A Single Source of Truth When Data Lives in Multiple Tools

The meeting you've already had

Two people open two reports. Sales says the month did £142,000. Finance says £128,000. Both are pulling from systems they trust. The next forty minutes go to working out whose number is real, and the decision that needed making — hire, hold, spend — quietly waits another week.

That meeting is the symptom. The cause is that your data lives in 10+ tools, and each one quietly answers the same question a slightly different way. A single source of truth fixes the argument at the root: instead of reconciling numbers after the fact, you decide once where each figure comes from and what it counts.

It is worth being precise about the term. As ThoughtSpot puts it, a single source of truth is "a unified system for managing decision-critical data across an organisation." Profisee frames it as "a data repository where the most reliable, up-to-date and complete version of business data lives." Both point at the same outcome — one place that wins when two systems disagree.

Why fragmented data produces conflicting metrics

The scatter is normal, not a failure on your part. BetterCloud's 2026 data puts the average company at 106 SaaS applications in 2024 — down from a 2022 peak of 130, but still more than a hundred places where a number can live. CRM, accounting, the ops sheet, the booking system, email, support, the warehouse export. Each was bought to solve one job well, and each keeps its own version of the truth.

Conflicting metrics between teams almost never come from one tool being "wrong". They come from definitions drifting apart:

• Revenue — does it count the invoice date or the payment date? Does it net off refunds? Stripe and your accounting package will answer differently, and both will be internally correct.
• An active customer — anyone who has ever bought, or anyone who bought in the last ninety days? Marketing and finance rarely mean the same thing.
• A lead — a form fill, a qualified conversation, or a name in a list someone imported in 2023?
• Timing — most exports run on a schedule, so a Monday-morning sales number and a Monday-morning finance number may simply be looking at different moments.

None of this is dishonest. It is the predictable result of asking ten tools the same question without ever agreeing on the question. That is why a tidier spreadsheet doesn't help — the spreadsheet inherits the same ambiguity. The fix has to happen one level up, in the definitions.

What it actually costs before you fix it

The obvious cost is the reconciliation meeting. The quieter cost is everything that meeting displaces. McKinsey's well-known work on the social economy found interaction workers spend close to 20% of the working week just hunting down internal information. That time doesn't show up on any invoice, which is exactly why it survives — nobody owns the line item.

There is a trust cost too, and it compounds. Once a team has been burned by a number that turned out to be wrong, they stop trusting the dashboard and rebuild their own in a corner. Now you have eleven sources of truth instead of ten, and the next disagreement is harder to settle because nobody agrees on which screen to look at. The single-source-of-truth question in 2026 is less about storage — cloud warehouses are cheap and good — and more about restoring confidence so people stop maintaining shadow reports.

How to build a single source of truth dashboard

The work splits into four moves. Skip any one and the dashboard quietly loses trust again within a quarter.

1. Map the tools and find the clashes

List every system that holds a number anyone reports on — CRM, accounting, ops, the spreadsheets too. For each headline metric, write down where it comes from and how that tool defines it. This is where the real disagreements surface, usually before you've written a line of code. You are not unifying data yet; you are finding the dozen places where "revenue" or "active" means two things.

2. Agree the definitions — once, in writing

This is the part teams want to skip, and it is the part that actually matters. For each metric, one definition wins: revenue is recognised on payment, net of refunds; an active customer is one who purchased in the last ninety days; a qualified lead is one a human has spoken to. The definitions are a business decision, not a technical one — finance, sales and ops have to sit in the same room and settle it. The dashboard simply enforces what they agree.

3. Pipe everything into one place

Now the plumbing. Each source connects — by API where one exists, by scheduled export where it doesn't — into a single store that becomes the golden record. Shopify describes this neatly as one hub that consolidates data "into one location, also known as a golden record". The store reconciles the agreed definitions as data arrives, so by the time anything reaches a chart, "revenue" already means the one thing you decided it means. Profisee's point is the practical one: "when everyone works from the same data sets, dashboards actually show reliable numbers."

4. Put it on one screen people will actually open

One dashboard across CRM, accounting and ops — the numbers everyone argues about, defined once, refreshed on a clear schedule, with the last-updated time visible so nobody wonders if they're looking at stale data. The win isn't visual polish. It's that when sales and finance both open it, they see the same £-figure and move straight to the decision.

Buy a tool, or build the screen?

Off-the-shelf business intelligence platforms — Power BI, Looker, Metabase and the rest — are genuinely good, and for many businesses they are the right answer. They connect to common sources and draw charts well. We'll say plainly: if your data sits in two or three mainstream tools with clean connectors and your definitions are simple, a configured BI tool will get you most of the way, and a bespoke build would be over-engineering.

The build case appears when the off-the-shelf path stalls, and it usually stalls in the same places:

• One or more of your critical systems is older, niche or homegrown, and no connector exists.
• Your definitions need real logic — proration, multi-currency, deduplication across systems that don't share an ID — not just a dropdown in a settings panel.
• You need it where work happens: a number pushed into Slack each morning, or an alert when two systems drift apart, rather than a dashboard someone has to remember to open.
• The reconciliation itself is the hard part — matching the same customer across CRM, billing and support when each holds a different spelling, email or ID.

That last one is the quiet killer. Unifying data from multiple tools is mostly an identity problem: deciding that this record here and that record there are the same customer. Generic tools assume you've already solved that. Often you haven't, and solving it is the actual project.

What good looks like six months on

You can tell a single source of truth is working by what stops happening. The reconciliation meeting disappears. Nobody emails a screenshot of "the real numbers" because the dashboard is the real numbers. New hires learn one screen instead of inheriting one person's private spreadsheet. And when someone challenges a figure, the answer is a definition you agreed in writing, not a forty-minute archaeology dig.

It is also honest to say what it won't do. A single source of truth won't make a bad number good — if your CRM data entry is sloppy, the dashboard will faithfully show you sloppy data, just in one place instead of ten. It surfaces the truth; it doesn't manufacture it. That clarity is usually the first real benefit: you finally see, plainly, where the underlying data needs cleaning.

Starting small beats boiling the ocean. Pick the three numbers that cause the most arguments — usually revenue, pipeline and one operational metric — unify those, prove the screen, then widen it. A dashboard that nails three trusted figures earns the right to grow. One that tries to show everything on day one tends to show nothing anyone believes.

Where to begin

If your data lives in 10+ tools and your teams keep arriving at meetings with different numbers, the first task isn't choosing a chart library — it's the definition map. Sit the people who own each number in a room and write down, for each metric, where it comes from and what it counts. You may find the disagreements resolve before any software is built. If they don't, you'll at least know exactly what the dashboard has to reconcile, which is the whole job.