Hi, I'm Glenn and this page is a permanent record of things I thought, built and published.

This week Human37 turns five. Half a decade. A milestone worth contemplating. I wrote the first version of this memo on a warm Wednesday evening on my balcony the week before the 21st of June, in the dark, surrounded by the sound of crickets. The perfect setting to think about the project I poured the last 5 years of my life into.

Could we build one?

Six years ago we were in our early thirties, running a successful business on behalf of a major group. Director seats, good salaries, a genuinely comfortable position. And yet, when you reach that point in your early thirties, the question starts surfacing whether you want it to or not: what does the next chapter look like professionally? At the same moment in time we were asking ourselves this question we could see the market shifting in a direction that excited us - customer data.

A follow-up question presented itself: "We know how to run a business but can we build one from scratch?"

Months went by and the question stayed with us. So we decided to find out. We built a plan, resigned, and started working out of bars and coffee shops. For six months none of us got a salary while we spent twelve hour days working to build something from nothing. It was an intense period. My mortgage application nearly got declined because banks do not love the phrase "early-stage consultancy." My co-founders had just become dads or were about to. The timing of it all was personally challenging but we did it anyway, because the opportunity felt more urgent than the risk.

Five years later

Five years later it turns out we were able to build a company. Five years in, the market we bet on materialised. It is real. Martech is standard vocabulary. Most organisations run stacks of interconnected tools and are asking the harder question - how do we actually use these together? How do I drive value? What do better experiences look like? RFPs arrive shaped around exactly what we do. Human37 is recognised as a leading reference in the field of Martech and customer data across EMEA and beyond. The shift we saw coming from those director seats happened. We were just early enough to be ready for it, and consistent enough to still be here when it arrived.

Of course we did not get here alone. We have been lucky enough to find smart(er) people to join us and clients who trust us. They are the ones who helped Human37 get to where it is today.

Where is Human37 heading

We have always been process-oriented. That does not change - it deepens. We are building out our organisational structure to support the next phase of growth, and rolling out Protocol37 as the operating framework that sits underneath everything we do.

We started as four people with a conviction. Today we are a team of well over twenty, working across EMEA, with partnerships built on the technology we genuinely believe in. The next phase is about making that scale without losing what made the first five years work.

Brussels is home, but we have always operated as a European firm and that only becomes more visible in our daily operations. Our team is international and we spend more time crossing borders to visit clients than we ever thought we would.

The market is shifting again. The organisations that were figuring out their stack five years ago are now asking what to do with it. How to activate it, how to govern it, how to make it work with AI tooling that is moving faster than their data infrastructure can support. That is exactly the problem space we have been building toward. We are not chasing the next wave. We have been standing in it for a while.

On a personal note: what did the last five years teach me about being a founder

Discipline is everything. Working while your friends are out for drinks is a real and recurring experience, especially in the first years. It does not feel like sacrifice when you are committed to what you are building - but it is still discipline, not motivation. Motivation fluctuates, discipline holds and gets the work done regardless of the mood.

We overestimate the complexity and underestimate the capacity to make it possible. That's especially the case in services. What it takes is planning and commitment. Looking back today we never really took a risk when we started. We prepared, leveraged our experience and network and rapidly built a first portfolio. Costs were low to start with anyways. Worst case we could go find another job.

Everything takes longer than you think, and then more time after that. The companies that make it are mostly the ones that did not run out of patience before the results showed up. Prioritise long-term thinking even when short-term pressure is the loudest thing in the room.

Your co-founder relationships are load-bearing. You will disagree but that is fine and necessary. What matters is whether you can quickly remember why you chose to do this together. I picked three people - Julien De Visscher, Maxime Vandenbussche and Vincent Crochet - I could trust and build with. That choice mattered more than almost any other and is the cornerstone of what we've built the last five years.

There is no growth without a team. In the beginning you can be everywhere all the time all at once. As you grow, you can't. The only way to deal with this is growing the business through people and making the people grow in turn.

On a personal note: what did the last five years teach me about myself

I never identified with the word entrepreneur. Five years in, I still do not. I do what I think needs doing - have an idea, develop it, chase it, force it into life. During these five years, my attitude towards risk seriously changed. I still consider myself risk-averse, but I take significantly more risks than I did 5 years ago. My baseline towards what I find risky has significantly changed which makes it easier for me to try to realise new things.

You need to be able to mitigate some of your (extreme) convictions. I love operations because it is the one thing that will get you results. I can be very direct (and annoyed) when client context makes things inefficient. Corporate politics is for instance one of the nails to my coffin. Then again, I do realise (very slowly) that as a founder you need to be able to navigate all types of corporate structures in the interest of the company. Managing that with grace will forever be a balancing act and working point for me. The day I master it should be the day I quit as it probably means I will have given up to some degree on my initial convictions.

You have to be able to recalibrate your perception of self value as the company grows and you move through different roles. Some of the roles I held were very scoped and value was measurable in billable hours. Today my role is mostly described as "client work and whatever needs to get done others don't have time for". I still struggle with the new measurement of value attached to that, but I do understand it is needed.

Owning a company means the work is never finished. There is always more. That is not a problem I can solve. It is the conditions I am operating in, and learning to work with them, including knowing when to step back, is part of the job. I think I am getting better at the last part, but there is still work to be done. It is not uncommon that my co-founders receive messages past midnight. A habit that needs to be tuned down.

We don't celebrate nearly enough. There is always something to build towards. I've realised that I need to allow myself to celebrate. Book the holiday I postponed. Stop earlier and go for a drink with friends. Have the morning coffee with my partner.

I find immense pleasure in creating. Nothing gets me more excited than a new idea or a concept or hypothesis that begs for validation.

Happy birthday

Five years later. The work is more challenging and more interesting than when we started. We are not in a comfortable seat thinking about what else we could build. Simply because there is still so much to build here. Five years, half a decade and it still feels like I'm sitting in that bar on day one. That's a good sign. Happy birthday Human37.

Agentic commerce is the bill for every governance shortcut you've been deferring.

Up until today you got away with poor governance. In a warehouse or a dashboard, dirty data has a grace period. You collect first, fix later. The transform layer is where the sins get cleaned up before anyone makes a decision on them. Governance was always deferrable because there was a margin between bad data and consequence.

Stripe walked through this on stage yesterday during the "off-the-radar" event. They showcases a PaymentIntent call. An agent discovers what it owes, authorises, and settles. There's no account creation, no pricing page, no human clicking confirm.

The safety margin you had in your data warehouse is gone. The data is the action. There's no staging layer between a wrong value and a settled transaction. The agent acts on what you declared, at machine speed, and the money has already moved by the time you'd have caught it.

Governance stops being cleanup you do later. It's the contract you encode up front, because there's no later.

It's time to get serious on governance on all fronts.

It's mid June. Holiday season is around the corner again. Working with enterprise organisations, I'm reminded of a pattern that keeps repeating:

In a twelve-month period, you're actually getting maybe four or five months of actual execution and decision windows. The rest is waiting or absence.

January through early February is slow - people are easing back in. Budgets land in February, roadmaps follow. By late May, holiday mode starts. Through summer, decision-makers aren't there. September hits and teams realise they've lost months. You work hard September through mid-December, then the cycle resets.

A lot of organisations doing digital analytics are obsessed with tracking. Not with making sure it leads to actual business growth.

A perfect schema with no decision attached to it is just expensive storage.

Semantic drift is when the meaning of an event or property changes over time without the name changing.

The classic example: you have an event called user_signed_up.

• At launch it fires when someone completes registration.
• Six months later, a developer adds it to a new onboarding flow.
• A year later it fires on SSO provisioning as well but the name stayed the same while the meaning didn't.

Your dashboards still show user_signed_up. Your data is now a blend of three different things. But nobody actually flagged it.

This is dangerous because it is a silent problem that fundamentally differs from other data problems we're used to. No errors, alerts or broken data pipelines. Just quietly wrong numbers that look right. What's worse is that it compounds. Every chart, cohort, or funnel built on that event inherits the drift. You can't trust historical comparisons because the denominator changed.

The only solution is governance in all its forms. Event versioning, taxonomy ownership, documentation, dependency mapping, etc. It's a process very few people want to own.

People learning about semantic drift usually realise they have a potential blind spot that might have impacted their business for months already.

During last week's webinar on analytics migrations an attendee asked the question: "What makes organisations pull the trigger on a migration?"

My answer was short: "When the pain of staying outweighs the pain of moving."

But there's a follow-up worth addressing. Most organisations overestimate what migration actually costs. The real cost of migration is lower than ever and will keep dropping. Tooling, like Sherpa, exists to automate the heavy lifting. Platforms are increasingly interoperable. The technical barrier is shrinking by default.

That said, migration cost has two components people tend to conflate.

• Reimplementation work - Rebuilding dashboards, reconnecting integrations, revalidating logic. Which I would say is a real cost.

• Cleanup work - Migration forces you to audit everything you've built and rethink it from scratch. That feels expensive. But that work was already owed but you avoided it because nothing forced the issue. Migration doesn't create that debt. It just makes it visible. That's not a migration cost but a lack of maintenance.

So when someone says migration is too expensive, they're often pricing in work they should have done regardless.

Stumbled into an old second hand book shop during my holidays in Athens last week and found myself a first edition (1974) of the book "2020 vision" by Jerry Pournelle.

The cover art caught my eye in the first place but it's the introduction that convinced to take it home.

"... but science fiction can build realistic scenarios that show what might happen, as a result of decisions that are made today.".

Even though this book was written in the 70s, these words couldn't be more relevant today.

For a lot of our clients we step in as a translator between legal and marketing. By the time we enter they usually don't get along. Which makes sense if you look at what they're tasked with:

• Marketing's role is to grow the org
• Legal's role is to protect the org

They lack a common language to book progress. Things we've seen work:

• Make them aware of each other's objectives. Everyone is just trying to do their job. Relax.
• Help legal understand technical terms. Don't yell "server-side" or "client-side". Show what it looks like. Draw it out.
• Help marketing understand which precise elements worry legal. Every architecture is a stack of bricks. Pin-point the problem brick. Design alternatives around it.

“I have sworn a vow of honesty. I’ve no choice but to be blunt.”

— Jakob of Thorn “The Devils” by Joe Abercrombie.

So - when do we introduce a vow of honesty in sales and consulting?

I post a lot on LinkedIn. I use it as others would use Twitter almost. I enjoy writing on the platform because it allows me to voice ideas, have them challenged and develop them as a result. The audience is right, the feedback is instant, and it's where my professional world lives.

But every time I publish something, it disappears into a feed I don't own. LinkedIn decides who sees it, when, and for how long. If the platform changes tomorrow, that content is gone. That bothers me.

Originally I solved this by maintaining a personal and public Notion page. This evolved quickly in some sort of portfolio page that allowed me to organise and find back things I had written that could be useful for customers, the team, etc. But it lacked the character and simplicity I loved so much during the early days of blogging and Tweeting.

So I built Mind, a simple personal CMS that acts as a permanent record of everything I think, build, and publish. It’s my professional archive on the web. It’s simple, it’s fast, it’s boring. And most importantly - it’s easy, not complex.

The name "Mind" is literal. This site is a representation of my mind. The ideas I care and have cared about, the projects I've worked on, the talks and articles that made it out into the world. All in one place.

I wanted it to feel like a micro-blog. No comment sections, no follower counts, no algorithmic feeds. Just content, chronologically ordered. I grew up on Myspace, Noxa and Tumblr and I miss that simplicity. A place where you just drop things and they stay. The stack is intentionally minimal. Next.js and React for the frontend, Supabase for the database and image storage, Tailwind for styling, and Vercel for hosting. A simple admin panel allows me to write and publish. No CMS platform, no WordPress, no dependencies I don't control. The whole process was a fun learning experience, as always.

So now you can find out what’s going on in my mind either here, or on glennv.com

A gentle reminder that more output ≠ more value.

Most meeting notes today are longer and less useful than ever. Used to be one page. Tight and relevant. Now it's three pages of bullet points documenting every comment someone made in passing. Everyone's just running a transcript or an agent through a summariser and calling it done.

Note taking became lazy and the output is mostly slop.

Volume went up and quality went down. Nobody reads them. Not even the people sending them out.

‘In early Terran history, during the dominance of the Sumaturan dynasts, naysmiths were employed by the ruling classes. Their job was to disagree. To question everything. To consider any argument or policy and find fault with it, or articulate the counter position. They were highly valued.’

— Horus rising, Dan Abnett

Sometimes fiction contains wisdom. Even if it is Warhammer 40K fantasy.

Companies should have more "naysmiths" and less people who agree for the sake of safety and career development.

Enterprise organisations have hidden the last decades behind the “we are too big to fail” mentality to ignore and minimise their lack of speed, service and quality.

In a lot of cases this resulted in a culturally accaptable lack of ambition.

The narrative no longer sticks. Timelines for everything just got dramatically shorter and enterprises aren’t able to deal with it. Their structure prohibits them from moving fast. That is even if their culture hasn’t killed the excitement for change already.

The timeline is no longer 10 years. It’s significantly shorter than any of us can estimate atm.

A great reset is coming.

During a dashboard walkthrough, a client stopped us mid-explanation: "Oef, that looks complicated. I'm just going to ask the agent to take care of it."

And it worked. People are moving to conversational analytics.

It’s here. It’s happening.

Is it me or are we going from:

"VC backed organisation with no LT revenue model burning cash on extremely expensive engineers"

to

"VC backed organisation with no LT revenue model burning even more cash on AI agents"

When everything can access the same data we finally start having conversations on what to do next, rather than why my number is different from yours.

"Technology without culture doesn’t work."

— Moonhaven on Netflix.

The way AI articles capitalise every word ("Amazing Article Title With A Punchline") makes me feel like it's 2014 again and I'm writing AdWords ads for my clients.

I’ve been told I’m very bad at corporate politics but that I always speak my mind.

And I take it as a compliment.

• Silence during a deployment gap is a choice to accept churn. Proactivity is a choice to build trust.
• Churn is rarely caused by a generic failure. It is caused by a failure to meet a specific expectation at a critical milestone.

The answer to a problem is rarely "more meetings".

The decades-old "Pull Model" where a human asks a question and a dashboard provides an answer is dying. The next generation of analytics tool will warn you proactively what to pay attention to and what you next move should be.

Blurb/thought I had this week when seeing a number of new ideas being shared in the form of open-source projects:

If there's an upside to AI generated code it is that it’s boosting the availability of open-source code (and thus sharing of ideas captured by that code).

Couple of thoughts on why this happens:

• A lot of projects that are being released are projects people used to get their hands dirty and are PoCs of sorts
• Because AI makes it so much faster and easier to turn an idea into code, people don't feel the need to guard their work or demand payment like they used to
• I guess they also realise that with AI, it’s now very easy for others to replicate what they’ve built anyway

That doesn't mean all the code is good code. But it does mean an influx of ideas that are openly being shared in the form of code. That's a win for all of us.

If I'm reading your technical docs and I can't figure it out because it's not clear (not because it's too technical) and I need to resort to using Gemini or ChatGPT to know WTF I need to do it means your docs aren't good enough.

"Make it as complex as we need to, not as complex as we can."

Picked up during an internal meeting at Human37. This one goes in the manifesto.

I'm intrigued by the fact that some organisation go through crazy hoops to make their reporting work "as it has always has" even though they have migrated to newer and better solutions that reflect reality better.

How many conversations I have had this year talking about retro-fitting Mixpanel or Amplitude reports to a UA session standard simply because exec reporting works that way is mind blowing.

The fact that you can choose conversion windows simply makes people bug. Break free from the standards Google told you to follow 10+ years ago. Reporting can really be custom for your business. Embrace it.

Real-time data ingestion for personalisation is often the first technical requirement we get when doing customer data strategies.

• (almost) Everyone wants real-time data for personalisation.
• (almost) No one is able to translate real-time data in meaningful customer interactions.
• (almost) No one is able to build a strategy/use case that is ROI positive when taking the cost of real-time ingestion into account.

Real-time use cases are what will move the needle marginally once you've done all the rest. For the vast majority near real-time is good enough because they still have so much ground to cover.

Prioritise, mature and make more money by saving money.

I believe that discipline is one of the most underrated traits someone can have, including in business.

The discipline to do the task whether it’s fun or not, because it is the task that needs to be done in order to move to the next step.

It doesn’t mean I like disciplined people because they take care of the annoying work. I like them because they understand that building something inevitably also includes dull work but that they have the ability to deploy discipline against it to get it done.

I can’t believe how undervalued and untapped the potential of loyalty still is today.

Everyone thinks of growth as acquisition. Growth is significantly easier when churn is low and CLV is up.

What PMs can learn from Deadmau5 and the creation of the song "The Veldt".

"The song was initially created during a 22-hour live streaming session in March 2012. The following day, Zimmerman (Deadmau5) found Chris James, who created his own vocal rendition of the song, via Twitter. Zimmerman was impressed with James' vocals, particularly with the lyrical references to the Ray Bradbury story, and confirmed that the official release of the song would include James' vocals." - From Wikipedia.

The discovery was actually captured during the livestream and is now hosted forever on Youtube.

The song currently has 86M streams on Spotify.

PMs - Even though you might think you know better. Listen to your users, you might be missing more gems than you know. You're never that good that you don't need user feedback.

Wishing you all a "The Veldt" moment.

I recently came across Paul Graham's blog of essays.

"Knowledge grows fractally. From a distance its edges look smooth, but when you learn enough to get close to one, you'll notice it's full of gaps. These gaps will seem obvious; it will seem inexplicable that no one has tried x or wondered about y. In the best case, exploring such gaps yields whole new fractal buds." source

The following reminded me of something I call "relentless curiosity". When I'm asked what makes people successfully in the role of solution architect my answer has always been being relentlessly curious. Asking more questions to build an understanding since understanding is required to craft solutions.

When everything connects to everything the risk is that data ends up in platforms for which customers never have given their consent. Think of data captured under analytics consent that's being streamed to marketing and advertising platforms.

Interestingly the vendors consider the lack of guardrails none of their responsibility while everyone is proud of their downstream steaming capabilities. When asked about it (I have :-)) everyone seems to be pointing towards the data warehouse as "the central point where the problem should be managed".

When a vendor says they're privacy centric but state that "privacy should be handled in the data warehouse" what they're saying is "you're on your own". It's a nice way of saying "solve it yourself".

Bottom line - Don't pay too much attention to privacy positioning of vendors. It's a topic that needs to be mentioned and requires some checkboxes from a gtm perspective. Very few are able to actually provide extensive guardrails. Ensure you have your own processes to put guardrails on how data gets piped from one tool to another. Cause you are really on your own.

Most companies don’t really do analytics. They simply do tracking.

And if you’re lucky some reporting.

I wish more organisations would “do” vs “talk about what they’re going to do”. So many people are paid to just do the latter. It’s mad. Most things aren’t that hard.

I want to build things, make progress. And slides don’t count.

The difference between customer support and excellent customer support?

Provide solutions, not only answers.

I don't care that the feature does not work this way, tell me which part of your product might be able to facilitate what I'm actually trying to solve.

The same goes for consulting btw.

If you work in analytics, data or are a data consumer in any way (yes, that includes asking questions to analysts) you need to be familiar with Kidlin's law.

Kidlin's law - "If you write the problem down clearly, then the matter is half solved."

Analytics is the side of the business where everything has been reduced to data points while the business questions we're trying to answer are usually "big", "opaque" or "vague".

Writing down these questions clearly forces you to specify different pieces of the problem and refine them. As a result, it helps everyone to understand which assumptions, hypotheses and data points are required to compose an answer to the questions.

I only recently learned about Kidlin's law even though we've been adopting something similar called "user stories" at Human37 since day 1.

This one is for all the Solution Architect (or whatever the equivalent role) at SaaS companies and was triggered by a SaaS vendor asking me what I thought of their Solution Architect / professional services team.

Note this is a reflection on SaaS SAs. I'm not speaking on behalf of service provider SAs like ourselves here.

As far as I'm concerned Solution Architects in SaaS (or whatever the equivalent role name) - are the "special forces" within SaaS organisations.

• They have a deep understanding of the product, its capacities and limitations. They understand all the tools, why they are used and what their limitations are.
• Unfortunately in most cases they're brought in to turn a bad situation around (this includes a bad sale in many cases).
• Adapt and overcome is what they do when the product hits a limit when executing a client case.
• They consistently get undervalued, are never put in the limelight and rarely get any credit.
• They usually operate like one man armies dealing with different client teams. As a result they need project management, people and technical skills. Most of them are well rounded profiles.
• When they're not working on a client they train. I know SAs who have a backlog of tests they "would like to run" to identify weak spots in the product to anticipate any upcoming scenario.

This is a call for SaaS companies to value their SAs. They are dealing with clients every day and know how the product is being used and misused. If you're in product make sure to have a direct line with them. Unfortunately it's rarely the case that SAs can directly feedback into product. This should change.

SAs are the SaaS heroes that don't wear capes and save the day. They deserve more credit.

I've been lucky enough to work with some stellar teams and have learned a ton from them.

I'm intrigued by the fact that there's organisations who believe they need server-side GTM tracking (or more complex setups) to "regain" X% of their tracking in order to then do ABSOLUTELY NOTHING with the data they regained 🤡.

Know what you build/buy and why. Don't just do it because "people say you need to".

And please, only capture data to use it. Not just to have it.

When starting with a customer data strategy marketers systematically overestimate the value of behavioural data when it comes to driving value for the business.

I've found that organisations who start with use cases that leverage confirmed 1st party data are able to drive results more effectively vs the ones that focus on use cases leveraging behavioural data.

Leveraging existing data from a single 1st party system (e.g: transactional data) can drive more value for an organisation than anonymous (or even identified) behavioural data. That last one will almost always only leads to marginal gains.

My hypothesis is that the reason for this focus on behavioural data is that marketers have always been using with anonymous data gathered through tags and pixels.

💭 Focus on confirmed behaviour vs implied behaviour. Especially when the actions you are taking based on this data are non-existent to start with.

"Banks got robbed because they hold money, companies get robbed because they hold data. So plan for when it happens, not if it happens."

That's an idea I shared with a number of organisations during a series of sessions around data & ethics.

But why am I sharing this today? Because this morning I woke up to a message from The North Face indicating that they suffered a data breach and some of my data might be impacted.

Even though the news is 💩 , it was "good" to read that a process was in place. They shared:

• What happened and how they reacted. Authorities were involved and the breach was submitted to the data governing authority.
• What information could be impacted. It looks like they are applying data minimisation principles to reduce the risk of WHEN a breach happens.
• How this could impact me. -A point of contact to raise questions.

Prepare for when, not if.

📣 Analytics is bigger than GA4 📣

There I've said it.

Every time I go to an analytics conference I'm surprised that 90% of the content is about GA4. The market has never been so diverse as it has been today. Go explore it.