trust this data"  

Nobody cares if your definitions, dbt logic or pipeline is 99% accurate. 
Trust only happens when user sees the value. 

To make that you need to uncover old sheets that might be hidding somewhere.

Itreate based on feedback from end users or data superuser. 

2. Assign Ownership 

Every metric, data product needs a business owner.
Not only IT. 
A business end user. 

You need this as this makes data everyone's responsibility for quality. 
If not, it's going to turn into pointing figures chaos.

3. Showcase Lineage 

Executives lose trust the moment numbers shift without explanation.
You need lineage. 
Executives don’t need perfect data; they need visible data quality.

Everyone in their mind is comparing data to some old spreadsheet
Showcase where it's from in simple, clear terms

4. Unify x Semantics 

If 3 of your systems have revenue differently something is going to be off. 
Fix it by creating a semantic layer, one place where all metrics and logic live.

Every data product is built on the layer that everyone agrees on. 
“Gross Margin” meant the same thing in every tool. 

5. Proactive vs Reactive

Most trust issues arise due to a bad process.

Most teams check data, the pipeline shows no errors, everything is fine and accurate. 
But right when VP looks, tries to use data, something is off. 

If most of the problems you get after they happen, nobody is going to trust that data. 
Even if it was right 99% time, that 1% wasn't. 

Make sure to set up: 
Automation handled data provisioning, lineage, and consistency checks across systems.
That meant quality wasn’t a manual process; it was enforced through architecture.

JSESSIONID

li_at

John Bucknavage

Your data problem didn't start in your warehouse. It started in that free-text 'Region' in your ERP. Spending $1M modernizing your stack, won't fix your data. 

Everyone wants accurate data.
But when you dig you realize their processes were never built to produce good data.

They’re trying to analyze chaos.

A few months ago, we were talking to finance company. 
They’d just spent 14 months modernizing their stack.
They hired the data engineers.
Millions spent.
Hundreds of dashboards.

And yet:
“Revenue” in Salesforce included refunds.
“Customer” in Marketing meant prospects too.
Operations had 15 different “regions” spelled 8 different ways.

The tech wasn’t broken.
The process was.

Their CRM, ERP, and sales systems were designed for convenience, not for data.

Every time a sales rep skips a CRM field..
You create a leak in your data foundation.
Until your warehouse is garbage. 

If your processes weren’t designed with data in mind nothing will save you  

Here is how to go about stopping bad data

1. Design Every Process as if Data Were the End Goal

If you’re setting up a CRM, ERP, or even a Google Form, build it like a data engineer would.

Even if it's yet. Develop a process with data in mind. 
As down the line, you will need, and rather than waiting 3-5 months to get data. 

Replace free-text fields with controlled dropdowns.
Enforce mandatory fields that align with business-critical metrics.

Executives say they want clean data but approve workflows that guarantee mess.

In my opinion, data should be clean from the source. 
Becuse if it's not, managing pipelines, modelling becomes a nightmare. 
And even that can't save it 

2. Treat Metrics Like Products 

Agreeing on definitions is not easy at all. 
People change, leave. 
2 VPs can't agree on it so they create their own spreadsheet. 

Every metric you report on should have an owner, version history, use case and single definition across the company.

If found in a situation can't agree, ask "What finding this info enables you" 
If can't answer it, archive it. 
Or if can't agree on metric. 
Seperate and define clear use case where each. 

3. Asssign Owner & Build Feedback Loops

Bad data comes from the frontlines, reps skipping CRM fields, creating custom objects in Salesforce. 

Assign owners of the metrics.

Answer: 
Who owns the data?
Who manages the inputs?
Who's keeping operational systems clean? (Data stewards)

If no one is accountable or owns it, how do you thing it will get fixed.
Tie accuracy to incentives. 

4. Enforce Standards, Not Opinions 

Everyone uses their own definition of “good data” 
Define how data should look: formats, naming, validation rules.

If “Region” is free-text in CRM, you’ve built chaos by design. 

5. Data quality isn’t a project or a one-time thing  

Start where it's most important. 
Track exceptions, expose results, fix patterns.
Embed it in the system, so it's proactive rather than reactive.

Your competitor made >$10M with AI predictive model. But your data team delivered 221 dashboards in 12-months. Your team not slow, just broken.

Been hearing this for the past weeks 
“Our data team is so slow.”

Iterations, requests take weeks.
Priorities keep shifting.
Everyone’s busy, but nothing moves.

Let’s be clear,  that’s not a talent issue.
It’s a structural failure.

Your data team isn’t drowning in work.
They’re drowning in dependencies, misaligned ownership, and bad incentives.

Mistake #1: You Built by Function, Not by Outcome

Most teams look like this:

Engineers ingest data
Analytics engineers model it
BI developers visualize it
Analysts most times nobody knows what they doing.

Looks efficient on paper.
In reality? It’s a garbage.

Each role optimizes for its piece, not the outcome.
The engineer ships a pipeline which already made. 
The analyst ships a dashboard which usless.
Nobody owns the impact that dashboard is meant to enable.

That’s why velocity dies.

Fix:
Organize around data products, not job titles.
Each data product (e.g., revenue insights, supply chain metrics, marketing attribution) should have cross-functional memeber that owns it end-to-end.

When the same squad owns ingestion → modeling → delivery, you eliminate 80% of the handoffs that kill speed.

Mistake #2: You Measure Output Instead of Impact

You’re tracking the wrong KPIs.

Executives love vanity metrics:
“How many dashboards did we build?”
“How many pipelines went live to prod?”

That’s noise.
What you should measure is throughput,  how fast the business goes from question, problem → trusted data → impact

If your don't have clera defintions of done, ETA and ROI. You doing nothing. 

Fix:
Velocity isn’t how fast engineers code,  it’s how fast the business gets results.

Mistake #3: You Built a Platform Without an Operating Model

Every company says they have a “data platform.”
Few have a data operating model, a clear system for ownership, collaboration, and accountability.

So you get chaos:

Five teams writing pipelines differently
Conflicting metrics definitions
Endless meetings to resolve “which number is right”
That’s not a tooling problem. It’s an org-design problem.

Fix:
Define how your platform and data teams interact.
Who owns ingestion? Who owns definitions? Who approves changes?
And for every dataset, model, and 'data product', name an owner.

Velocity doesn’t come from adding tools.
It comes from removing friction.

Your team isn’t slow because they lack skills.
They’re slow because you built an org that rewards activity over impact.

You don’t fix that with more headcount or another vendor.
You fix it by rewiring focusing on impact, trust and adoption.

One team, one outcome, 90-day delivery cycles.

If your team’s buried under tickets and “urgent requests,” it’s not a workload problem, it’s an operating model problem.

🧬 Repost if you think most data teams creating bureaucracy.

Everyone trying to 'become AI-driven.' But less than 10% of you actually are. You can't even trace where your data is coming from or what is difference between 2 metrics. 

Let’s call it what it is:
Most AI or ML initaitives fail because your data isn’t ready.

When we walk into mid-market organizations we see the same pattern every time:

Sales, finance, and ops each define metrics differently.
BI and AI pull from different data sources.
Data pipelines are built in silos, no lineage, no ownership.
80% of the data is still trapped in PDFs, or spreadsheets.
AI pilots stall in POCs 

Few months ago one firm came to use after burning through $400,000 on “AI enablement.” 

When they tried to build AI use cases (recommendation engines), the models failed.

Their models couldn’t differentiate between 'accounting-adjusted revenue' and 'store-reported sales'. There was no business meaning attached. 

Here’s the Data That Should Scare Every CIO

Over 55% of companies cite inconsistent data architecture as the #1 blocker to AI readiness. But only 14% invest in Data Strategy, the part that makes all investment pay off. 

Here is how to do it 

1. Impact over complexity  

Start with use cases, data products, not ML or AI. 
Don't due AI for sake of it. Don't skip foundational work.
Most use cases I saw are hidden logical statments.

Data products deliver business outcomes. 
Then design data products, curated, governed datasets aligned to those outcomes.

The smartest organizations aren’t moving data,  they’re federating it.

They build logical data products that query where data already lives, wrapped in a semantic consumption layer that abstracts complexity from the business. 

2. Semantic & Consumption layer 

A semantic layer starts by translating business concepts into reusable logic.

Instead of 20 “revenue” tables with different definitions, you get one trusted “Revenue” concept everyone agrees on defined once, reused everywhere. 

3. Add Business Context, Metadata & Lineage

Don't just add metadata. Add meaning. 
If not, AI will just assume. 

Column names like rev_net_v3_final don’t help at all. 
You need business metadata, written in the language your teams use.

You need business metadata, written in the language your teams use. 

Answer: What does each data point mean in business terms?  Where from? Where used etc?

This is not only for AI but also for trust and adoption. 

4. Don't need 99% data quality but 

The companies that are winning treat AI readiness like data hygiene.

Data doesn’t stay clean.
Set process ensuring data accurate in upstream. 
Freshness checks ensure data timeliness. 

5. Simplify to Scale.

Complexity kills agility.
If your data architecture diagram looks like a Gartner Expo, it’s already broken.
Every component should have a measurable purpose or be removed.

Start fixing the foundation that actually makes it work.
AI Readiness is a byproduct of good data enviroment.

You hired $200k engineers to clean your data. But a 3-year-old spreadsheet is more trusted than your $1M data warehouse. Here is why 99% accuracy doesn't matter 

“We can’t trust the data.” said every 'data-driven' org, so they

Buy another observability tool.
Hires two more data engineers.
Runs another “data cleaning” project.
Trys to change the logic in the warehouse. 

Six months later?

MD notices 2x higher revenue before the meeting.
Dashboards don’t match finance.
VPs trust his 4-year-old sheet more than the 1M warehouse. 

Here’s the truth 
80% of data problems have nothing to do with pipelines, tools, or code.

They start upstream, in the way your business creates the data.

We worked with a cement manufacturer running plants in multiple.
One plant was 100 years old. The other was new.

Different equipment, processes.

Everyone recorded data differently.
One used a shared Excel file.
Another logged reading in a legacy system.
A third called maintenance over the phone, logged in free text field in ERP.

By the time that data reached the warehouse, already broken.
Mismatched units, timestamps, duplicate entries.

The failure happened long before ingestion.

Most data quality mistakes are people, process, and trust issues. 

Most often that we see are: 
-No validation because “it looks fine.” 
-Treating quality standards as static.
-Ownership of IT.
-Logic scattered from BI to semantics.
-Long data lags not up to date. 

Data always fails, silently but that's to late.

Here’s How You Actually Fix Bad Data Forever

1. Map Where Data Is Born

Where does each data originate

Who enters it?
In which system?
How often?
You’ll find that half of it starts in Excel or free text fields in CRM.
Start where there is the highest impact, like regulatory etc. 

2. Standardize the Process.

Map out the process, ensuring that bad data is not possible to be entered. 

Standardize the process:
Define what, when, and how data is captured.
Document it.
Enforce it across.

Having a free text field in CRM is perfect for bad data, replace it.

3. Validate at the Source

Don’t wait until Snowflake to check for errors.
Build validation rules where the data starts. 
Catch mistakes before they hit the warehouse.

4. Assign Real Ownership

Every dataset, metric needs two people:
Owner: A business person who cares about accuracy.
Steward: A person who ensures reliability.

If no one owns it, it won’t get fixed.

5. Don't forget trust 

Accurcy ≠ Trust. 
Seen VPs trust old Excel file more than $1m warehouse. 
Most treat DQ as reactive play, fix it when it fails
But by then, trust is broken

Need to be always getting feedback from end users for trust.

Tip: Know important dates. This ensures specific QA done prior important meetings

Most data issues are reactive, when VP presents 3x higher revenue on board meeting.

Comment "DQ" and I'll sent you a detailed playbook on how to ensure to stop the bad data that results in trust and adoption in weeks

Until you can't agree on what a 'new_customer' means, you’re not ready for AI. You're not even ready for data. Here is what to do.    

Every exec wants better data.
Another warehouse.  
Another “AI-ready” stack.

Six months later, the story’s the same:
– Dashboards that don’t match
– Execs arguing over numbers
– Data team firefighting definitions instead of building

The problem isn’t technology.
It’s not a “data quality” issue.
It’s a semantic alignment issue.

When we ask a simple question,  “What does revenue mean here?”, the room goes silent.

Finance says “booked.”
Sales says “pipeline.”
Ops says “fulfilled.”

Three teams.Three definitions.
One guaranteed way to burn your data budget before you even start.

Here’s what nobody tells you about Data and “AI readiness”

Data doesn’t fail because your engineers are bad or need another tool.
It fails because your data doesn’t speak the same language.

If your business can’t agree on what a “customer", actually means, then everything gone fail. 

A huge finance corporation figured this out and here before leader in the space.
They unified all the core data, market, risk, and transactions into one shared semantic layer.

So when someone said “return,” it meant the same thing in every tool, system or model.

That’s why they could evolve data into intelligence. 
That's why they got trust, adoption and agility. 

That’s what made their data usable, scalable, and productizable.

Not because of tools.
But because of alignment. 

Here’s how to align on definitions

1. Map the Chaos.

Start with the data that causes the most friction: revenue.
Audit and gone see half of them are all over the place.  

You’ll find at least three, and that’s where the mistrust begins.

2. Define & Agree 

Define each metric. 

It should explain:
Tehnical and Business definition 
Where it lives
Who owns it
Use Case 
Products and more 

Create data dictionary dynamically, include version control as things change fast.

If can't agree on metrics, or data, make sure to assign a clear use case. 
One used for ML model in operations, other in board meetings. 

3. Assign ownerships 

Data ownership often dies in IT. That’s the mistake.
Every critical metric should have a business owner who’s accountable for its definition, accuracy. 

That way you get buy-in, as without it everyone is pointing fingers. 

3. Codify It in Your Data Model.

Once definitions are locked, codify them:
In your semantic layer
In your sources 
In documentation 

That’s the foundation of a data platform that’s trusted. 

4. Run Trust Loops Weekly.

Get feedback weekly from business users:
“What feels off?” “What’s inconsistent?” “Where are the anomalies?”
That loop is where adoption and trust, are built. 

Things change a lot.

5. Productize It.

Once the core language is established, you can package it as data products.
Each product = a trusted, documented dataset serving a clear business purpose.

Your $1M 'modern data warehouse' is world's most expensive junkyard. Tracking 2TBs of usless data 'just in case' but still takes 2-weeks to answer simple question.

Every company says the same thing:
“Let’s capture everything, for down the line”

That sentence has quietly destroyed more ROI, Snowflake credits, and data trust than any bad tool ever has.

A few years ago, I met a retail data leader that proudly told me,
“We’re tracking everything. Every click, scroll, hover, API event.”

A billion events a day.
Storage bills through the roof.
And still, the FP&A didn’t trust a single revenue number. 
Irony right

Analysts couldn’t even tell if someone added or removed an item from a cart because every event stored the entire basket state.

So the engineers were drowning in data.
The business was starving for insight.
And leadership thought the solution was… to collect even more for 'in case'.

That’s not data maturity. That’s data garbage.

Most mid-market firms believe more data = more answers.
But that logic is backwards.

When you collect everything:

Your warehouse becomes a landfill.
Pipelines grow, meaning more debugging.
Your team spends 80% of time cleaning, 20% analyzing.
Governance dies because no one knows what matters.
Every metric becomes a debate, not a decision.

“Collect everything” doesn’t make you smart, it makes you slow, expensive, and untrusted.

Here's how to fix this

1. Start with the business question, not the data source

Before tracking another event, ask: “What question are we trying to answer and who needs it?”

If it doesn’t tie to a decision, action that enables the end user to achive their objective 
Don’t collect it.

Data only valuable if used, storing it it's worth $0. 

2. Move from data hoarding to event intent.

Stop dumping entire states  
Start tracking facts, atomic events one line that captures who, what, when, and why.

They represent a business action or customer intent.

For example:
- Don't “Full basket snapshot” → 200 useless fields.
- Do “Product_Added_to_Cart” → who, what, when, how much.

That’s where clarity starts.  
This is gone build foundation for it as AI needs context, not just more data.

3. Build a tracking plan that acts like a product spec.

Define which events are allowed, what’s mandatory, and who owns them.
If it’s not in the plan, it doesn’t get into your warehouse.

Most warehouse look 'modern' outside but 'legacy' inside. 
You didn't fix nothing, just made it worse. 

4. Kill the “just in case” mindset.

You don’t need another terabyte of noise.
You need confidence in the 1% of data that drives 99% of value.

One org had TBs of data, queries that cost  > $5,000 month, that they didn't need at all. 

If it takes three department, two weeks to answer “What’s our active customer count?” 

You’re not data-driven. 

Companies winning with AI and analytics aren’t hoarding petabytes.
They’re the ones who know exactly which 1% of their data actually matters.

Data Architecture is worthless. You just spent 12 months fighting with pipelines and $500k building data platform. But then VP announced move to HubSpot, and now all your work is useless. 

Most “modern data platforms” break the moment your business changes.

You spend 12 months modernizing, Snowflake, dbt, BI, governance, the works.
You spent hours on meetings aligning on definitions. 
You finally set processes in CRM so data gets accurate in the source. 

But then suddenly…
→ The CEO changes the pricing model.
→ VP says "We migrating to Hubspot" after you spent last 5-months combating with existing sources. 

Suddenly, everything you worked on became usless 

Half your dashboards are wrong.
Data models break.
Pipelines you built are usless.
New source + process needs to be redone.

You didn’t build a data architecture right way. 

Most companies build static systems in dynamic businesses.
You’ve got cloud storage and transformation pipelines…

But every change request still takes 6 weeks.
Every definition debate lasts 5 meetings.

You are communicating throught some 3rd persona that all value gets lost through translation. 

Building 'data architerture' in itself is not valuable. 
That's step 0. 
Step 1 is actually using it and generating ROI. 

Best data leaders and teams are building adaptive data architectures and evolve with the business. Not the other way around. 


Here is how to build data architerture that ensures trust, agility and long term ROI.

1. Map business capabilities, not systems

Forget the org chart,  map what your business actually does for each business area. 
From finance, sales and so on how do they work. 

This becomes the foundation of your data platform design.

2. Architect for change, destruction, not perfection.

Keep architecture modular.

Data Integration: Connect and standardise data
Data Layer: Raw → curated → business model
Smart Data Services: Reusable data products and APIs
Search & Consume: Self-service tools and governed access 
.
You should be able to change and add without breaking or rebuilding. 

3. Build data products, not pipelines.

Each product serves multiple teams,  with clear ownership and measurable value.

E.g., “Customer Revenue Product” used by Sales + Finance + Marketing = one truth, three users. 

Everyone can accsess lowest hanging data, aligned to their objectives, which allows agility, but remains control. 

4. Make governance adaptive

Governance that moves at the pace of business,  flexible, iterative, and owned by both data and business.

Governance = light guardrails, clear ownership, fast iteration

5. Build feedback loops

Your architerture should adapt to your business. 
Attack lowest hanging fruits, only build according to value. 

Identify a business outcome
Deliver a small, testable solution
Measure impact
Iterate

When you build adaptively, everything accelerates

Your business changes every quarter.
Your data architecture should adapt, scale with you

You promised to build source of truth, add predictive, dashboard annoying VP but in 12 months got 313 broken dashboards. If your Head of Data says "Yes" to every request, fire them

I can walk into any $20M, $1B org and I’ll find:

-89 dashboards nobody trusts.
-A “modern stack” that costs more than it saves.
-A data team is exhausted from chasing every request.
-CEO asking, “Why we still didn't fix this data mess?”

Most fail because leaders try to fix everything at once.

“Let’s rebuild the warehouse. Add AI. Roll out dashboards. VPs screaming "We need this...." 

When everything’s a priority, nothing is.

Most act as 'ad-hoc' service desks like. 
Trying to please everyone, say YES to all usless requests. 

After 6-months you spent 80% building usless things, that don't solve the problem.

You can’t modernize your architecture, add AI all at once.

12-month 
They care about whether something finally works and gets results.
And fast. 

Here is how to balance and prioritize accordingly 

1. Create clear priortization criteria

Revisit the business pain points identified in your earlier assessment

Pick one high-visibility, high-value problem 

Define Clear Scoring Criteria

Evaluate every initiative using these dimensions:
Business Value - Quantify the ROI or risk reduction.
Complexity - How mature is your current data environment?
Urgency / Risk - What happens if we don’t solve this?
Frequency - Are these data products used recurring daily or quarterly?
Reach - How many departments or users used by?
Time-to-Value - How fast can we deliver?

Additional Factors:

Staffing & Availability 
Budget & Capital Constraints 
Competing Current Projects
Upcoming Milestone 

You can use AI to help but need to feed a lot of context

2. Map it out, Rank, Communicate, and Confirm

Now score rank each use case
Validate this with your executive sponsor:

If it doesn’t meet the criteria, it doesn’t make the cut.

High impact, low complexity → Do now
High impact, high complexity → Plan next
Low/low → Park it
Low/high → Drop it 

3. Create a “Stop Doing” list.

Think in terms of data products.
If the 'quick thing' that VP asked doesn't have an end user, outcome, timeline, current way, definition of done assigned, don't do it. 

Thats why your data strategy or priortization roadmap needs to be reflected monthly. 

At the same time, you don't need to plan 12 months, Focus on now 3-6 months MAX

Things change fast 

4. Sequence, don’t scale

Your first few wins buy trust and set foundation for future strategic initiatives.

For example 
Start with a data product where the finance team spends 50+ hours manually reconciling data. Fixing this builds your pipelines, and a semantic layer,  this create foundation for next initiatve

For example can't do 'predictive' prior having a foundation set. Start with quick wins that enable that

→Comment "Matrix," and I'll send you a Priortization Matrix Playbook on how to balance short vs long term and setup data as growth function.

You cloud migration failed becuse you didn't build a 'modern data warehouse.' You built the world's most expensive junkyard. 

Your migration didn’t fail because of Snowflake or Databricks.
It failed because you tried to move too early, too fast, or without proving value.

You picked the wrong timing, treated it like IT initaitve and treated as a goal

Seen this all the time

The CIO announces a “cloud-first data strategy.”
$2M+ budget approved.
Consultants everywhere.
18-month migration roadmap.

Two years later, Finance is still in Excel.
Operations still doesn’t trust dashboards.
The business still runs on exports and emails.
200-redunded pipelines running into cloud (10% actually used) 
Over 1TB of usless data that shouldn't be there in first place. 

The stack changed.
But you just moved shit data, to the cloud. 

Here’s what most teams get wrong (and how to fix it):

1. You moved everything at once.

You don’t need to migrate 100% of your data.
You need to migrate the 20% that drives 80% of business decisions.

You gone see that 80% of parts you don't need, and are usless. Start

Fix: Start with one high-value use case,  financial close, revenue cycle, or ops reporting. Prove value before moving anything else.

2. You treated the migration like a sprint, not a relay.

You can’t “big-bang” your way to trust.

Most migrations lose sponsorship halfway because they take too long to show results.

Most migrate data but nobody 
They fund impact.

Fix: Deliver one visible business impact every 45-90 days.
Show speed-to-value, think in terms of data products focuson on trust, adoption and ROI. 

3. You sold the end state, not the next milestone.

Your roadmap said: “100% cloud by 2026.” 
Terrible strategy.
No one funds a vision they won’t live to see.

Cloud isn’t the destination or outcome,  the only thing it matters what you do with data. 

 If your first conversation starts with tools or roadmap instead of business problems, it’s already dead. 

Fix: Reframe the migration as a sequence of self-contained 90-day wins.
Each one should earn credibility and fund the next phase.

4. You forgot that migration ≠ adoption.

The new data warehouse is live,  but if business users still export to Excel, OR don't trust your data is usless.

Fix: Make adoption part of your migration plan.
Enable self-service by standardize foundation, but keeping outside agile. 
Celebrate business wins, not engineering milestones.

5. Avoid Over-engineering.

Don't add tools, until you have trust, adoption and semantic layer completed. 
Organization end up with 10 tools, real time pipelines but can't trust their data. 

Don’t wait for perfection

Moving to cloud doesn't make you data driven or AI readiness
Focus on impact use cases, speed, adoption and creating foundation that is trusted.  

Most just move bad usless redundat data, to cloud

Comment below, "Cloud Migration" I’ll send you our playbook that gets you adopted, and ROI in weeks.

Most data teams are set up for failure. They create bureaucracy, not value. You can hire best engineers, buy best tools and still fail.

One mid-market company finally approves their “data investment.”
They hire their first data engineer. Maybe even head of data.

Spin up Snowflake. 
Maybe dbt.
Add 2 BI tools with an AI feature.

Six months later: same issues amplified

The engineer is buried in random Slack requests.
The CFO says, “We’re not getting ROI.”
Leadership starts whispering: “Maybe we just hired the wrong person.”

If your dashboards aren’t trusted and your data team looks “slow”
It’s because your operating model.

The truth is, 70% of data teams are set up to fail.

Seen this across our clients and 90% of the time, the problem wasn’t the stack or the people. 

It was how the team was positioned inside the business:

-No clear ownership model (who owns what data? who maintains it?)
-No definition of what success looks like beyond “more dashboards.”
-A data team buried under IT with 100 bosses and no mandate.
-Endless overengineering, copying Netflixs stack with a team of three.

By the time they realize it, the engineer’s burned out, adoption is low, fixes takes 12 months and the CEO’s lost trust data.

You can have every tool, hire the best engineer, and still have the same data problems. Tools and people just amplify an already good data team 

Most data teams create bureaucracy then value.  

Here is how to set up the data environment for succsess 

1. Start with problems, not roles.

Don’t hire anyone until you can name the 3–5 ways data can impact business directly. 
Most hire data scientists 

Otherwise, you’re building a reporting factory.

2. Define ownership before tools.

Map who owns what, by name. 
Every key domain (sales, finance, ops) has one accountable owner. 

No “shared responsibility.” 
That’s how data chaos starts.
The data team also decides and has the final word. 
No more hidden spreadsheets and so on. 

3. Set outcome-based goals.
Data teams shouldn’t be measured by tickets closed.

They should be measured by business impact.
That’s what executives actually care about.

4. Move Reporting to the Business.

This one is hard as I've seen working everything under the sun. 
If your Head of Data reports to IT, they’ll always optimize for infrastructure.
Move them under the people who actually care about ROI and impact. 

5. Centralize Data. Federate Execution.

Have one central team that defines rules, security, and data architecture 
but let embedded data partners execute inside the business units. 

They build one standardized data platform, then give teams freedom within that system with correct guidrails. In short, self-service right way. 

This way, you maintain speed, agility and control. 

The failure started in how you structured ownership, expectations, and accountability long before the first hire.

Your machines are collecting millions of data points. But PLC says “Pump_4A” while your ERP says “Equipment_1403.” Your market edge is trapped inside 20 year old machine and 2008 Excel file. 

Pumps. Sensors. PLCs.
They’re all generating thousands of data points every second.

The problem? 
No one’s listening.

Not because the data isn’t there, it’s everywhere.
It’s because your systems, teams, machines don’t speak the same language.

Machines OT are speaking MQTT or OPC-UA.
Corporate IT systems are speaking SQL and APIs. 
The business speaks “Excel and PowerPoint.”

So you’ve got noise.

I sat with a COO & CIO of a 300-person industrial manufactuing firm.
Their “digital transformation roadmap” - AI, predictive maintenance, digital twins,

But when asked:
“Can your systems tell you in real time when a batch is about to fail?”

Silence.

Here’s what was happening behind the scenes:
-Quality results took 6+ hours to show up in reports. 
-Each factory ran different generations of equipment and software. 
-Production systems (SCADA, MES) weren’t connected to corporate IT (ERP)
- Engineers couldn’t access business data, finance couldn’t trust operational numbers. 

Because here’s the truth:

Most industrial or heavy operational orgs already own the data they need.
They just can’t use it - because it’s locked inside 20-year-old machines, local servers, and 2008 Excel templates.

The result?

Every report was hours late, every meeting a data debate, and AI pilots stayed stuck on PowerPoint. 

All while you are losing $260,000-$2M per hour due inefficency

Here is how to combine OT and IT data into a single, trusted layer to create moat

1. Start with a Unified Namespace.

Stop wiring point-to-point systems.
Standardize how data talks MQTT, OPC UA and normalize it at ingestion.

If one site logs “pump_2B_temp” and another “Pump_2B_Temperature" its problem

If every new sensor needs a custom connector, your architecture is already broken. 
All pulled from UNS from there. 

2. Unify OT and IT data in one foundation.

Make finance, operations speak the same language.

If finance’s “efficiency” and operations’ “efficiency” don’t match, losing money.

3. Govern before you automate.

Everyone wants AI. No one wants accountability.
Assign data owners, document definitions, build lineage.

Every dataset without an owner becomes a liability.
Governance isn't seperate project its part of the architecture. 

4. Democratize, but with guardrails.

Give operators, analysts, and managers access to governed data through a portal or tool they like. Excel not enemy.

It's not giving them all data, but one aligned to their needs. Federate impact, give flexibility.

5. Prioritize by value, not hype.

Start with quality prediction, maintenance alerts, where value is visible and measurable.
If a project doesn’t have a impact, defintion of done, stop.

You have data worth millions

When your OT, IT, and finance speak one language,  you can create a moat

You have a data strategy, but your $1M+ data team is stuck operating like a support desk, not a product team. Why? 

Every exec thinks their data strategy failed because of tools or “lack of adoption.”
But the real reason?
You never built the Data Operating Model,  the system that makes your strategy actually work.

A few months ago, I walked into a company that had rebuilt its entire data stack three times already. 

Each time, the Head of Data promised this one would finally “scale.”
A company invests six figures into a “data strategy.”
New tools, new team, new architecture.
Twelve months later? Same story. 

They’re still firefighting, and dashboards aren’t trusted. 
All those data strategy initiatives are just words on paper. 

Let’s be clear:
Your data strategy tells you what you want to achieve.
Your data operating model defines how you achieve it.

And 90% of mid-market firms skip that part entirely.

Here’s what it looks like when there’s no operating model:

-Everyone is busy, but nothing gets delivered.
-Data engineers are stuck doing ad-hoc reporting instead of building systems.
-Every project is “urgent” because there’s no intake or prioritization process.
- No one owns adoption, dashboards, ML models die in 5 days.
- No metrics to prove ROI, just a list of “completed projects.”

You don’t have a data strategy problem.
You have a delivery problem.
An operating model problem.

Here’s how to fix it:

Stop thinking about “what tools do we need” and delivering 200 long data strategy looking 24-months in advance. 

Start designing how you operate.

1. Define Roles Around Outcomes, Not Titles.

Stop hiring for tools, or 12-year expirence at F500. 
Hire or assign roles based on accountability and value they own.

Who owns adoption?
Who owns trust?
Who owns delivery speed?
If it’s “everyone,” it’s no one.

2. Standardize How Work Moves.

Most companies start with “what dashboards do we need” 
Build a lightweight delivery framework:

Requests come in through a business brief (not a Slack message).
Every initiative has a business owner and an ROI target.
Delivery runs in short, iterative cycles, with feedback not in 12 months
Your team should operate like a product organization, not a support desk.

3. Create Oversight That Actually Adds Value.

Governance is not about blocking work.
It’s about connecting data delivery to business outcomes.

Set up a monthly review with data and business leaders, track ROI, adoption, prioritization, re-frame the strategic roadmap etc. 

4. Trust, Adoption & ROI 

The end goal of your operating model isn’t to deliver 'dashboards',  it’s to deliver trust and impact 

Communicate ROI, impact. 
Build iterative with feedback, not 12-month timelines for 'version_1' 

Every failing data strategy I’ve seen has one thing in common: no operating model.
Everyone is rowing hard, but in different directions

Everyone has ambitions to get "AI readiness", "predictive models", but easier said then done.

Your $500K data foundation will need to be rebuilt. Your main lead engineer left, and now no one knows how anything works. Is it final_v2 or final_final_v3? 

Here’s what that looks like in real life:

Your $250k data project is “done.”
The dashboards are live.
The pipelines work.
The business claps for a week.

Then the lead data engineer or Head of Data leaves.

Two months later, no one knows how anything works.
Everythings break.
Metrics stop matching.
The new engineer spents months asking, “What’s the difference between customer_final_v3 and customer_final_v4?” 

So, the team says what every team eventually says:
 “We need to rebuild it.”

And suddenly you’re back where you started 
another 12-month rebuild, another $200k burn, another “modernization” cycle with the same outcome.

You don’t have a data problem.
You have a documentation problem.

Every undocumented pipeline, model, or KPI is a future rebuild waiting to happen. 

- No lineage → every fix and onboarding takes 10x longer 
- No standardization → Everything duplicated, pipelines, dashboards
- No definitions → finance and sales never agree 
- No ownership → nobody knows who’s responsible.
- No blueprint → every new engineer restarts from scratch.

That’s why your data team spends months “stabilizing,” and your VP wonders why data never shows ROI.

Then reality hits. It's boring and hard. 
You have 400 tables, and no one knows what they do as all context lives guy who just left.

Here is how to use AI and prevent this chaos happening 

1. Treat Documentation as Insurance, 

If your entire data stack would collapse because one engineer quits, you’re uninsured.
Documentation protects your investment. It’s risk mitigation, not a nice-to-have.

If your project doesn’t include documentation as a milestone, it’s already failing.

2. Document for Business, Not Just for Engineers

Most teams document SQL, not context.
Business documentation (what KPIs mean, where from) is what builds trust.
Without it, even the best data model fails adoption.

Use AI to auto-generate first drafts of table summaries and lineage. 

3. Create a Living Blueprint, Not a Wiki Graveyard

Confluence, Spreadsheets pages work for 2-weeks, then die.
Your documentation needs to live where the work happens, dbt, your semantic layer. 

It must evolve automatically as the system changes.
You don’t need another catalog or tool to fix this.
You need clarity, ownership, and documentation discipline. 

4. Assign Ownership

Every dataset, KPI, and model needs one owner.
Not “the data team.” A name.
Without accountability, documentation dies the same week it’s written

5. Modularize It 

Best documentation is invisible. 
Don't creat 200 pages that are irrelevant and impossible
Start incremental, 
Every time you touch smth.

Every undocumented project is a future rebuild waiting to happen

If your data project would collapse tomorrow if one person left, you’re carrying documentation debt.

Your complex $1m data architecture is built in 2015 and looks like vendor expo at a Gartner event. You are spending more time maintaining it than driving impact. 

Tightly coupled. 
Over-engineered.
And allergic to change.

Doing litterly more harm than good. 

You built data foundation like it’s 2015.

Meanwhile, the defense sector, the one with more risk, regulation, and bureaucracy than anyone, is doing it differently.

They call it data flexibility and being agile. 

Was talking to the data leader of a big defense manufacturer and realized their operations were being slowed by 200+ systems. 
They didn’t buy more tools, created more pipelines. 
They stopped, looked at the mess, and asked one hard question:

“How can we move faster without rebuilding everything?”

Flexible scalable data architeriture.
They built a foundation that can absorb change,  not crumble under it.

What they did differently: 

1. Rationalized first.

They killed redundant systems before they modernized.
No more patchwork of “legacy + shiny new.”
One unified platform, one source of truth, one model.
Centrelize, Standradize data, but federate the use of it. 

Stop chasing perfection, and 'future tense'. Data only valuable if used.

2. Standardized everything.

Every pipeline, schema, and permission model followed the same rules.
No cowboy connectors. 
No hidden spreadsheets 
No redunded tables or pipelines ( like v3 or v3_realfinal)

Create a semantic and consumer layer where you centrelize all, but remain agility.

Most teams spent 80% time uplifting then providing impact. 

3. Built governance into the foundation.

I hate 'governance' documentation. 
Not as policy docs, but as code.
They automated trust: lineage, access, compliance.

Hits all from trust, adoption, scale and impact. 

4. Designed for agility.

They didn’t just centralize data, they made it modular.
Imagine like building legos. 
Thats how you should you think about data foundation. 

Add a new source, new data product, deploy an AI use case…
without ripping up the foundation each time.

Most teams spent more time fixing rather providing impact. 

5. Agility, People & Literalcy 

They invested in data literacy, not just data engineering.
Literlcy is not training or workshop. 
It's providing clear ideas, feedback on what is possible with data, ai etc. 
End users could  trust, and act data without anyone. 

Compared to others, and one one manufactuerer who recently audit said: 
“We have everything, but nothing works together.”

Their architecture looked like a vendor expo at Gartner event. 
Data team spends 80% time firefighting pipelines instead of delivering impact.

You’re slow because you’ve built rigid systems. 
You need fewer tools and scalable architecture.

Architeirture needs to adapt to business, not the other way around. 

→ Comment below "Scale"  and will sent you full playbook on building a scalable agile data architecture without risking never-ending rebuilds

You think you’ve migrated. But 80% of your business logic is still stuck in that Teradata box no one dares to touch. Now you have two sources of truth. 

Snowflake is live.
Pipelines in products look slick.
Cloud migration in progress. 

But here’s the uncomfortable truth
Half your data is still stuck in Oracle, SAP, or that Teradata box no one dares to touch.

You didn’t modernize.
You just built a duct-taped mess hybrid trap, legacy + cloud

Hybrid environments look harmless at first.
But duplicating logic, moving everything with no clear priortization, is a dead game. 
Running duplicated systems is not goal.
It's a phase. 

I’ve seen it across dozens of companies:

→ Two data teams are maintaining duplicate pipelines.
→ Two budgets paying to process the same data.
→ And one VP still asking, “Which number do we trust?”

Every month you delay decommissioning legacy systems,
you’re paying double. 

That’s your “hybrid tax.”
And most execs don’t even know they’re paying it.

This includes 

1. Double Spend.
Duplicated Licenses, compute, ETL jobs. 
You’re not saving by “phasing it out slowly.” 

2. Compounded Complexity.
Your engineers are maintaining SQL and shell scripts, dbt and old Informatica
Engineers are babysitting legacy. That’s why your best people quit.

3. Broken Trust.
When half your dashboards pull from the cloud and half from legacy,
Nobody believes the data. Self-service, compliance reporting, AI become impossible.

4. Lost Speed & Innovation.
You can’t run AI or scale insights when half your business logic still lives on an old system. 

This isn’t about technology.
It’s about focus and ruthless simplification.

1. Quantify the Hybrid Tax.

Audit total cost of ownership,  storage, ETL, maintenance, labor.
Show the CFO the overlap. 

2. Decouple Before You Decommission.

Inventory every report, pipeline, and dashboard touching legacy systems.
Then kill them one by one, not “all at once.”

Move only the highest impact that relevant now not 12-18 months. 

The goal isn’t “all in the cloud.”
The goal is no dual dependencies. 

3. Redesign for Reuse.

Hybrid environments expose one ugly truth: 
Your data models are too brittle to scale, usless and 89% deleted. 

Lock in your business definitions (semantic layer).
Build modular, reusable data products.
Document everything.
 
4. Redefine “Done.”

“Cloud migration complete” isn’t done.
 Replicating everything into the cloud is just duplication at scale.

The goal isn’t to move all your tables.
The goal is to translate what matters, retire what doesn’t, and prove ROI at each step. 

5. Institutionalize the Reset.

Every month, review how much of your data still touches legacy.
Everything should be priortized based on impact x complexity from the data strategy 

What % of workloads still depend on legacy?
What costs can we retire next? and so on 

Have seen companies depend on 3 cloud systems + legacy, everything just amplified the chaos.

Every company has that one custom pipeline that breaks right when VP looks at it. If your engineers are spending 70% of time fixing pipelines, you don't have a data team. 

Every company has that one custom data pipeline.
Built by the smartest engineer on the team.
Works (most of the time).
And supposedly “saved” you hundreds of thousands in vendor fees.

The problem?
It’s a time bomb.

When I talk to mid-market firms, I always hear the same story:

“We built it ourselves because it was cheaper, needed something fast to work”

Six months later…

That “free” pipelines has turned into a full-time maintenance job.
Your data engineers spend nights fixing schema drift.
Your new hires waste weeks trying to understand undocumented code.
It breaks at times right when VP looking at the data, 2min ago worked fine.
And your CFO still doesn’t trust the numbers in the dashboard.

Let’s do the math:
Two engineers, three months of work.
$150K annualized salary.
You just spent ~$30k to “build” it. 

But here’s what you didn’t factor in:

30–40% of those engineers’ time now spent keeping it alive.
2–3 months of onboarding every time someone leaves.
Countless hours lost when it breaks right before an exec review.
Missed opportunities because your team’s firefighting instead of innovating.

That $75K build?
Turns into $1.2M+ in lost productivity and opportunity cost over three years.

Here is what to do rather. 

1. Audit the True Cost

What is your data team working on? 
What is the requests, and data initatives (data products) they are shipping? 
Calculate time spent of your data team

Most data teams 70% of time spent on maitainance rather then driving impact.

2. Kill the Zombie Tools

Every organization has 2–3 internal tools or legacy pipelines no one dares to touch.  

Archive, replace, or sunset anything that:
Has no clear business owner. Don't bring ROI. Was built to just have smth.  

Seen recent finance org running redundent pipelines, they didn't need. 
They took more time to maintain, for 0 impact. 

3. Buy the Plumbing, Build the Differentiation

Your engineers should be building the data products that drive growth - not re-inventing connectors to Salesforce.

Buy what’s solved. 
Build only what creates a business advantage.
Buy boring. Build brilliant.

4. Force ROI Discipline

Before any “custom build,” or "data product"  ask one question:

“What’s the measurable business outcome?”

If there’s no answer, it’s not a project, it’s aligning on wrong outcomes.

Can't start building nothing, if dont have outcome, data, problem solving, definition of done, why and sign off. 

5. Move From Maintenance Mode → Innovation Mode

If your engineers are spending more time fixing than delivering, you don’t have a data team  you have a repair crew

Automate the repeatable
Buy managed infrastructure
Let your team focus on data, AI, and decisions that move the business

Your “free” pipeline isn’t free
It’s costing you growth, talent, and trust

Everyone wants AI to talk with their data. But every meeting sounds the same: ‘Wait… where did that number come from? You have trust issues.

What they actually want is confidence in their numbers and forecasts.

But somewhere between the Snowflake bill, the dashboards, and the AI hype deck… that confidence disappears.

And every board meeting sounds the same:

“Wait… where did that number come from?”

At that point, everything breaks.
Your $250K data stack.
Your shiny new AI pilot.
Your data team’s credibility.

Because if no one believes the output, the input doesn’t matter.

AI isn’t your next frontier. Trust is.

Here’s the uncomfortable truth:
Most data programs don’t fail technically

We’ve seen this play out across dozens of mid-market firms:

Dashboards no one logs into.
AI models that contradict intuition.
Executives reverting to Excel, because it’s “more trustworthy.”

This isn’t about tools or pipelines.
It’s about belief.

When your team says, “We have a data quality issue,” what they really mean is:

“No one trusts the numbers.”

So how do you fix it?
Not with another tool.
Not with a new dashboard.
But by rebuilding organizational trust in your data.

Here is what to change. 

1. Shift the Mindset: Trust is an Output 

You don’t “implement trust.”
You earn it.

Stop running weekly “data quality reviews” and start assigning data ownership at the source.
If Sales owns sales data, and Finance owns finance data.  

2. Replace ‘Single Source of Truth’ with ‘Single Source of Trust’

Truth is technical. Trust is emotional.
You can have the perfect data model and still lose credibility.

Fix that by making transparency visible.
Add lineage, timestamps, and ownership right. 

Executives don’t need more tables. They need assurance.

3. Kill the Shadow Data Systems - Openly

Those private Excel sheets and “Finance’s version” of the truth?
They’re not harmless, they’re silent trust killers.

Don’t ban them.
Reconcile them live.

Compare, explain, document, and publish the aligned metric.
You’re not fixing a formula.
You’re fixing belief.

4. Assign a Name to Every Metric

Every data point should have a human next to it.

Business Owner → defines what it means
Data Owner → ensures it’s right.
A metric without an owner is an orphan - and orphans aren’t trusted.

5. Make Governance Invisible 
:
Governance shouldn’t feel like paperwork.
It should feel like safety.

Automate your data validation and lineage tracking.
Keep it out of sight — but never out of control.

If governance feels visible, it’s already slowing you down.

AI fails for the same reason any data product fails, no one believes it.

AI isn’t a tech problem.
It’s a credibility problem.

If leaders don’t trust the source data, they won’t trust the model.
Especilly as models like to agree on a lot of things. 

You don’t have a data problem.
You have a trust problem.
And AI won't fix that.

Everyone wants AI and predictive. But most of your data is locked in 1990s ERP system, 31 spreadsheets, and Joe shouting across the hallway. Moving from legacy ERP won't make you data-driven. 

Most mid-market run on systems older than their CEOs.

That “temporary” ERP you implemented in 2010? Still there.
The homegrown Access database that powers finance? Still critical.
Half your ops? Run on Excel, emails, and shouting to Joe 

You don’t need to replace your legacy systems to move forward.
Moving from legacy ERP won't make you data-driven. 
After 18-months gone be in bigger data mess but in cloud

You just need innovate around it. 

Was talking to CIO at huge industrial company facing this nightmare.

Their ERP was from the 1990s. 
Their factories ran on siloed systems.
Data lagged 6 hours behind reality.

Replacing the ERP wasn’t an option. 
So they built around it.

They built an integration “hub” so all systems could talk without touching core ops.

They normalized data at ingestion.
They governed every metric from the start.

And suddenly, they had real-time visibility, data to leverage a 24-month delay.
They led with value, not tech. 

Here is how to deal with legacy systems

1. Stop Thinking in Systems. 

Stop treating your ERP as your source of systems
Legacy record transactions. 
Build innovation around them, not inside them.

You don’t fix legacy by buying new tools.

You fix it by decoupling the old stack from the new.
Build a central data layer (lakehouse/warehouse) that real source of truth, and let your ERP, CRM, and OT systems feed into it. 

2. Introduce an Integration & Translation UNS Layer.

If your ERP, legacy, OT systems can’t talk API, don’t panic.

Add a lightweight API gateway or event-driven architerture 
Make every legacy system publish standardized events through that. 

If ERP calls it “Material ID", Iot some code and MES calls it “Product Code” you need apply consistent data model, naming conventions

That one layer eliminates point-to-point chaos, everyone speak same language  and gets data with context. 

3. Normalize Upstream, Not in Dashboards.

Most firms want AI but have "machine_21" in warehouse but "M12" in ERP. Not gone work

Standardize naming conventions, master data, use canonical schemas , validate

If we can't align on metrics, nothing will happen

4. Go Hybrid, Not All-In Cloud

Not every system needs to move or even can

Keep sensitive or legacy systems on-prem if that case
Send structured, validated data to your cloud lakehouse
Sync it in near real-time if have clear value there

5. Govern Before You Scale.

When someone pulls “Margin,” they need to know:
What it means
Where it came from
When it was last refreshed 

Assign data stewards per domain  
Define data flows between systems.
Establish business definitions for critical metrics

Don't let your legacy systems stop you

Comment “LEGACY” and I’ll send you the blueprint how to deal with legacy, without breaking the business

taplio

The Battle

Darrell Alfonso

Ali Z

Followers

Posts

Engagements

Likes

Performance ScoreDarrell Alfonso's LinkedIn Performance

Performance ScoreAli Z's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Darrell Alfonso

Ali Z

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Darrell Alfonso's LinkedIn Performance

Performance ScoreAli Z's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreDarrell Alfonso's LinkedIn Performance