Bad data only gets fixed when the business can see what it is costing

September 16, 2026

Bad data is easy to tolerate when its cost is invisible.

A duplicate record, inconsistent product code, stale field or broken lineage rule can look like a technical defect. Inside the business, however, those defects accumulate into something far more expensive: delayed projects, manual rework, missed sales, poor decisions, regulatory exposure and teams that stop trusting the systems they are meant to use.

Recent UK data-leader discussions convened by The Leadership Board surfaced a consistent commercial reality. Data quality becomes easier to fund when leaders can connect the problem to business value. When the conversation stays at the level of completeness scores, duplicate rates or governance maturity, sponsorship is harder. When the organisation can see the money, time or opportunity being lost, priorities change.

That should change how data vendors sell.

The data quality problem is often priced into the business without being named

Many enterprises already pay for poor data. They just do not always label the cost as a data problem.

An analyst spends hours reconciling two reports. A project is delayed because the source data is incomplete. A commercial team manually checks numbers before presenting them. A customer receives the wrong information. An operations team keeps extra inventory because demand data cannot be trusted. A transformation programme adds another layer of controls because nobody is confident in the underlying definitions.

Those costs sit in different budgets and different teams, which makes them difficult to see as one problem. The technical data issue may be small. The business consequences are distributed.

This is why the strongest data-quality business cases often start outside the data function. The buyer needs to quantify where bad data changes an outcome, not merely where a rule fails.

The roundtable examples were commercial, not theoretical

Several individual organisations shared concrete examples of this shift. One product business described a data-quality issue that could be tied to roughly £200,000 in stock and sales impact. Another organisation described missing data updates contributing to annual losses of several million pounds. A separate automation initiative was associated with an annual saving of around £800,000.

These are individual examples, not market-wide benchmarks. Their importance is in the mechanism. Each organisation was able to move the data conversation from an abstract improvement programme to a business outcome that senior leadership could understand.

The moment poor data can be expressed in pounds, hours, risk or lost opportunity, it stops being a back-office quality issue and becomes an executive decision.

That is closely connected to the broader funding challenge explored in why proving business value is becoming a deal breaker for data vendors. The technology may be credible, but enterprise buyers still need a reason to prioritise it over every other credible investment.

Data quality metrics are useful only when they lead to a business decision

Completeness, accuracy, validity, uniqueness, consistency and timeliness all matter. But a dashboard full of data-quality scores can create the illusion of progress while leaving the business indifferent.

Senior leaders discussed the value of scorecards, quality matrices and threshold-based reporting. The strongest versions of these approaches were tied to priority datasets and business needs rather than applied indiscriminately across every possible field.

That distinction matters. Not every defect deserves the same investment. A missing field in a low-value archive is not equivalent to an incorrect margin calculation used in a board report. A vendor that helps the buyer prioritise quality according to materiality will usually create more value than one that simply exposes more defects.

The most useful question is not “how many data issues do we have?” It is “which data issues are changing decisions, delaying work or creating avoidable cost?”

Visibility changes behaviour

One of the recurring challenges in the discussions was business engagement. Data stewards are often time poor. Subject matter experts have client work, operational responsibilities and revenue targets. Asking them to attend another governance forum can feel like additional administration.

The engagement problem becomes easier when the business can see how poor data creates more work for them. Repeated cleansing, mapping, reconciliation and manual checking are not data-team inconveniences. They consume the time of the very people who are reluctant to engage with governance.

This creates a useful vendor angle. Do not sell stewardship as another responsibility that buyers must persuade the business to accept. Show how the operating model reduces the recurring friction that poor data already creates.

That is also why data literacy and ownership matter. We have previously explored why data literacy is moving up the buying agenda. People are more likely to care about quality when they understand how their actions affect downstream decisions and when the consequences are visible in their own work.

The vendor should help build the economic model

Data vendors often arrive with a technical ROI model: fewer manual tasks, faster pipelines, lower infrastructure cost, reduced tool sprawl. Those measures are useful, but they are only part of the value.

The more powerful commercial case often sits one level higher. What happens to revenue, margin, customer experience, working capital, project delivery or risk when the data becomes more trustworthy?

For example, a quality platform may detect missing inventory attributes. The technical benefit is better completeness. The business benefit may be fewer stock-outs, fewer manual corrections or more accurate replenishment decisions. A semantic layer may standardise calculations. The technical benefit is consistency. The commercial benefit may be faster executive decisions and less time spent debating which number is correct.

Vendors that help buyers make that translation are easier to fund because they give internal champions a stronger story to take to Finance and the executive team.

What a stronger data-quality business case looks like

Technical issue	Business consequence	Commercial proof to capture
Incomplete or inconsistent records	Manual correction and slower operations	Hours spent, delay avoided, process cost
Conflicting metrics	Decision delay and loss of trust	Reconciliation time, duplicated reporting, decision cycle
Stale data	Missed sales or poor planning	Revenue impact, inventory impact, forecasting error
Weak lineage	Risk and expensive investigation	Audit effort, remediation time, control cost
Poor metadata	Low reuse and duplicated engineering	Build time, repeated work, adoption levels
Unowned data	Persistent defects and slow resolution	Issue age, escalation volume, time to resolution

Do not promise perfect data

Enterprise buyers know that perfect data is unrealistic. A pitch built around eliminating every defect can therefore sound detached from operational reality.

A stronger proposition is prioritised trust. Help the organisation identify which datasets and metrics are material, set acceptable thresholds, surface exceptions and create a clear route for remediation. That allows the buyer to direct scarce resources toward the problems that create the greatest business consequence.

One roundtable discussion described a quality matrix used before projects begin so teams can assess whether basic data requirements are good enough to support delivery. That is a practical model because it moves quality upstream. The cost of discovering bad data before a project starts is usually easier to manage than discovering it after timelines, people and budgets have already been committed.

The strongest sales conversations begin with the cost of doing nothing

Data vendors understandably want to demonstrate product capability. But enterprise prioritisation is often driven by comparison. The buyer is deciding whether your project deserves funding now, later or not at all.

That makes the cost of doing nothing central to the sale.

How much manual reconciliation continues every month? How many projects start with avoidable data discovery? How often are business teams rebuilding the same definitions? How much working capital, revenue or staff capacity is affected? What risk remains because the organisation cannot trace or trust the data behind an important decision?

These are not scare tactics. They are the commercial baseline against which the investment should be evaluated. Our earlier analysis of the boardroom story that gets data programmes funded points to the same requirement: technical merit needs to be translated into executive consequence.

What data vendors should change

Start discovery with business pain, not platform features

Ask where teams are reconciling, reworking, delaying, checking and escalating. Those behaviours often reveal the real economic cost of poor data faster than a generic maturity assessment.

Make materiality part of the proposition

Show buyers how to distinguish a critical quality issue from a low-impact defect. Prioritisation is a feature of a mature operating model, not a compromise.

Give internal champions numbers they can take upstairs

Build business-case outputs into assessments and proofs of value. Measure time saved, incidents avoided, decisions accelerated and commercial outcomes affected.

Prove that governance reduces friction

Buyers do not need another layer of process for its own sake. Demonstrate how ownership, issue management, standards and automation reduce recurring work for the business.

Bad data becomes urgent when the cost becomes visible

The enterprise data market is full of technically credible solutions. The vendors that win budget are increasingly the ones that connect technical capability to an outcome the business already cares about.

Data quality does not need to be sold as a moral obligation or a maturity milestone. It can be sold as a practical way to reduce waste, protect revenue, speed decisions and make expensive transformation and AI investments more dependable.

The Leadership Board gives technology vendors access to senior enterprise buyers while these priorities are still being shaped. If your proposition can turn data-quality problems into measurable business value, that is the conversation worth having before the budget is allocated elsewhere.