Someone presents a number in a meeting. Before anyone reacts to what the number means, someone else asks, “Where did this come from?” The meeting stalls while two people compare it against a different report, find a small mismatch, and spend the next ten minutes debating whose version is right. Nobody has questioned the strategy yet. They are still arguing about the input.
This scene repeats often enough that it shows up in industry research, not just anecdotes. Precisely and Drexel University’s 2025 Data Integrity Trends report found that 67% of organizations don’t completely trust their data for decision-making, up from 55% the year before. A widely cited KPMG and Forrester Consulting study of data and analytics leaders found only 38% had high confidence in their customer insights, and only about a third trusted the analytics generated from their own business operations.
Trust is going down even as spending on data and analytics goes up. That gap is worth understanding before trying to close it.
The problem is rarely the data itself
When trust breaks down, the instinct is to blame data quality. Sometimes that is the cause, but more often the data is fine and the problem is what happens to it after it is collected.
- The same metric has more than one definition. Finance calculates “active customers” one way, product calculates it another way, and both dashboards are technically correct by their own logic.
- Nobody can trace a number back to its source. A figure shows up in a deck with no visible path to the table, query, or transformation that produced it, so verifying it means asking around instead of looking it up.
- Spreadsheets fill the gaps dashboards leave. When a dashboard cannot answer a specific question, someone builds a manual workaround, and that workaround quietly becomes the real source of truth for one team while the official dashboard stays untouched by everyone else.
- Access is inconsistent. If two people with different roles can see different slices of the same report without knowing it, they will eventually compare numbers that were never meant to match.
What low trust actually costs
The TheyDo 2025 survey of senior decision-makers across the US, UK, and Netherlands found that 77% of executives admit they only sometimes or rarely question the data they rely on daily, even though 67% worry that leaning on static dashboards risks missing real opportunities. That is the quiet cost of a trust gap. People do not necessarily stop using the data. They stop questioning it, which is arguably worse, because bad numbers stop getting caught before they turn into bad decisions.
The louder cost is the one every data team already feels: meetings that turn into reconciliation exercises, decisions that wait on someone to manually confirm a figure, and a data team that spends more time defending numbers than producing new ones.
What actually rebuilds trust
More dashboards do not fix this. Neither does a data quality initiative on its own, if the underlying definitions and lineage stay scattered. What closes the gap is making three things true at once:
- One certified definition per metric. “Active customer” should mean exactly one thing, owned by one team, used everywhere it appears.
- Visible lineage on every number. Anyone looking at a figure should be able to trace it back to the table and logic behind it without filing a request.
- Access enforced at the point of the question. Role-based permissions applied at query time, not just at the report level, so what someone sees is consistent with what they are allowed to see.
This is the architecture QuaerisAI is built around. Every answer maps to a certified metric definition, comes with a source citation, and is logged in a full audit trail, so a number can be traced back to the table and query that produced it in one click. Access is enforced at the moment a question is asked, based on role, not after the fact. None of this requires moving your data. QuaerisAI connects to Snowflake, BigQuery, Databricks, Redshift, and other sources in place.
Teams that move to this model tend to stop treating “where did this come from” as a rhetorical question. Self-serve adoption has climbed from roughly 30% to 60% of business users in QuaerisAI deployments, largely because people no longer need to double-check a number with the data team before they can act on it.
A starting checklist, regardless of platform
If a full platform change is not on the table yet, these are worth doing on their own:
- Name one owner for every metric that appears in more than one report
- Require a visible source and definition next to any number in a board deck
- Find the shadow spreadsheets your team already relies on and ask why the official dashboard doesn’t cover that case
- Check whether access controls are enforced at the data layer or only at the report layer
Trust in data is not restored by asking people to believe harder. It is restored by making it possible to check. Talk to us to see how QuaerisAI builds that checking in by default.