Sales Metrics to Track: The Nine That Change a Decision

A sales dashboard tends to grow the way a garage does. Someone asks a question once, a metric gets added to answer it, and then it stays — past the question, past the person who asked, past the quarter where it meant something. Nobody removes a number, because removing a number looks like hiding from it.

So by year three you have forty tiles, a weekly meeting where they are read aloud, and a room full of people who could not tell you what would be done differently if any single one of them moved four points.

That is the actual problem with sales metrics, and it is not solved by finding better ones. It is solved by throwing most of them away.

The test: which decision changes?

Here is the whole filter, and it takes about ten seconds per metric:

If this number moved significantly, what would we do differently on Monday?

If you can name the action — and the person who takes it — the metric earns its place. If the honest answer is "we would look into it" or "we would know how we are doing", it is not a metric. It is a number you report.

The distinction matters because reported numbers are not free. They take up room in the meeting, they create the appearance of management, and they reliably crowd out the three or four figures that would actually have changed what happened this week.

The nine

Each of these is here because a decision hangs off it. The decision is the point of the metric, so it is named first.

1. Stage-to-stage conversion

The decision: which part of the process gets attention this quarter.

An overall win rate is a blend — of segments, sources, deal sizes and stages — and when a blend moves you cannot tell which ingredient moved. Broken into stage-to-stage conversion, it stops being a scoreboard and becomes a map. If eighty percent of first meetings become qualified opportunities but only a quarter of those survive to a proposal, you have a qualification problem wearing a closing problem's clothes, and every hour spent coaching closing technique is an hour spent in the wrong stage.

This is also the metric most sensitive to a definition nobody has written down. If two reps disagree about what makes an opportunity "qualified", the conversion rate between those stages is measuring the disagreement rather than the pipeline. A shared qualification standard — MEDDIC or anything you actually enforce — is what makes this number mean the same thing twice.

2. New opportunities created, per rep, per week

The decision: whether to act this week on a gap that lands next quarter.

Almost everything on a sales dashboard tells you about a quarter that is already determined. This one tells you about the next one, early enough to do something. If your cycle is ninety days, opportunities created today are the number you report in three months, and a quiet fortnight now is a hole you cannot fill later no matter how hard anyone works in the final week.

Track it per rep and per week, not as a team total per month. A team total hides the one person who has stopped prospecting behind the two who have not.

3. Pipeline coverage, against your own win rate

The decision: prospect, or work what you already have.

Coverage is the only metric on this list that directly answers the question reps ask most often — should I be finding more, or pushing what I have? — and it is also the one most often taken from a blog post rather than from your own data. The multiplier everyone quotes is a win-rate assumption in disguise, and using someone else's assumption quietly imports their business into your forecast. The arithmetic, and why the 3x rule is not a rule.

4. Meetings held to opportunities created

The decision: fix getting in, or fix what happens once you are in.

Two teams can both miss the number with identical activity. One cannot get meetings; the other gets plenty and cannot convert them into anything real. These require completely different interventions — outreach and targeting versus discovery and qualification — and without this ratio the two failures look the same from the outside.

When this is the leak, it is almost always the same leak: the meeting happens, the rep presents, nobody finds out what the buyer is actually trying to do, and the call ends with a polite agreement to reconnect. The questions that prevent that are cheaper than more meetings.

5. No-decision share of losses

The decision: fight competitors better, or qualify harder.

Split your closed-lost into two piles. One lost to a named competitor. The other lost to nothing — no decision, no budget, went quiet, "revisit next year".

The split tells you where your losses actually come from, and it is usually not where the investment goes. Battlecards, competitive matrices and objection drills all serve the first pile. If the second pile is bigger — and for most B2B teams it is — then the constraint is that buyers are not convinced anything needs to change at all, which is a different problem with a different diagnosis.

Sort your last twenty closed-lost into "lost to someone" and "lost to nothing", then look at where your team's prep time goes. Most teams find they are heavily armed for the smaller pile.

6. Days in current stage — per deal, not averaged

The decision: which specific deals get worked today.

Average sales cycle length is one of the most-reported and least-useful figures in sales, for a reason worth understanding: it is computed from deals that closed. The deals that stalled forever and were eventually written off are not in the average, so the number systematically describes your successes and tells you nothing about the pipeline you are actually holding.

Days in current stage fixes this, and changes the character of the metric entirely. It is not a statistic, it is a worklist: sort descending, and the top of that list is the set of deals where something has gone wrong that nobody has said out loud. Some need a call. Some need closing as lost, which is the decision this metric most reliably forces and the one teams most reliably avoid.

7. Close-date slip rate

The decision: how much to discount the forecast, and whose dates to trust.

Each month, count the deals whose close date moved out. Not whether the forecast was right in aggregate — that can be accidentally correct, two errors cancelling — but how often the individual dates hold.

This gives you two things. A team-level discount you can apply to next month's commit with a straight face, and a per-rep picture of whose dates mean something. A rep at ninety percent accuracy and a rep at forty percent can both make quota; only one of them lets you plan. And slip rate is a coaching conversation with an unusual property: it is about forecasting honesty rather than performance, so it can be had without anyone getting defensive.

8. Realised price against list

The decision: change the price, or coach the negotiation.

If discounting is deep and consistent across every rep and every segment, the price is wrong and no amount of negotiation training will fix a positioning problem. If it is deep for three reps and shallow for the rest, the price is fine and you have a skills gap that shows up most often in one moment — the rep hearing a price objection and treating it as a request for a smaller number rather than a question about value.

The distinction is invisible in a revenue number and obvious in this one.

9. Deals closed per rep

The decision: whether you are entitled to draw any conclusion about a rep at all.

This is not a performance metric. It is a sample size, and it belongs on the dashboard because it governs how much weight every other per-rep number can carry. A rep who closed four deals this year has a win rate with an enormous error bar around it; the difference between their 50 percent and someone else's 70 percent may be one deal going the other way.

Enterprise teams run into this constantly and act on the numbers anyway, which is how a good rep gets managed out over statistical noise. How much evidence a rep's number actually contains is a prerequisite for using any of the eight metrics above at the individual level.

The ones to stop reporting

Not because they are lies. Because nothing changes when they move.

Total activity — calls made, emails sent. No decision hangs off the aggregate, and the moment it is reported it becomes a target and stops measuring effort. If you need an activity number, use new opportunities created, which counts outcomes of activity rather than activity itself.

Overall win rate. A blend. When it moves you have to break it into stage conversion to find out why, so report the decomposition and skip the headline.

Average sales cycle length. Survivorship, as above. Useful once a year for capacity planning, useless weekly.

Total pipeline value. A single large deal makes the whole number move, and a pipeline of the wrong shape can look identical to a healthy one. Coverage and stage distribution say what this is trying to say.

Quota attainment, used as a diagnosis. It is the outcome every other metric exists to move. Report it, be judged on it, do not mistake it for an explanation.

Email open rates. Privacy proxies pre-fetch images on the recipient's behalf, so a pixel fires whether or not a human ever looked. The measurement broke; replies did not.

How often each can honestly be read

The most common dashboard failure is not the wrong metric. It is the right metric read too often, producing noise that then gets acted on.

Weekly: new opportunities created; days in current stage. Both can genuinely move in a week, and both have an action attached that is worth taking within the week.

Monthly: close-date slip rate; meetings to opportunities. A month is roughly the shortest window where these contain signal rather than the calendar.

Quarterly: stage-to-stage conversion; no-decision share; realised price; deals per rep. These need volume before a change is distinguishable from randomness, and a quarter is usually the first point at which you have it.

Continuously, but only as a question: pipeline coverage. It is not a performance number, it is a prompt — prospect, or work what you have.

The rule underneath: read a metric no more often than it can meaningfully move. Reading a quarterly metric weekly does not give you an early warning. It gives you twelve chances to overreact.

They are a chain, not nine gauges

The tidiest thing about a dashboard is also the most misleading thing about it: nine tiles, side by side, each looking independent. They are not.

Coverage is computed from win rate. Win rate is the product of your stage conversions. The forecast is stage conversion with slip rate applied. Realised price feeds deal size, which feeds coverage again. Pull one and the others move — which is why "improve win rate" is not an instruction, and why a team that tightens qualification will watch its opportunity count fall, its coverage ratio fall, and its win rate rise, all at once, and have to decide whether that is success.

It is. But only someone reading the chain rather than the tiles can tell.

That is the real argument for nine metrics instead of forty. Nine, you can hold in your head as a connected system. Forty, you read aloud.

Common questions about sales metrics to track

How many sales metrics should a team actually track? Fewer than you have. The count matters less than the test: every metric on the dashboard should have a named person who does something different when it moves. In practice that constraint lands most teams somewhere under ten, because the tenth one rarely survives the question.

What is the single most important sales metric? There isn't one, and the question usually means "which number should I put in front of the executive team?" — which is an outcome question, so the answer is attainment or bookings. If the question is really "which number should I manage from", it is whichever one is currently your constraint, and that changes.

What is the difference between a KPI and a metric? In practice, a KPI is the outcome the organisation has agreed to be judged on, and a metric is diagnostic — something upstream that explains the KPI and can be acted on. Most sales dashboards are stacked with KPIs and short on metrics, which is why they describe the past accurately and change nothing.

Should I track activity metrics at all? At the individual coaching level, sometimes: if a rep's pipeline has dried up, looking at what they did last week is a reasonable diagnostic conversation. As a reported team metric it does more harm than good, because it becomes a target and then it stops being information.

How do I know if my pipeline coverage ratio is healthy? Derive it from your own win rate rather than adopting a published multiplier, because the multiplier is a win-rate assumption and someone else's does not describe your business. The arithmetic is here.

Why is my forecast accurate in aggregate but wrong on individual deals? Two errors cancelling. A deal that slips and a deal that pulls forward produce a correct total and a false sense that the process works. Close-date slip rate measures the individual dates rather than the sum, which is why it catches this and the forecast number does not.

How long before a change in these metrics means anything? Longer than most teams wait — the lower the deal volume, the longer. A team closing a few hundred deals a year can read quarterly movement with some confidence; a team closing thirty cannot, and should treat quarter-to-quarter changes as a prompt to go and look at the deals rather than a finding.

What should I do first if my dashboard is a mess? Run the test on every tile and delete anything where no one can name the decision. Do the deletion before adding anything new — a cluttered dashboard is not improved by better metrics arriving alongside the ones already being ignored.

A note on sources

There are published benchmarks for every number in this article — average win rates, "healthy" coverage multiples, standard cycle lengths, typical discount depth. This piece deliberately quotes none of them, because they are not comparable between any two companies that contribute to them.

The reason is definitional rather than statistical. Two companies measure win rate over different denominators, count an opportunity as created at different moments, define "qualified" differently, and record no-decision losses under different reasons depending on what the CRM dropdown offers. Averaging across that produces a number with a decimal point and no meaning, and the decimal point is the dangerous part.

What is checkable is your own data, and every claim above can be tested against it this week. Sort your last twenty losses into "lost to someone" and "lost to nothing". Count how many close dates moved last month. Sort open deals by days in current stage and read the top ten. Check whether discounting is uniform across reps or concentrated in a few. None of those require a benchmark, and all of them will tell you something a benchmark cannot — which is what is true here, rather than what is average somewhere else.

The framing of metrics as diagnostics upstream of outcomes follows the sales-management literature on managing activities and objectives rather than results, and the caution about targets corrupting measures is the standard observation usually attributed to Goodhart. The nine-metric selection, the decision test and the chain argument are ours — a practitioner's way of deciding what to delete, which is the part most metrics writing skips.