The valuable data in your business is not the data you manage. It is the residue of doing the work. Not the systems you bought to hold information, but the record that accumulated as a by-product of being where you are, doing what you do, for as long as you have done it.
Every organisation that has operated for decades has quietly accumulated a record of observed reality in its own domain: what actually happened, what actually failed, who actually delivered. Almost none of them can query it, and almost all of them assume it is worthless because it is messy.
The one thing a competitor cannot buy
Software can be bought. Talent can be hired. Market data can be licensed. A forty-year record of what actually happened in your sector, gathered by being present while it happened, cannot be acquired at any price. Three things make it different:
- It is longitudinal. Value comes from depth in time, not volume. A competitor starting today is forty years behind and cannot close the gap by spending. The only way to acquire it was to have been there.
- It records behaviour, not claims. Most industry data is what organisations say about themselves. Yours is what you observed them actually do, repeatedly, under conditions you controlled. That is a different class of evidence.
- It comes from a position. You hold a vantage point, whether regulatory, contractual, transactional or physical, that generates the record automatically. The position is the moat; the data is the yield.
The vantage point differs by business model, and so does the prize. Certification, inspection and audit bodies hold decades of non-conformities, corrective actions and repeat offending, the only independent record of how well a supply chain actually conforms. Membership and trade bodies hold the sector's aggregate picture, and think of themselves as a lobby rather than the evidence base. Asset owners hold decades of condition, intervention, failure and renewal, scattered across maintenance systems, GIS and contractor reports that have never been joined. Contractors and consultants hold every tender priced, won and lost, beside what the work actually cost and how long it took, and almost nobody joins the two halves. Regulators hold everyone's submission side by side, as PDFs. Professional services firms hold their matter history, and knowledge management is the perennially unsolved problem, because it is a retrieval problem.
It is not a data asset. It is a franchise on observed reality, and unlike a data asset, it cannot be replicated by anyone with a budget.
Whose permission do you need?
This is the question that governs the whole exercise, and the one that gets skipped. Sort what you hold into bands before doing anything else, because the answer changes completely from one band to the next.
Already public (P0) is what you published yourself: registers, directories, published reports. No permission is required, and it is frequently forgotten precisely because it is public. Your own work product (P1) is your findings, assessments, decisions and history. Nobody's permission is required at all, and it is almost always the most under-exploited band by a distance. Data supplied to you (P2), such as returns, submissions and test data, is confidential when identifiable, and defensible in aggregate or anonymised against a stated threshold policy. Co-produced data (P3), created jointly in a shared platform or process, has an owner nobody has ever decided; that must be settled, not assumed. Locked data (P4) is identifiable, commercially sensitive or held under a confidentiality obligation that is itself the source of your credibility. Spending it is an asset sale, not a strategy. And then there is the void: the part of your value chain where the record simply stops. It is the largest prize and the largest piece of work, and it is worth naming honestly rather than glossing.
The permission problem blocks about half. Most organisations behave as if it blocks everything. P0 and P1 together will usually support a substantial set of products and need nobody's permission, yet the conversation stalls at P2 and P4, and the whole idea is shelved as legally impossible. Someone says "we cannot touch customer data", everyone agrees, and nothing is built from the material that was never constrained in the first place. The better course is to sort the estate into bands first, build from the unconstrained bands straight away, and answer the permissions question for the constrained ones properly, in parallel.
Why this has not happened already
Usually the opportunity has been spotted before, sometimes years before, sometimes in a report the organisation commissioned itself. Understanding why it stalled matters more than restating the prize. The data is not in a usable state: findings from the 1990s are not a queryable dataset, and returns are spreadsheets. Nobody owns it: it sits across operations, compliance, IT and commercial, and work with four part-owners and no single owner does not start. The governance question is unanswered: not "can we technically" but "may we lawfully, and what does it do to our standing". And it looks like becoming a software company, which for most organisations would fail. The capability is rentable; the position is not.
A model of your world, not a report about it
Dashboards describe what happened. What an organisation needs is a structure that holds what exists, how it connects and what may be done about it, so that decisions run on the model rather than on somebody's recollection. It has four parts. Objects are the things the business is actually about, such as assets, suppliers, obligations, batches, matters and sites, named once and defined once. Links are how they relate: this supplier made that batch, which went to that asset, under that obligation. The relationships are where the answers live, and they are what documents destroy. Actions are what can be done, by whom, under what authority and against which clock. Learning is the part that compounds: every decision made on the model becomes evidence in it, so the next programme starts where the last one finished.
The document then becomes an output, not the record. A certificate, a report or a submission is printed from a governing model rather than typed and then filed as the truth. Software is replicable in months. A structured model of a domain, refined over time, is not.
We ran it on ourselves first
We tested the method at our own cost, on the hardest case we could find: the complete national record for development consent. That meant 268 applications and more than 154,000 documents from the published Planning Inspectorate record, extracted and structured, and an obligation model with 141 requirements traced to their source provision [1].
The question worth asking is not "what should you do with your data". It is "you have known about this for years: what stopped it?" That answer is where the work actually starts. It is also where Synaform's Data service starts: connecting the record you already hold and modelling it in Playground, beginning with your own work product, which needs nobody's permission.
Sources
- [1] Synaform internal counts as at 8 September 2026: 268 applications and 154,000+ documents, a Synaform extraction from the published Planning Inspectorate record; 141 obligations traced to source, from the Synaform DCO master process model. Source: deck slide 18, "Provenance". The Consenting and DCO delivery Solution page (15 Oct pack) uses 154,723 documents across 268 projects from the DCO Explorer.