If You Can't Trace a Number, You Can't Defend It

Emily Goh · 2026-10-30

A board member points at one line in your deck. "Pipeline was 4.1 million last quarter. Now it's 3.4. What happened?"

There are only three honest answers. The business changed. Somebody changed the records. Or somebody changed what counts as pipeline. If you can't say which, in the room, the number stops being evidence and becomes an opinion.

That is what traceability is for.

What data traceability means

Data traceability means you can walk any number back along two trails.

The people trail: who did what. Who created the records behind the number, who moved them, who changed the amount, the stage or the close date, and when.

The definition trail: which rule counted them. Which definition of the metric was in force on the day the number was produced, who approved it, and what it replaced.

You need both. The people trail without the definition trail tells you that 40 deals moved, but not whether they should have counted in the first place. The definition trail without the people trail tells you the rule, but not why the records stopped matching it.

Who did what: accountability you can see

You can't hold anyone accountable for a change nobody can see.

That sounds like blame. It's the opposite. When the trail is missing, the meeting fills the gap with guesses, and guesses land on whoever is in the room. When the trail exists, nobody has to guess. You can see that the close dates moved on 12 deals the day after a territory change, and the question becomes "what did the territory change do?" instead of "who broke the pipeline?"

Accountability isn't about catching people. It's about knowing who to ask. (Your Pipeline Meeting Is a Blame Ritual. It Doesn't Have to Be. is what that meeting looks like without the trail.)

Which definition: memorialize it, or lose it

Definitions change for good reasons. Marketing raises the MQL threshold. Finance starts excluding services from bookings. Sales adds a stage. Each change is reasonable. Each one quietly breaks every comparison with the months before it.

A definition that lives in someone's head leaves when they do. A change agreed in a meeting or a chat thread is gone in a month. Writing it down once isn't enough either: you need the version history, because last quarter's number was counted under last quarter's rule. (The Definition Drift Tax is what that missing history costs, quarter after quarter.)

So memorialize it. Every definition change gets one entry, with six things:

1. The definition, in one sentence.
2. The owner, a named person.
3. The date it takes effect.
4. What it replaced, word for word.
5. Why it changed, in one line.
6. Who approved it.

For example: MQL = a lead with a score of 60 or more from a target account. Owner: Head of Marketing Ops. From 1 October. Replaced: score of 50 or more from a target account. Why: sales accepted fewer than half of the leads scoring 50 to 59. Approved by: CMO and VP Sales.

Now a jump in MQLs in October has an answer before anyone asks. (If you haven't written the definitions down at all yet, start with What Is Business Data Accountability?, and for the one that moves forecasts most, how to define closed won.)

Run the trace test on one number this week

Pick one number from your last board deck or QBR. Then try to walk it back, one hop at a time. Write down where the trail breaks.

1. The report. Which report or filter produced the number? Can you find it and rerun it for the same period? If today's answer differs, note the gap; usually records have moved since, and hops 3 and 4 will show it. If nobody can find the report, or the number came from a spreadsheet, the trail breaks here.
2. The definition. Which definition was used, and was it the one in force on that date? If nobody can say, or the only answer is "the usual one", the trail breaks here.
3. The records. Can you list the records behind the number? Every deal, every lead, every row. A number you can't open up can't be checked.
4. The changes. For the records that matter most, who last changed the stage, the amount, the close date and the owner, and when? In Salesforce, stage, amount and close date are in the opportunity's Stage History; owner and custom fields need Field History Tracking switched on. In HubSpot, property history is on by default: open a property's history on the record, which also shows whether a user, a workflow or an integration made the change. If tracking was never on, the trail breaks here. Switching it on today won't recover last quarter, but next quarter's number will be traceable.

Expect to stop at hop 2. That's useful to know. The first hop that breaks is the first thing to fix, and it is usually cheaper than another cleanup project.

Then ask the owner of the number to explain it using only what the trail shows. If they can do it in five seconds, you own that number. (The 5-Second Rule for Data is the longer version of that test.)

Why it matters more as you automate

Every workflow, enrichment tool and AI agent you add can change records. Each one is another hand on the data. If those changes aren't traced back to a person or a rule, you can't tell a real shift in the business from a tool doing something nobody asked for. That's why we surface issues instead of silently fixing them: a fix nobody can see is a change nobody can trace.

Where Gohvio fits

Gohvio keeps the definition trail. You upload your own guardrails, the definitions your teams actually use, as a document. A person approves them, and every change is recorded with who changed it and when. Gohvio then checks your CRM data against those guardrails and shows what is broken, what it does to pipeline and forecast, and the rows behind every finding. It is read-only today: it reads your CRM and does not write to it.

Stop debating your data. Start making evidence-based decisions.