Skip to main content

What is Return on Data Spend?

Return on Data Spend is the business value an organisation gets back from everything it spends on data — collection, storage, transformation, governance and consumption — measured against that total cost rather than against infrastructure cost alone.

Most organisations can produce the cost half to two decimal places and cannot produce the return half at all. That asymmetry is the problem: what gets measured is the spend, so what gets managed is the spend.

What most organisations actually measure​

Cloud consumption. Licences. Engineering headcount. Project budgets. Storage growth.

Every one of those is an input. They describe what the estate costs to own and say nothing about what it produced. An organisation can reduce all five in a year and be no better informed than it was before — and often less, because the reductions usually come out of the consumption end rather than the machinery.

Why the return half is the hard half​

Data spend is invoiced, so it is easy to total. Value from data arrives as things that did not happen: a question answered without a project, a reconciliation meeting that was not needed, a definition nobody had to re-agree. Absences do not generate line items.

That does not make them unmeasurable — it makes them uncounted. The four measures on Enterprise Data Efficiency exist precisely to count them.

What counts as return​

  • A business question answered without commissioning a new pipeline
  • A definition reused rather than rebuilt for the next consumer
  • A board number that did not have to be reconciled between two reports first
  • A report retired because something upstream made it redundant
  • An AI answer a business user could act on without having it checked by the data team

The part of data spend that produces nothing new​

The largest single category of data expenditure in most enterprises is not storage or compute. It is the repeated re-establishment of business meaning — the same definition of revenue, customer or margin implemented again in the next pipeline, the next mart, the next semantic model, the next dashboard and the next AI project.

Each implementation is individually defensible. Collectively they mean the organisation pays many times for one answer, and still gets several different versions of it. That is spend with no return attached, and it grows with every new consumer rather than amortising across them.

How to work it out from your own figures​

Three calculators on this site take your numbers rather than ours:

Your Enterprise Data Efficiency Score combines them into a single number, with the arithmetic shown, and it is yours whether or not you ever adopt CryspIQ®.

Why this is not project-level data ROI​

Project ROI asks whether one initiative paid back. Return on Data Spend asks whether the estate does.

The two routinely disagree. An estate can consist entirely of projects that each cleared their business case and still return poorly overall, because every one of them budgeted to establish meaning that an earlier project had already established. Nobody approved the duplication; it is what happens when meaning is a property of each consumer rather than of the data.

The CryspIQ® perspective​

Return on Data Spend improves by removing repetition, not by adding capacity.

CryspIQ® decomposes records on entry and holds the business definition with the stored data, so meaning is established once and every consumer inherits it. The effect on the numerator is direct: the second, fifth and twentieth consumer of a business term cost nothing to define, because they read a definition that already exists.

Reported outcomes are argued separately, with their working, on productivity, time to value and cloud cost.

Frequently Asked Questions​

Is Return on Data Spend the same as data ROI?​

No. Data ROI is usually calculated per initiative, asking whether one project paid back. Return on Data Spend asks whether the estate as a whole does. The two can disagree sharply: an estate can consist entirely of individually justified projects and still return poorly, because each one paid again to establish business meaning that an earlier project had already established.

How is it different from cloud cost optimisation?​

Cloud cost optimisation reduces the denominator. It makes the same work cheaper to run without changing how much work there is. Return on Data Spend can be improved from either side, and the larger gains are usually on the numerator: removing the repeated effort of redefining the same business terms, rather than paying less per unit of that effort.

Can it be measured without replacing the data platform?​

Yes. Every input is already available inside most organisations: total data spend, the number of distinct definitions in use for key business terms, the time between a business question and a trusted answer, and the proportion of transformation logic that reimplements logic existing elsewhere. None of those require a platform change to count — and counting them is usually the first time the return has been stated as a number at all.

What is a good Return on Data Spend?​

There is no published benchmark, and anyone offering one should be asked where it came from. The useful comparison is against yourself: the same measures taken twice, a year apart. A rising number means new consumers are inheriting meaning rather than rebuilding it.