Lessons from Shantanu Narayan: Three Ideas That Still Shape How I Work

Shantanu engaging customers at one of our Adobe events, National Gallery, London.

The news that Adobe has (finally) found a successor to its 19-year tenured CEO, Shantanu Narayan, prompted me to reflect on an extraordinary leadership tenure.

19 years as CEO is rare. Over that period, Shantanu led Adobe through fundamental shifts in how software is bought, used and marketed - from perpetual licences to SaaS, from creative tools to a broader creativity-and-data strategy, and into AI.

I was fortunate to work at Adobe during some of those transitions. Now, let me be the first to say by no means was I a close colleague of Shantanu’s, but I was in enough meetings to pick up a few simple ideas that have stayed with me as a B2B marketer:

1. The website should be at the centre oF EVERYTHING WE DO

“I think this web thing’s going to catch on, don’t you?”

Amongst other things, Shantanu had a knack for the cutting remark. I remember hearing this during a global marketing review in San Jose, when the website was not yet central to the B2B buyer journey.

The message was deceptively simple: the website is not a supporting asset. It is where marketing activity should lead. Paid, earned and owned channels should all help create attention and direct it somewhere meaningful. That remains a useful discipline. Make the website the hub, not an afterthought.

2. Interest matters more than survey scores

“I don’t care about brand surveys. All I care about is that people are coming to check us out.”

There is nothing wrong with measuring awareness. But awareness is only useful if it translates into action. Are people searching for your company? Are they coming to your site? Are they spending time with what you have to say?

Those behaviours signal genuine interest. They are why I continue to see engaged web visits as one of the most useful top-of-funnel measures.

3. BUSINESS Reviews should lead to different actions

“I just need to understand what happened and what you’re going to do differently.”

Marketing reviews can become overloaded with slides, analysis and detail. The risk is that reporting replaces learning. This question cuts through the noise:

• What happened?

• What did we learn?

• What will we do differently?

It is a powerful standard for any periodic review. The goal is not to document everything that occurred. The goal is to improve the next period.

A lasting leadership lesson

Shantanu’s tenure at Adobe was defined by change, and by a willingness to make major bets before they were universally accepted.

For me, these lessons remain practical rather than abstract: build the hub, measure real interest, and turn every review into a commitment to do better next time. Cheers Shantanu!

The LinkedIn algorithm isn't ruining LinkedIn. We are.

My eldest has just graduated from university.

I'm immensely proud. And as a parent, of course I want to tell people about it.

But should I tell my 8,000+ professional connections?

I did some rough maths.

There could be around 1.25 million UK parents every year with a child reaching a major school or university milestone and of that ~225k are likely on LinkedIn and ready to post.

So, with my 8,000ish connections, I could potentially see well over 100 graduation posts - directly, through comments or because someone reposted them.

Now add birthdays. Weddings. Anniversaries. Holidays. Marathons. New houses. Children's achievements. Pets. The occasional inspirational encounter with a taxi driver.

If everyone posts everything that's important to them, is it really any surprise that our feeds become clogged with content that has absolutely nothing to do with why we're on LinkedIn?

We love to blame the algorithm but the algorithm can only work with what we give it. I'm not arguing for a LinkedIn without personality. Personal stories can be powerful when they illuminate something relevant to our professional lives.

But "this is important to me" isn't necessarily the same as "this is relevant to you".

Maybe that's a test worth applying before we hit Post.

Anyway, I'm incredibly proud that my eldest has just graduated from university.

And I've managed not to post about it on LinkedIn.

Oh.

Bugger.

Personalization or volume. Pick one. That was the deal

𝗣𝗲𝗿𝘀𝗼𝗻𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻 𝗼𝗿 𝘃𝗼𝗹𝘂𝗺𝗲. 𝗣𝗶𝗰𝗸 𝗼𝗻𝗲. 𝗧𝗵𝗮𝘁 𝘄𝗮𝘀 𝘁𝗵𝗲 𝗱𝗲𝗮𝗹.

Until now.

Sat through 90 minutes of fascinating AI startup pitches at LettsTalk-Tech earlier this month. That line - from a 25-year-old founder in Toronto tackling AI-native outreach - was the one that really resonated with me.

Because it's not just an email problem. It's the defining tension of modern marketing. And it's collapsing fast.

A few things I noticed:

𝗧𝗵𝗲 𝗰𝗮𝘁𝗲𝗴𝗼𝗿𝘆 𝗶𝘀 𝗯𝗲𝗶𝗻𝗴 𝗿𝗲𝗻𝗮𝗺𝗲𝗱. Gartner just rebranded "sales engagement/outreach" as "revenue action orchestration." That's not a rebrand. That's an admission that AI agents - not humans - can now run outreach end to end.

𝗧𝗲𝗮𝗺𝘀 𝗮𝗿𝗲 𝘀𝗵𝗿𝗶𝗻𝗸𝗶𝗻𝗴. Across the AI-native startups I’ve worked wuith, no team is larger than five people past Series A. Not because they're bootstrapping. Because agents are doing the work that used to require headcount.

𝗧𝗵𝗲 𝘀𝘁𝗮𝗰𝗸 𝗶𝘀 𝗰𝗼𝗹𝗹𝗮𝗽𝘀𝗶𝗻𝗴. Apollo. Clay. Hunter. Instantly. A warm-up tool. A LinkedIn tool. Most growth teams are carrying five or six subscriptions that don't talk to each other. "Death by a thousand tabs" is now a category problem, not just a workflow annoyance.

The CMOs I respect most aren't asking "how do we add AI to our marketing?” they're asking something harder: if an agent can run our outbound, what is our marketing team actually for?

That question doesn't have a comfortable answer yet. Personally I don’t believe its a run lean & junior decision - increasingly I’m seeing the need for experience to provide context and decision making. But the companies being built today - lean, agent-first, AI-native from day one - are going to force every incumbent to find one.

The future of go-to-market looks less like marketing and more like software.

Five startups. One 90-minute session. Building a company when AI is the co-founder.

£𝟮𝗺 𝘁𝗼 𝗴𝗲𝘁 𝘁𝗼 𝘀𝗲𝗲𝗱. 𝗧𝗵𝗮𝘁 𝘄𝗮𝘀 𝘁𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿.

It's now £200k.

That was the message from the AI-native startups at this week’s LettsTalk-Tech event. The economics have shifted so fast it's hard to keep up.

Months of work are being compressed into days. Teams past Series A are capping out at five people - not because founders are cutting corners, but because agents are doing the work that used to require twenty.

One startup was rebuilding the reinsurance industry from scratch. Another was dismantling the £50B+ Bitcoin lending market. A third was dismantling a $9 billion sales engagement industry with two founders and an agent.

None of them looked like startups from five years ago.

A few things stood out:

𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝘄𝗮𝘀 𝗮𝗹𝘄𝗮𝘆𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺. The number one reason startups failed wasn't the idea. It wasn't the market. It was execution - fragmented tools, ad hoc processes, too much dependency on who the founder happened to know. AI is systematically removing that bottleneck.

𝗧𝗵𝗲 𝘀𝘂𝗽𝗽𝗼𝗿𝘁 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗶𝘀 𝗯𝗲𝗶𝗻𝗴 𝗿𝗲𝗯𝘂𝗶𝗹𝘁. The old model - advisors, mentors, accelerators - was slow, expensive, and inconsistent. The new model is step-by-step, AI-guided, available at 2am, and doesn't charge you a percentage of your round.

𝗧𝗵𝗲 𝘀𝗲𝗹𝗳-𝗱𝗿𝗶𝘃𝗶𝗻𝗴 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗶𝘀 𝗮𝗿𝗿𝗶𝘃𝗶𝗻𝗴. Agents that autonomously execute strategy, file for IP protection, run legal checks - not as a future concept, but running live in production today.

The metrics VCs have used for decades - runway, headcount, time-to-market - are being rewritten in real time.

What does this mean for established businesses?

The structural advantages incumbents relied on - capital, talent density, execution infrastructure - are being commoditised faster than most boards are meeting to discuss it.

Miss the event? Recording here.

The rise & rise of GTM Engineering

Spent some time this week at Clay's "GTM in GMT" event (possibly the most high-energy B2B tech event to hit London in a long time!).

One theme kept surfacing - AI is collapsing the boundaries between marketing, sales, operations and data. The companies growing fastest are building systems, not campaigns. A phrase that stuck with me: "Your job is no longer running campaigns. Your job is solving the problems constraining revenue."

That mindset showed up everywhere:

𝗔𝗱𝘆𝗲𝗻 is saving tens of thousands of hours by automatically researching accounts, identifying buying signals and delivering weekly intelligence briefings to sellers.

𝗩𝗲𝗿𝗸𝗮𝗱𝗮 is transforming email replies, call transcripts and CRM history into a proprietary first-party data asset that competitors simply cannot replicate.

𝗧𝗵𝗼𝘂𝗴𝗵𝘁𝘄𝗼𝗿𝗸𝘀 is connecting alumni data, CRM signals and audience intelligence to create entirely new pipeline sources.

𝗘𝗹𝗲𝘃𝗲𝗻𝗟𝗮𝗯𝘀 has moved beyond campaigns altogether, building repeatable growth systems where launches, webinars and content become scalable engines.

𝗖𝗹𝗮𝘆 itself? Their growth engine is built around education, community and user-generated content, amplified by AI and orchestrated at scale.

None of these companies were talking about prompts. They were talking about infrastructure. The new GTM stack is shifting:

From: CRM + Marketing Automation + SDR Team

To: Signals + Memory + Orchestration + AI + Human Judgement

It’s the rise of GTM Engineering sitting alongside alongside CMOs, CROs and RevOps leaders. Because the future of go-to-market looks less like marketing and more like software.

AI hacking isn’t just getting better at prompting

My new Google Fitbit Air (the so-called "Whoop-killer") reminded me of a conversation from a few years back. Someone told me they were "body hacking." Intriguing, I thought. Complex area.

Then they revealed their hacking: tracking steps, heart rate and sleep.

Not exactly what I'd call body hacking. I was actually doing more - blood glucose, HRV, the works. But it got me thinking about AI and the never-ending FOMO around it.

If you're only using AI by prompting ChatGPT or Claude, that's great. But you haven't really got started.

It can be overwhelming I know, but I'd always recommend three things that move you along:

𝗩𝗼𝗶𝗰𝗲 𝗽𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴. Stop typing. Start speaking. I use Whispr Flow - it applies AI to help you prompt your favourite engine. Really transformative. It self-corrects as you go, so you don't need to be perfect. Just think out loud.

𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼-𝗪𝗼𝗿𝗸. Buy the upgrade. Get off basic prompting and start dipping your toe into agentic capabilities, without the fear of coding. It'll do things like go into your analytics dashboards, your CRM, your documents and bring back a weekly report in a usable format. Really easy to set up. Game changer for complex tasks. Then move on to Skills!

𝗡𝗼𝘁𝗲-𝘁𝗮𝗸𝗶𝗻𝗴 𝘁𝗵𝗮𝘁 𝗯𝘂𝗶𝗹𝗱𝘀 𝗺𝗲𝗺𝗼𝗿𝘆. Take notes, not for notes sake. My favourite is Granola as it takes notes without joining a video call as "John's Notetaker". Transcribe every productive meeting you can. Product roadmaps, customer conversations, ICP discussions, prospect calls. You're recording your corporate memory. Put those transcripts into a project folder that your AI can access, and you're building a goldmine that gets smarter every time you ask it questions about your business. Paul Cash recently posted about this. He calls it building a Collective Intelligence System.

Hack your way through it. Don't be like the person who thought body hacking was counting steps.

Less fear. More building.

AI breaks all the time. And that's actually reassuring.

Great to attend a session on Pavilion’s latest AI Pulse Report - what’s changing across GTM teams, what matters now, and where leaders should focus next.

In summary, the fundamentals of go-to-market haven't changed. The P&L still matters. Strategy still matters. We're just reimagining how we execute. But don’t accept that we’re just "in the loop" of AI workflows. We're the last mile. The final check, the endpoint. The accountability.

A few more things struck me:

𝗦𝘁𝗼𝗽 𝘄𝗮𝗶𝘁𝗶𝗻𝗴 𝗳𝗼𝗿 𝗽𝗲𝗿𝗳𝗲𝗰𝘁 𝗱𝗮𝘁𝗮. You're never going to have it. Use AI and knowledge graphs to clean as you go. Every time you correct an output, you're enriching the data. Start with what you have. Get moving.

𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗰𝗼𝗻𝗰𝗲𝗿𝗻𝘀 𝗮𝗿𝗲 𝗿𝗲𝗮𝗹. 𝗕𝘂𝘁 𝘀𝗼𝗹𝘃𝗮𝗯𝗹𝗲. Same fears we had moving from on-prem to cloud. The infrastructure exists - hyperscalers, local LLMs, controlled file access. Don't let compliance be an excuse for inaction.

𝗧𝗵𝗲 𝗖𝗹𝗮𝘂𝗱𝗲 𝘀𝗵𝗶𝗳𝘁 𝗶𝘀 𝗿𝗲𝗮𝗹. Power users have migrated. Hard. One person went from being in the top 0.1% of ChatGPT users to using it maybe 5% of the time. The reasoning, the workflow capabilities, Co-Work - it's just more powerful for complex work. I agree!

𝗖𝘂𝗿𝗶𝗼𝘀𝗶𝘁𝘆 𝗮𝗻𝗱 𝗿𝗲𝗶𝗺𝗮𝗴𝗶𝗻𝗮𝘁𝗶𝗼𝗻 𝗯𝗲𝗮𝘁 𝗳𝗲𝗮𝗿. We used to have gas lamp lighters and switchboard operators. Jobs evolve. AI is less than 4% of job losses today despite the headlines. And here's the thing - if AI was really replacing everyone, we'd all be working four-hour days. We're not. We're cranking harder than ever.

𝗗𝗼𝗻'𝘁 𝗴𝗼 𝗶𝘁 𝗮𝗹𝗼𝗻𝗲. There are no experts here. Just people at different points on the journey. Being able to share where you're stuck, learn from others building in public - that's becoming a necessity, not a luxury (happy to share my own experiences!)

Less fear. More building.

Thanks to Jonathan Moss, Andy Jolls and Sam Jacobs for the discussions.

Less Mad Men. More operators

Spent the afternoon at HotTopics CMO Studio at Abbey Road. Smart people. Strong opinions. A few themes kept surfacing:

𝗦𝘁𝗼𝗽 𝗰𝗲𝗹𝗲𝗯𝗿𝗮𝘁𝗶𝗻𝗴 𝗴𝗿𝗲𝗲𝗻. Most teams still present an overly positive picture, desperate to dial up the green. The real leverage is in what's broken - and having the courage to say it out loud and fixing it.

T𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗶𝘀𝗻'𝘁 𝗮 𝗽𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗲. 𝗜𝘁'𝘀 𝗮 𝗵𝗮𝗯𝗶𝘁. The best operators aren't doing big-bang change. They're fixing 1-2% inefficiencies, relentlessly. Over time, that compounds. It’s about operational efficiency.

𝗔𝗜 𝗶𝘀 𝘆𝗼𝘂𝗿 𝗮𝗺𝗽𝗹𝗶𝗳𝗶𝗲𝗿, 𝗻𝗼𝘁 𝘆𝗼𝘂𝗿 𝘁𝗲𝗮𝗺𝗺𝗮𝘁𝗲. Speed, scale, efficiency - yes. But accountability, judgment and sleepless nights still sit firmly with humans. Co-pilot, not auto-pilot.

𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝗔𝗜 𝗿𝗶𝘀𝗸? 𝗦𝗮𝗺𝗲𝗻𝗲𝘀𝘀. Everyone’s using the same tools = same outputs = diluted brands. The edge will come from how you think, not what you prompt.

𝗧𝗵𝗲 𝗯𝘂𝘆𝗲𝗿 𝗷𝗼𝘂𝗿𝗻𝗲𝘆 𝗶𝘀 𝗮 𝗺𝗲𝘀𝘀. Non-linear, multi-person, and for many B2B brands often starting years before "intent" shows up. If you're still clinging to a neat funnel, you're modelling a world that no longer exists.

𝗠𝗤𝗟𝘀 𝗮𝗿𝗲 𝗼𝗻 𝗹𝗶𝗳𝗲 𝘀𝘂𝗽𝗽𝗼𝗿𝘁. The shift is clear: from leads to signals. From volume to pipeline + velocity. From activity to revenue impact. Plenty of teams quietly admitting they've killed them.

𝗠𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴 𝘁𝗲𝗮𝗺𝘀 𝗮𝗿𝗲 𝗯𝗲𝗶𝗻𝗴 𝗿𝗲𝘄𝗶𝗿𝗲𝗱. New roles emerging (hello "marketing engineers"). Old structures breaking. More expected with less budget. This isn't evolution - it's a rebuild.

The big takeaway? Marketing is becoming less about campaigns and more about systems, signals and continuous improvement.

Less Mad Men. More operators.

6 marketing fundamentals I keep relearning

Grateful for a busy few months working with several SaaS and data businesses. A few reminders along the way:

𝗦𝗽𝗲𝗲𝗱 𝗯𝗲𝗮𝘁𝘀 𝗽𝗲𝗿𝗳𝗲𝗰𝘁𝗶𝗼𝗻. Theory doesn't grow companies. Execution does. You don't need the perfect plan - you need to start. I've built a messaging framework "live" in a one hour workshop, a digital strategy in a morning.

𝗖𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝗰𝘆 𝗯𝗲𝗮𝘁𝘀 𝗰𝗹𝗲𝘃𝗲𝗿𝗻𝗲𝘀𝘀. Consistent messaging is better than better messaging. Endless tweaking is a great way to ensure the market never understands what you do.

𝗔𝘁𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻 𝗶𝘀 𝗼𝘃𝗲𝗿𝗿𝗮𝘁𝗲𝗱. When you're small, pipeline beats precision. Yes, learn what's working, but don't get lost in the weeds trying to prove it. Teaming is more important than departmental victory laps.

𝗗𝗲𝗰𝗸𝘀 𝗮𝗻𝗱 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸𝘀 𝗮𝗿𝗲 𝗲𝘃𝗲𝗻 𝗺𝗼𝗿𝗲 𝗶𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁. But not for humans. For AI. Those plans/style guides/messaging frameworks that no one read are now the lifeblood of an AI-forward marketing team. Just don't labour the (power)point.

𝗔𝗜 𝗱𝗲𝗺𝗼𝗰𝗿𝗮𝘁𝗶𝘀𝗲𝘀 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝘆. This feels like the early days of marketing automation all over again. Except faster. And bigger. Small teams can now punch way above their weight. Yes, there's a risk of beige content - but the upside is enormous if you get it right. Oh, and Claude all the way - great writing, much better file creation and Co-Work really is a game changer.

𝗥𝗲𝗳𝗲𝗿𝗿𝗮𝗹𝘀 𝗮𝗿𝗲 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴. My business has been built on them. No funnel beats trust. No campaign beats a recommendation. In the spirit of which - check out Thomas Coles if you need help. More people should be thinking this way.

And finally, eternally grateful to fashion designer JW Anderson for producing branded JWA merch 😀

20 years of friendship. 150 years of brands.

Finally caught up properly with B2B marketing legend Jason Miller the other week. We go back nearly 20 years, but like most things in life… work, travel, life… we hadn't really connected in a while.

We met at the Museum of Brands in west London. A brilliant place that charts 150 years of brand marketing, largely through a UK lens. (BTW you pay once and get free access for a year).

What struck me (again) is how universal brand memories are. No matter your background, there's always something in there that pulls you back to a moment, a product, a feeling.

For me, it was computing. My first computer was a Commodore PET that my dad lugged home from work – took two people to carry it and it was roughly the size of a small fridge. That led to the ZX81, the ZX Spectrum, and a path that eventually took me into computer science… and then into sales and marketing. Less love for coding itself, more love for what technology makes possible.

If you don't know Jason, he's a brilliant example of a true learn-it-all. From content superemo (I first met him at Marketo) to brand, demand and growth, he's built serious T-shaped depth – with creativity right at the centre. He's also a rock photographer and runs a refreshingly honest, no-BS marketing podcast that's well worth a listen (https://lnkd.in/eQ-a_kbM).

A great reminder of everything that's good about marketing: curiosity, creativity, and people who never stop learning.