Seven Deadly Claims 1/7: "We Have a Unified Data Model."
Almost every software company has a unified data model slide. One box across the bottom labeled "unified data model" or "common platform," with every product sitting on top of it. In my experience, it's rarely true. The truth is that it's usually <partially> true, which is harder to catch. As investor/acquirer you should care a LOT about this. It has incredibly important EV, AI, growth, margin and defensibility implications.
Candidly, over the last 15+ years, PE firms could afford to ignore or downplay this Deadly Claim in diligence because as long as we can produce Rule of 50 performance, who cares? There will be a buyer at 20-25X + EBITDA in 5 years, right??
What the slide usually describes is shared authentication and authorization, a common customer or tenant record, and a handful of shared services like notifications, audit logging and billing. Underneath, each product still owns its own database and its own definition of the core data objects. In fact, a customer in one product isn't the same record as a customer in another. Data moves between them through nightly syncs, exports, integrations someone built one at a time or NOT AT ALL.
That's a shared platform. It isn't a unified data model. The difference plays an enormous role in whether the product cross-sell plan works - not to mention credibility with the buyer. It is also a good indicator of what remains to be done to get it where it needs to be done (i.e. R&D allocations to tech debt vs. innovation). Most importantly in 2026, is the AI strategy real or surface level nonsense?
I think about it as four levels:
Shared login. Single sign-on across products. Supports a suite story in the deck, not in the product.
Shared services. Identity, tenancy, notifications. Lower cost, no cross-product workflow value.
Synced data. Core objects copied between products on a schedule. Cross-product reporting works, but with lag and reconciliation headaches.
Unified data model. One definition of the core objects, read and written by every product in real time.
Most companies claiming level four are at level two or three.
The fastest test I know is to ask what happens when an existing customer buys a second product “on the platform?” Walk me through it, step by step – show me. If the answer is "we turn it on," ask them to show you. If the answer involves a data migration, a services project, or a mapping exercise, you've learned where they really are.
Then ask three follow-ups and ask for a realtime demonstration to prove it.
Does data entered in one product show up in another immediately?
Can a single workflow read and write across two products, in production, for a real customer?
Can your AI features see and use data from every product, or only the one they sit in?
That last question is the one that matters most right now. AI and agentic workflows can only reason across the data they can reach. A company with fragmented data will ship fragmented AI. Crucially it’s also not going to generate AI specific revenue – ever. The value for the customer won’t be there, I guarantee it.
You will get a bunch of push back from the portcos who know they can’t do this. The CEO, CFO and bankers will protest that you’re nitpicking and provide cover for the poor CPO/CTO/CCO tasked with responding to your diligence request. There’s probably a formula to be created for the number of creative reasons why this can’t be shown, won’t be shown, or doesn’t matter that should be correlated to the discount you should apply to the valuation.
Don’t let it go and don’t allow the clock to be run out. You have leverage, use it.
Next week: "We Integrate with Hundreds of Systems."
