First-party data: the build order most companies get backwards
A founder told me this recently.
“We are having our best months ever. Growth is steady.”
I asked him two questions. What is your retention rate, and how do you know where the growth is coming from?
His answer: “Eh, we just look at overall revenue. No data platform yet. We will add that later.”
He can see revenue. Revenue is the one number that arrives whether or not you built anything to explain it. It lands in the bank account by itself. Everything that explains it has to be built on purpose.
He is right that a platform can wait. He is wrong about what cannot. The thing you cannot add later is not software. It is the record of who did what, kept from the start.
Most companies build first-party data in the wrong order. They buy tools, collect broadly, then find the foundation was never laid. This article argues for the opposite order. Build the spine on day one. Refuse everything else until you can name the decision it changes.
1. Revenue is the only number that arrives on its own
Every other number is constructed. Retention, cohort behaviour, channel contribution, payback period. None of them shows up by themselves. Someone has to decide what to record, then record it.
This is why growing companies feel informed and act blind. The number that is easiest to see explains the least.
Salesforce surveyed 4,450 marketers in June 2026, including 100 in Singapore. 87% of Singapore marketers said they still run generic campaigns. Every single Singapore respondent reported a barrier to personalisation. They named siloed systems and poor data quality.
Ambition is not the missing part. The record is.
💡 Key Takeaway: Revenue tells you something worked. Only your own data tells you what.
2. The only thing you cannot add later is the spine
The spine is two things. A persistent identifier attached to every visitor, known or anonymous. And a raw log of what each one did, timestamped, stored somewhere you control.
That is the whole list. No platform. No vendor.
Here is where founders push back, and they are half right. Modern tools do stitch backwards. The moment a visitor logs in or hands over an email, a customer data platform (CDP) can attach their earlier anonymous sessions to the new profile.
But it can only attach what was already captured. Stitching joins records. It does not invent them. If nobody tagged that visitor in March, March is gone.
So the spine is the one exception to “we will add it later”. Definitions can be rewritten. Models can be retrained. Tools can be swapped in a quarter. History cannot be recovered at any price.
I built funnel tracking infrastructure at Hooq. It cut time to insight by 50%. That speed came from the record existing, not from the reporting layer sitting on top of it.
💡 Key Takeaway: You can rebuild every layer above the spine. You cannot rebuild the months you did not record.
3. Everything above the spine waits for a named decision
Once the spine exists, the instinct is to collect more. Resist it.
Gartner surveyed 405 marketing leaders in 2023. They were using 33% of their marketing technology capability, down from 42% the year before and 58% in 2020. The stated causes were skills, governance and stack sprawl.
The same pattern shows up in the goal itself. Only 14% of organisations had achieved a full view of the customer. More telling: 72% of marketers who believe they need every data point also report that the more they collect, the less benefit they see.
Collecting more is not the same as knowing more. Past a point, it makes knowing harder.
The test fits in one sentence. Name the decision this field changes, and name the person who makes it. If you cannot do both, you are not collecting data. You are collecting maintenance.
4. In Singapore, collecting without a purpose is a liability
Elsewhere, this is an argument about waste. Here it is also a legal one.
Section 18 of the Personal Data Protection Act (PDPA) sets the Purpose Limitation Obligation. You may collect personal data only for purposes a reasonable person would consider appropriate. You have to tell people what those purposes are.
The Commission goes further. It advises organisations to avoid over-collection where there is no business or legal need for it.
Read that against “collect everything now, find a use later”. That strategy is not free, and it is not neutral. Every field you cannot justify is exposure you chose to carry.
💡 Key Takeaway: “We might need it one day” is not a purpose. In Singapore, it is not a defence either.
5. The build order, staged by what you can see about yourself
Stage by behaviour, not by size. You will recognise yourself in one line.
Stage one. One product, founder-led sales, one or two channels.
Build the spine and nothing else. A persistent identifier on every visitor. An event log with timestamps. One written definition of what counts as a customer. This costs close to nothing and fits in a weekend.
Stage two. Repeatable acquisition, a second channel, a team that did not build the product.
Now definitions matter, because more than one person reads the numbers. Agree what counts as active, as churned, as a qualified lead. Record consent next to the data, not in a separate system. Tie every new field to a named decision and a named owner.
Stage three. More than one product or market, and paid spend that has to be defended.
Only now do models earn their place. Lifetime value, cohort decay, propensity to buy. These need history, and you have it, because you built the spine three years ago.
Tooling comes last at every stage. A platform is a convenience layer over a record you already keep. Bought first, it gives you a faster way to produce numbers nobody trusts.
Final Thoughts: build the record first, buy the platform last
The founder who told me he would add the data platform later was solving the wrong problem. The platform was never the hard part. Vendors are queuing up to sell him one.
What he actually decided, without noticing, was that this year’s customer behaviour would go unrecorded. In two years he will want to know which cohort stayed and which channel brought them. The answer will not exist. Not because a tool was missing, but because nothing was written down.
Only 21% of Singapore marketers say they fully use the first-party data they already hold. That is not a tooling gap. It is a discipline gap, and discipline is cheaper than software.
The order is the strategy. Spine first, on day one, before you feel ready. Definitions when a second person needs the numbers. Models when spend has to be defended. Tools last, every time.
If you are about to buy a platform, book a call and let us check the spine first. Or connect with me on LinkedIn and tell me which stage you are in.
A note before you close this tab. The fact that you read this far tells me something. You already sense that the way you’ve been thinking about growth might be incomplete. That instinct is worth following.
Mervyn Chua is a growth-transformation consultant helping founders and CEOs build the strategic clarity and systems to grow in an AI-first world. If this raises questions worth exploring for your brand, let’s talk.
