Oren Greenberg

B2B Marketing Led Revenue | Marketing Advisor Scaling Tech Companies

Smart companies aren't scaling their GTM teams. They're scaling systems powered by AI.

I help B2B tech revenue leaders (CEO/CRO/CMOs) build AI-first GTM systems that generate pipeline. Typically, technology mid-market.

GTM engineering has shifted from optimising what exists to building entirely new revenue engines, where one AI-powered marketer produces output equivalent to that of 5 traditional team members.

Efficiency is the route to increased effectiveness.

My toolkit includes Python, Google Colab & Cursor. Also Clay.com

I leverage data science & statistics to shift from anecdotal debates to scientific conclusions.

Starting with foundations by looking at patterns in your customer data e.g.:
- Who your ideal customer profile is
- Why they buy from you specifically
- How to find similar accounts to target

Using these insights, I help the sales team know which prospects & accounts to prioritise.

This results in:
- Increased sales velocity
- Less churn
- Lower cost of customer acquisition
- Higher lifetime values

Companies using AI-native GTM systems achieve:
* 5-25% revenue increases
* 30-70% higher conversion rates
* 5-15 hours/week recovered

What does that look like in practice?

Strategic direction:
* Ideal customer profile segmentation & prioritisation
* Automated competitive intelligence monitoring

Insights:
* AI-driven account research
* Signal-based triggers
* Predictive lead scoring

Automation:
* Content engines producing assets at scale
* AI lead routing & qualification
* Multi-channel sequences 

Pipeline acceleration systems:
* AI-augmented outbound
* Predictive lead scoring

I help you leverage the intersection of GTM with data science to drive growth.

Background:
* 20+ years building revenue systems for B2B SaaS (HandsHQ, Redbrain, Peakon, SeedLegals + enterprises such as BT, Lenovo, Canon, Investec Bank)

* Managed millions in marketing spend

* Scaled bootstrapped agency to 7-figures, now I focus on consulting

I help teams become category leaders through system building. 

Find out more + get your AI marketing capabilities score → orengreenberg.com

Founder & Fractional chief marketing officer

Kurve is a boutique digital marketing agency scaling tech businesses into hypergrowth.

I’ve been hospitalised.

Got hit with the flu that’s developed into a respiratory complication.

Fighting an unpleasant fever; I asked the doctor what the best remedy is and they said posting on LinkedIn.
I don’t argue with experts.

For all the flaws of the NHS it’s pretty remarkable how efficient hospitals are. 

Doctor was amused we were chatting to Claude about medical history as that’s all living in a project I setup (Covid messed me up with a nasty dust mite allergy so the past five years have not been easy).

Gutted I won’t make it to TuringXSystm
fest on Thurs. Brilliant speakers and lovely humans - recommend you check it out.

Scaling B2B SaaS & AI Native Companies using GTM Engineering.

No alternative text description for this image

And people say cold email outbound is dying. Can’t for the life of me figure out why.

A leaked OpenAI file reveals interesting insight from user interaction tracking.

Specifically just how how few clicks chatGPT sends.

For one top-performing page, the stats:

- 600k impressions
- 4k clicks

That’s a 0.69% click through rate.
Not great. On most paid search & social platforms you’re looking at 2-7x more on average.

More stats revealed:
Best individual page CTR: 1.68%
That’s the max you’ll be getting (at an unlikely stretch).

Most other pages: 0.01%, 0.1%, 0%.

0% means you could be putting in a lot of effort to optimise for rankings in gen AI without viable ROI.

Across my clients & network gen AI is generating good quality enquiries & customers. So even trace amounts of (relevant) traffic can make a difference.

Keep in mind that although the tech is sweeping it is new which introduces ambiguity & volatility. This doesn’t make planning any easier for revenue leaders.

This is a symptom of the overall shift we’re seeing with moving from quantity to quality in GTM.

graphical user interface, text, application

Here's my view (backed by research) on the marketing skills most at risk due to AI disruption.

Beyond language-related skills (basic writing, email marketing) the tasks next to be on the chopping block are:
• Repetitive and procedural
• Include limited decision-making authority
• Lacking in specialist know-how

Yes, entry-level roles are at risk.
The shift will result in students & entry-level functions spending longer academically developing specialised skills before entering the job market.

The traditional generalist approach to hands-on experience across multiple domains is in decline. The shift is now primarily technical & data-led + system design & orchestration.

This doesn't mean hiring AI-generalists who'll manage numerous specialist agents. This POV underestimates the importance of ecosystem context in knowledge-based roles. Companies need specialists with AI chops, not AI-first generalists.

Understanding which categories are growing, which players in each category are hot, which channels present growth opportunities. The best GTM experts I know live this daily - accessing specialist knowledge only available in dark social & private relationships. Generalists aren't in those circles.

Marketing will evolve to become less 'soft' and 'hazy' to more predictive & systemised. Just as marketing automation, performance marketing, product-led growth and data-analysis continue to grow in importance.

The data I've presented before supports this - generalist marketing roles have shrunk 5-7% over the past 5 years.

This is already happening in more advanced companies. With specialised terms like GTM engineering, AI ops and variants of growth marketing evolving.

So what skills are needed now and in the near future?

1. Context engineering (previously known as prompt engineering). Understanding and managing the right information to feed in and manage. With basic prompt-structure know-how.

2. Vibe-marketing/coding. Being able to use Loveable / Bolt and similar tools to create assets. e.g. ROI calculators, dynamic quizzes. Lite software tools for buyer enablement.

3. Deploying AI agents. These serve as always-on LLMs to take on specific tasks. Mostly recurring mundane ones like enriching inbound leads with info from website.

I cleaned up 25k unread emails in 20min using AI. You can do it too.

Context: I’ve managed to accumulate many newsletters over the years. In a desperate attempt to consume as much as I can to become a super-brain.

At one point the majority ended up on the promotions tab in Gmail. Because my super-plan became an overly ambitious super-nightmare (common sense for most people but I like learning things the hard way evidently).

Step 1. Ask Claude what the process is to do this.
Step 2. Follow the steps outlined.

Here’s what it looked like for me:
- Admin credentials in Google admin to setup API access
- Python script run locally from Terminal

This sounds technical but it isn’t, more like learning new tools. No coding required.

Gmail can identify newsletters via those that have unsubscribe option.
Then mass delete all emails older than 30 days.

I’ve now setup multiple forwarding rules & labels. The way I do that:
- 3 priority buckets
- Newsletters labelled under each priority bucket

For examples emails from software subscriptions I pay for are all nested in own labels under priority1.

So now I have all new newsletters neatly categorised into their own labels with only important emails hitting my inbox.

Yes I could just pay $15/mo for email management software. But that isn’t as much fun.

Some of the cooler uses of Nano Banana Pro I've seen.

Implications for marketers with the rate of advancement:
- Less reliance on designers for infographics
- Less reliance on designers for creating ads

Meaning more creative variants to find best performers (within constraints of distribution budgets).

But also more noise making it harder to push out remarkable ads.

I predict that in time there will be a strong counter-culture shift to having to show the authenticity and human-touch behind creatives produced. It'll happen in B2C first but will eventually hit in B2B (500-600 years later or so like most things).

This will result in a cull of low-level designers. Where the top ones shift focus to creating new creative concepts & design-led innovation.

There's still areas top designers won't be affected:
- Logo design (for now)
- UX/UI components (for the near future)

Tried nano banana to create visual from a PDF. Total fail. 

But, then asked it to summarise and extract key points.
Then create image.

See attached output.

Input file was Semrush's recent report on AI visibility (131 pages so not surprising AI couldn't ingest and make a clear image out of it).

Point is: break down your tasks when you make an ask for better output.

Don't think folks in traditional GTM/SaaS are quite aware of how hard they are going to get hit due to AI. Mostly due to vibe-coding.

All interactions related to:
- Learning a specific UX workflow
- Having to use Regex / Javascript / Python to work with data
- Stitching up multiple sources via ETL tools
- Setting up automations (Zapier/Make/N8n) through knowledge of specific nodes/modules/vendors

Put another way - any tool that follows a procedural or linear process related to data that would live in a database or Excel file.

Are all going to be made entirely defunct in the next 12-36 months. AI is already doing a decent job of catering to the above jobs-to-be-done.

MCP integrations will accelerate this due to ease of access but the real transformation will be due to foundational model improvements combined with progressively lower costs.

Lower costs means running thinking-models for longer plus running more agents to gather richer contexts leading to better quality and more accurate outputs. As the semantic-layer (helping the AI agents understand relevance) improves, this will further accelerate adoption.

Cutting-edge tools like Clay and Notion are already implementing chat-first UX.

This doesn't mean GTM-related expertise will shrink and be replaced by generalists. Rather a shift to orchestration & management of multiple agentic workflows.

This includes a deeper understanding of who the best data vendors are for specific use cases (e.g. contact data, social engagement scraping, technographic enrichment) which forms the basis of all effective GTM.

The many pains that block growth, such as poor quality records in CRM, manual and conflicting marketing automations, are being replaced by AI-first mechanisms.

The barrier for entry will drop, resulting in more GTM-engineers/strategists as traditional SaaS sales & marketing roles evolve.

"Wanted your thoughts about gtm these days, how to win attention through all this noise" someone asked me on a DM this week.

My answer:
There's a few ways to win attention with so much noise.

Triple caveat: strategies vary depending if you're: 
1. Small/mid/large
2. Targeting SMB/mid-market or enterprise segments
3. Your target account market size

What I'm seeing:
1. Spend lots of money. On ads. Influencers. Flashy big events and intimate dinners. If you've got the budget.

2. Do the above, but don't spend a fortune. Be selective with which influencer you work with. Which keywords you bid on. Which audiences you target. Run fewer events.

3. Produce unique content.
Smaller companies like Wynter or HockeyStack pack a punch there.

Mid-size ones: 6sense, Semrush, Gong, Mixpanel.

The large ones...well, we all know the multi BN ones.

If you don't have unique data then leverage partnerships or conduct qual/quant research for differentiated content. As long as it's insightful/actionable, it'll work.

4. GTM engineering is working.
Leveraging AI for workflows, automations, enrichment and prioritisation.
- Signal-based prioritisation
- Automated personalisation engines
- Research automation to support SDRs/AEs
- Smart lead routing (based on ICP)
- AI cold email outbound (only for SME, doesn't perform well for mid-market & enterprise)
- Tracking champion movements across businesses

Having clean data (e.g. in your CRM). Having your tech stack integrated and working well.

5. Baking in distribution into content creation rather than leaving it as an after-thought.

6. Leveraging founder brand. Although this is hard. It's fairly slow. It works best when the company is growing quickly. And not many examples of people doing it really well (but the ones who do are seeing great return on effort). Pulling in your team as well to maximise reach & opportunity.

7. Including warm outbound and intent or inbound-led outbound, there are also opportunities for partner marketing. Someone I spoke to the other day said a fair chunk of their pipeline comes from a niche industry-specific CRM partnership.

ABM & social selling work. When done well.

What doesn't work:
1. High volume content production in a world moving towards quality over noise - this is true across most B2B channels.

2. Going broad multichannel beyond your budget means you don't cut through on any specific channel.

3. What's waning in potency: lead gen, while demand gen is getting momentum. SEO while AEO/GEO is growing.

4. Complaining that a few giant platforms monopolise the world and we're all forced to pay taxes to them in one form or another. It's true, but complaining about it won't help you drive growth.

I didn't highlight specific channels. That's because they are often dependent on a hyper-audience-specific. Broadly speaking the lower the account value the more digital marketing plays a driving force. The higher the account value the more face-time you need.

Pretty nuts if you haven't seen this, Elevenlabs cloned my voice in under 15min.

I've been deep-diving into their platform recently. If you haven't heard of them, they are one of the most interesting rocketship unicorn businesses to come out of Europe in recent years.

Because I'm obsessed with cutting-edge AI tools, I'll be sharing more about their platform features & capabilities - with a focus on agentic use cases.

I've already started building workflows and free tools for your pleasure, which I'll share soon (make sure to sign up for the newsletter so you don't miss future updates).

Excited to be one of their new partners with a focus on agentic for enterprise.

#ElevenlabsPartner

Looks like AEO ("Answer Engine Optimization") as a term might be winning the race.

I put SEO in for contrast.

You *almost* always want to say 'research this' for any response from a large language model.

The models are trained on data that's 6-12 months old.

Example:
Ask chatGPT: "how to intergrate phantom buster in clay"

Response: "Clay doesn’t have a native PhantomBuster integration".

Ask it to run research, the response then is:
"What the native integration supports:"

It isn't stupid, it's just operating from a limited or outdated knowledge base.

Why almost?
Because if you're looking up the capital of France that isn't something that's changing often. But for almost anything else GTM related you want to make sure you've got fresh context.

From what I've seen more problems happen due to not inputting the right quality information into AI-systems (user error) than AI systems going haywire (system error).

Common mistakes:
- Context overload
- Irrelevant context
- Not breaking tasks into granular steps
- Using the wrong systems
- Expecting the AI to do the thinking for you

I'm running an analysis on the entire AI landscape - 28,945 websites.

I'll post tidbits as I uncover them, but here's one that just struck me and I wanted to share with you.

Big AI brands get all the media attention. But we often forget there's an entire ecosystem of companies built on top of the giant foundational models.

40% of all AI companies (being AI powered, or AI-native etc') have fewer than 10 employees.

There's a massive long tail of solopreneurs and small businesses.

Some of the biggest power users of chatGPT API are companies with fewer than 15 people.

Instagram had just 13 employees when Facebook acquired it for $1 billion in 2012.

This breaks the traditional approach to B2B where employee count was historically a critical variable when defining target prospects.

Bigger doesn't always mean better.

For all the hype with go to market engineering and in particular signals there are two specific trusted & reliable events that consistently perform across clients.

The first is tracking champions/users who move to a new business.

Companies spend an inordinate fortune on building brand, educating the market and creating relationships with new people & accounts.

Aka cold prospecting. Which is becoming less efficient across all channels across all industries in B2B - not just SaaS.
The increase in automation is met by the rise in privacy constraints. From cold email filters, LinkedIn cracking down to Apple caller id - uninvited guests are facing increased barriers.

Past customers - the ones who’ve experienced direct benefit are of disproportionate value as a result.

When a person who’s used your product migrates to a new company that moment opens up a hot opportunity in an entirely new account.

Making it a foundational signal.

The second signal is when someone starts a new role. When your target ideal buyer persona starts a new challenge they are in deep evaluation mode and more open to new solutions.

Across my client base these 2 carry the most consistent predictive power for generating growth than any other signal.

As the years progress I end up viewing most things in GTM through a lens or jobs-to-be-done. Timing is a critical dimension to nailing traction.

Shift to focus more resource on tapping into the moment where there’s increased probability a potential buyer is open to your solution - you’ll get higher ROI.

One of the more prolific use cases for Gen AI I'm seeing lately is personalised web experiences - landing pages, demos, emails.

This tech existed previously but was quite pricey and required manual inputs. Now it's proliferating and will soon become a commodity (given the number of players I know of doing this).

It's still a critical piece of kit for marketing/sales to increase conversion rates by offering a personalised journey - the tech will only improve in time.

There's no such thing as 'too personalised'. As noise levels increase, it becomes critical.

The most important thing to know is that all personalisation efforts depend on the quality & depth of the data you have about said person/account in your CRM/CDP.

Which means the infrastructure layer for gathering real-time data about said person/account is becoming critically important. Fetching info from website, but also job ads, social posts, reviews on third-party sites, etc, for each contact record.

Having clean data that can be used was a differentiator before the advent of Gen AI - but that advantage is having a multiplier effect now as a result of how it can be leveraged.

Invest in clean data. Invest in enriching your data.
Only then invest in leveraging your data.

You don't have a GTM challenge.

You likely have a GTM prioritisation challenge.

You're not limited by resources. 
You lack the clarity re where to allocate your resources more effectively.

Which motion to commit to - and go all in.
Which channels to double down on.
Which ICP to define as tier 1.

When you spread your GTM team too thinly, you end up with ambiguity over how to build a scalable & repeatable growth engine.

Here’s a good alternative:

• Get the leanest tech stack, not the fanciest
• Take your time with org design, can't afford to mess it up
• Use data to define your tier 1 ICP only (see my post pinned on profile if you missed it)

Focus is how you win.

I've recently shifted to recommending clients use Clay's new reverse IP (what they call: deanonymization) upgrade.

The reason is due to the waterfall cascade.
Instead of 1 vendor, Clay aggregates them to give you more coverage.

Some businesses I support with sales teams of 20-70 heads rely on this signal as part of their GTM strategy.

(Thanks to Yaniv Mazor for the tip).

Interesting new research that breaks up a task into over a million sequential steps, each handled by a focused agent call.

Three core components:
Break tasks into the smallest possible steps – each agent focuses on a single action
Get multiple agents to vote – different agents independently solve the same step. An answer is only accepted once it leads by enough votes.
Throw away dodgy outputs – if a response is too long or badly formatted, discard it and try again
The system solved a puzzle requiring over one million steps with zero errors.

The individual error rate wasn't eliminated; it sat around 0.2%. But the voting mechanism caught mistakes before they could cascade.

One counterintuitive finding: don't try to fix malformed outputs. When an agent produces a messy answer, it's more likely to have reasoned poorly. Bin it and resample.

Another surprise: smaller, cheaper models won even though they have higher error rates on average. This works because the task is broken into tiny steps. 
Then each step gets multiple independent attempts. 
An answer is only accepted once it leads by enough votes so errors get outvoted.

Before running a million-step task, you can estimate which model will be cheapest by testing error rates on a small sample. They ran 10,000 random steps first, measured error rates, then picked the most cost-effective model.

Another interesting point: smaller models focused on tiny tasks are easier to audit, control, and sandbox than one powerful model doing everything. They argue this approach reduces the risk of uncontrollable AI behaviour.

Longer responses = more errors: once responses crossed about 700 tokens, error rates jumped from ~0.1% to ~10%. When an agent overthinks a simple step, it's probably confused.

Takeaway: chunk your tasks into smaller steps + add a validation layer. Bigger models doesn’t mean better if your prompt is poorly defined or your context layer is thin.

New open source Kimi2 outperforms ChatGPT 5.
Or does it?

GPT-5 beats K2 on multimodal. GPT-5 is also a better writer (which isn't saying much as Claude does a better job there than either model).

K2 Thinking, however, is a better planner. Can execute 200-300 sequential tool calls autonomously.

You can inspect reasoning traces & fine-tune for your domain so for regulated industries, K2 is a better option.

K2 Thinking trained for mere $4.6M.

Open-source models commoditise margins so the only way to justify the huge cost in infrastructure is if investing in larger models means it delivers better outcomes.

Research suggests that smaller models tend to outperform larger models. So it's possible that the giant infrastructure projects from OpenAI will be one of the largest financial black holes in history.

GPT-5 must remain proprietary because the math only works if developers stay locked in.​

K2 costs 1/10th what GPT-5 does per token.

Eventually, the question isn't 'which is better' but it's 'can you justify the 10x markup?'

That's historically the moment proprietary dominance starts to crack. e.g. last year ANZ Bank switched from OpenAI to fine-tuned open source Llama because of quality-cost consideration.

Don't ask 'which model is the best?'
Ask:
- Which model is the best for this project
- What's the business case

Here's I think about prioritising smart signals for GTM motions. And why most companies are still working with an outdated playbook when assembling their target account list.

Many folks are still using generic firmographics to build their target list. Using industry + employee count. 
But they're missing out on establishing relevant context.

I've been working with the Cargo 🧱 team to set this up for a client recently.

Layer 1:
Identify new hire in closed lost accounts. 

When a new hire begins in closed lost accounts, you get notified and can reach out. You do this for the ideal buyer persona who meets your criteria.

Can also do this for new target accounts you're focusing on - when a new hire starts in that role you add them to your ABM-mix.

Why does it work?
Typically a higher propensity to purchase in first 90 days of starting a new role.

Layer 2:
Identifying champions moving to new company. 
If a previous user has had positive experience and then leaves to new company they have context of your solution. 

I'm seeing this one work across multiple clients at the moment.

Layer 3:
Finding target accounts using customer Lookalikes.

There's a few ways to build out lookalikes. I prefer to build a custom formula using multiple factors (https://lnkd.in/eUBWfxRv).

Sometimes that isn't viable if public characteristics are weak indicators, statistically speaking.

Layer 4:
New users' past employers
Someone who's worked in a target account previously, but is now a user who can give you an intro into their old company.

Layer 5:
Reach ex-employees of your new customer who've moved to prospect companies.

Every time you win a new logo, automatically surface all qualified personas who used to work at that company and are now somewhere else.

Essentially running an alumni playbook in reverse.

--
People don't believe what you say, but what other people say about you. These tactics milk that to the max.

Thanks Maxence de Villepion  🧱 for input on this post.

taplio

Oren Greenberg's Linkedin Analytics

Oren Greenberg

Performance ScoreOren Greenberg's LinkedIn Performance

What is Oren talking about?

Oren Greenberg's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Oren Greenberg's Linkedin Analytics

Oren Greenberg

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);}Oren Greenberg's LinkedIn Performance

What is Oren talking about?

Oren Greenberg's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Oren Greenberg's Linkedin Analytics

Performance ScoreOren Greenberg's LinkedIn Performance

Oren Greenberg's Best Posts (last 30 days)