Before a sales metric informs a decision, it owes you five answers. What action did the buyer take? What establishes that it happened? What's the denominator? Which decision will the result change? Where does the evidence stop? Growth Orbit puts the numbers it shares with clients through these questions. A metric without the answers is a number, with no basis for treating it as evidence.
Consider a LinkedIn campaign for a SaaS company. Connection requests were accepted by 5 of 26 decision-makers and 4 of 71 influencers. These aren't the totals people rush to show a board. Yet they answer all five questions and point to a specific next step: direct the next wave toward decision-makers. Most dashboard figures don't give you that clarity.
What are the five questions every sales metric must answer?
Use these five questions before putting a sales metric in front of a board or using it to make a decision:
What did the buyer actually do? Identify something the buyer did, rather than an activity your own team performed.
How do we know? Specify the recording system and the period its records cover.
Out of how many? Supply both the count and its base instead of a percentage alone.
What will we decide because of it? State which decision follows from the figure.
What can't this tell us? Explain the boundaries without obscuring them.
A modest total with answers to all five is more useful than an impressive one with none.
What did the buyer actually do?
Begin by identifying whose behavior was measured. A buyer might accept a connection request, attend a meeting, or return a redline. Those actions happened on the other side of the sale and can be checked by someone outside your team. Your team's activity is a different measure.
This first test exposes the weakness in most dashboards. Booking meetings, creating opportunities, advancing stages and scoring leads record your team's actions or judgments. They help manage the team, but they don't establish decisions made by buyers.
How do you know where a sales number came from?
For question two, identify both the source and the time window. Which system recorded the buyer's action, and through what date? "Accepted, according to the LinkedIn event log, through June 25" supplies an answer. "The dashboard says so" supplies none.
Because systems conflict, this question uncovers more problems than the others. The campaign's outreach tool said 110 requests had been sent. Its internal dashboard counted 144 contacts pushed into outreach. The event log recorded 120 actual sends. Which source you accept determines the rate you report.
The definitions behind metrics can conflict as well. In Gartner's 2020 survey of 299 sales leaders, just 25% had standard definitions for sales metrics across their organizations.
Why does "out of how many" matter so much?
Without a denominator, the same count can carry very different meanings. Ten acceptances might seem disappointing, while ten out of 26 seems encouraging. Put those ten against 1,000 and you have a concern. That's why the base matters.
Our instincts often miss the distinction. For a 1994 experiment, psychologists Veronika Denes-Raj and Seymour Epstein offered a dollar to people who drew a red jelly bean. Many participants selected 7 red beans in a bowl of 100 rather than 1 red bean in a bowl of 10. The smaller bowl offered better odds, but the larger numerator was more appealing.
Present counts to counter that instinct. In 1995, Gerd Gigerenzer and Ulrich Hoffrage found that natural frequencies such as "5 of 26," help people reason much more accurately than percentages do. A dashboard should therefore display "19% (5 of 26)," with the count attached to the rate.
What decision should a sales metric support?
Decide which decision a sales metric will inform before examining the result. A figure that changes nothing you do serves only as decoration.
You can name an actionable response: focus the next wave on decision-makers, transfer budget between lists, or stop calling companies below your revenue floor. Compare that with "to track it" or "for visibility." Neither says what you'll decide.
Should you tell your board what a number can't prove?
Tell the board where the evidence ends. Precise statements of a number's limits make the remainder credible while costing very little trust.
Anne Marthe van der Bles and colleagues reported five experiments involving almost 5,800 people in PNAS in 2020. Communicating numerical uncertainty produced only a small decline in trust. Describing the uncertainty precisely preserved trust better than an imprecise caveat.
Be especially explicit about small samples. Amos Tversky and Daniel Kahneman showed in 1971 that trained researchers, too, overestimate how representative a small sample is. Five of 26 can justify action without justifying certainty. Your answer to the fifth question must make that clear.
What does a sales number that answers all five look like?
A figure that passes all five tests identifies the event, its evidence, and the sample size. It also tells you what action to take and how far the conclusion can be trusted.
| Question | Answer |
|---|---|
| What did the buyer actually do? | A buyer accepted a LinkedIn connection request |
| How do we know? | The LinkedIn event log, through June 25 |
| Out of how many? | Decision-makers: 5 of 26 (19%). Influencers: 4 of 71 (6%) |
| What will we decide because of it? | Direct the next wave toward decision-makers |
| What can't this tell us? | It establishes neither meetings nor revenue. Samples are small, and decision-makers received more research, so the effects of seniority and research depth cannot be separated. |
The modest result, with its limitations stated honestly, supported a specific decision that could be defended.
What does a sales number that fails look like?
A metric can appear impressive while failing to identify a buyer's action. One B2B fintech viewed a 67% demo-to-qualified-opportunity conversion rate as very strong. Here's what the five questions reveal:
What did the buyer do? The buyer attended a demo. What the 67% rate captured afterward was the seller deciding the deal qualified.
How do we know? The SDR edited a stage field at the point in the process where SDRs were paid.
Out of how many? The base was 397 demos, so that answer held up.
What will we decide? No decision followed; a positive assessment was the entire conclusion.
What can't it tell us? It doesn't establish whether a buyer intended to purchase. Eighty percent of the opportunities went no further; only 5% closed.
How do you put the five questions to work?
At Growth Orbit, we recommend making the five questions part of everyday business decisions rather than another form to complete:
Test last month's most scrutinized board figure. Take the number the board focused on most, work through each of the five questions, and record the answers.
Require all five answers for board materials. Every number needs them before it enters the deck, including numbers in the appendix.
Attach dates and counts to each rate. Percentages on the executive dashboard should never appear without them.
Place the intended decision beside the metric. Record the decision you ultimately made as well.
Apply the questions to AI output, too. Discard an answer if the AI tool cannot identify the records it drew on.
Make uncertainty safe to admit. Reward "I don't know." Punishing it teaches people to conceal what they don't know.
Begin with the good news. The figures you welcome tend to escape scrutiny.
Frequently asked questions
What is a verified sales metric?
A verified sales metric answers every one of the five questions. It links a buyer's action to an identified source, a date range and a stated count, then specifies the decision and the limits. A reported metric simply reflects what the system reports. Its trust score is the share of that reported number that has been verified.
How should I report pipeline to my board?
Present the evidence alongside the pipeline figure: buyer actions, source, underlying count, intended decision, and limits of the conclusion. Put the verified figure beside the reported figure and explain any difference directly.
Is a small number ever good enough to act on?
A small result can support action when all five questions are answered and the limits are explicit. Five of 26 provides a basis for a specific, low-risk choice about the next target. A much larger figure that can't identify a buyer's action doesn't meet that standard.
The five-question card
-
What did the buyer actually do?
-
How do we know?
-
Out of how many?
-
What will we decide because of it?
-
What can't this tell us?
Try this with the figure you presented to your board last month. Do you have answers to every question?
Steve Schilling founded Growth Orbit. For a closer look at why data sources matter, read Buyer Intent Data vs. Usable Insight and Is Your CRM Data Ready for AI?