The real cost of a system of record nobody trusts

A system of record nobody trusts costs two things you can actually measure: hours, as reps and leaders manually reconcile what the CRM will not tell them straight, and forecast error, as real decisions get made on a number nobody quite believes. The licence fee is the smallest line on that bill.

Where the hours go

Every workaround a team builds around a CRM it does not trust is time paid twice: once to do the work, once to verify it somewhere else.

A rep checks the deal stage against their own memory before a call, because two owners have touched the account since the field was last accurate. RevOps reconciles the pipeline number by hand before the forecast call, because the raw export never matches what the leader expects. Marketing pulls a segment into a spreadsheet because the lifecycle stages in the CRM do not match what the campaign actually needs to send to.

None of that shows up as a cost centre. It shows up as a slower week, repeated every week, for as long as the system stays untrusted.

Validity's State of CRM Data Management in 2026 report, released 25 August 2026 from a survey of 500 B2B and B2C marketing professionals, found that nearly a third of teams now spend six or more hours a week just fixing and reconciling CRM data rather than doing revenue work.

Where the forecast error comes from

A forecast built on a system nobody trusts is a story told in confidence intervals, not a number drawn from the record. We have written before about what it takes to build a forecast you can defend: exit criteria on every stage, auto-logged activity, conversion rates drawn straight from the CRM rather than a private spreadsheet.

When that discipline is missing, the forecast is only as good as the last person who reconciled it by hand, and that person changes every quarter. The same Validity survey found that 62% of organisations report losing revenue directly because of poor CRM data quality, a direct line from untrusted records to a number the business acts on.

The board question that exposes this is simple: "how do you know?" A leader with stage conversion data and time-in-stage benchmarks drawn from the CRM answers it calmly. A leader with the spreadsheet open in another tab does not.

The one number that predicts both

We use a single diagnostic across client engagements: the percentage of open opportunities updated in the CRM in the last seven days. We first published the thresholds in our piece on why CRMs become graveyards, and they hold here too, because hours and forecast error move with the same number rather than separately.

Above 80%, the record is current enough that reconciliation is occasional, not weekly. Between 60% and 80%, reconciliation becomes a routine, unbudgeted tax on every rep and every report. Below 60%, the CRM and the forecast the leader defends in the boardroom have quietly become two different documents, and closing that gap is manual, recurring work someone is already doing without a line item for it.

What the external research says, and where it is misquoted

The most repeated figure in this space is that poor data quality costs an organisation an average of $12.9 million a year. It is real, and it is still directionally right, but it is not new: it comes from Gartner's Magic Quadrant for Data Quality Solutions, published in July 2020, based on a survey of reference customers across sixteen vendors. Most content answering this exact question recycles the number without saying when it is from, which makes a six-year-old finding read like this year's research.

The genuinely fresh data point is Validity's, and it is dated correctly above: a 2026 report, a named survey size, a named date. That distinction, checking the figure, the organisation and the report against each other rather than trusting whichever page paired them, is the whole discipline this piece is trying to demonstrate.

The third cost: the tools bought to work around it

The most common response to a system of record nobody trusts is not to fix it. It is to buy a second one: a reporting layer that sits on top and promises a clean view, a data enrichment tool to patch the gaps, a BI dashboard that recombines the same records into something the leader will actually open.

In our audits, that pattern shows up almost every time, and it never actually solves the trust problem, because the new tool inherits the same underlying records. It adds a licence, a sync job that can silently break, and one more place the number can disagree with itself.

The tell is a stack where two or three tools each claim to be the source of truth for the same metric. None of them are. The CRM still is, whether anyone trusts it or not, because it is where the transaction actually happened.

How to cost your own instance

Do not borrow our number or anyone else's. The inputs are specific to your team, and the formula is simple enough to run in a spreadsheet this week.

What to do this month

Run the weekly-update-rate check today. It takes one saved search and gives you the headcount for the table above.

Then engineer the hygiene loop before anything else. It is usually the cheapest of the four loops we describe in the graveyard piece, and it is the one that buys back the most hours per pound spent. Everything downstream, forecast confidence included, follows from a record people stop having to double-check.

For teams running the system of record itself as the constraint rather than the workflow around it, our CRM as system of record guide and the Revenue Signal Field Manual cover the build in more depth than an answer post can. Our own approach to the diagnosis itself, before any of that build work starts, is the diagnosis is the deliverable.

The licence is never the expensive part. The expensive part is a team that has quietly agreed to stop trusting the system they are all still paying for.