Sales KPIs are the metrics you steer a sales team by. Useful ones split into two kinds: leading indicators you can still influence this week (first contacts per channel, meetings booked, pipeline created) and lagging indicators that report an outcome after the fact (revenue, win rate, cycle length). A dashboard made only of lagging numbers is a rear-view mirror with a steering wheel painted on it.
Most KPI articles hand you a list of 15 or 20 metrics. The list is not the problem. The problem is that nearly every metric on it measures activity, and according to the Salesforce State of Sales, 6th edition, reps spend 70 % of their week on non-selling tasks. Measure activity hard enough and you will mostly be measuring administration, in high resolution.
Leading or lagging: the only split that matters
Sort every candidate metric by one question: if this number looks wrong on Monday, can anyone change it before Friday? If yes, it is a leading indicator and belongs in the weekly routine. If no, it is a lagging indicator and belongs in the monthly review, where it validates whether the leading ones were the right bets.
Most dashboards fail because they mix the two on one screen with the same visual weight. Revenue and dials per day are not peers. Revenue is the exam result, dials are homework, and homework only matters if it correlates with the result in your business, which is something you have to check rather than assume.
The tooling rarely helps here. In the Atreus B2B-Vertrieb 2025 study of 288 sales executives, 61.4 % named CRM optimisation as their top technology priority and 48.5 % named data analytics and BI tools. A CRM configured around what is easy to record hands you activity fields by default, while a leading indicator like follow-up latency has to be built on purpose.
Leading: review weekly
First contacts per channel
Qualified meetings booked
New pipeline value created
Follow-up latency on open quotes
Share of deals with a named decision maker
Lagging: review monthly
Revenue and margin
Win rate by segment
Average deal size
Sales cycle length
Customer concentration
Check the correlation once, then trust it for a year. Take four quarters, plot your favourite leading metric against closed revenue two months later, and look at whether the lines move together. If they do not, you have been coaching a number that does not predict anything.
The eight sales KPIs that explain something
Eight is not a compromise, it is a limit with a reason: a weekly meeting cannot hold more than eight numbers before people start reading the colours instead of the values. Each row names what the metric explains, which is the column most KPI lists leave out.
| KPI | How to calculate | What it explains | Type |
|---|---|---|---|
| First contacts per channel | New conversations started, split by channel | Whether your acquisition model actually works, per channel | Leading |
| Qualified meetings | Meetings with a named decision maker and a budget question asked | Whether reach turns into access | Leading |
| New pipeline value | Sum of newly created opportunities in the period | Whether next quarter has a chance | Leading |
| Follow-up latency | Median hours from inbound enquiry to first human reply | Whether your process loses deals it already won | Leading |
| Win rate by segment | Won deals divided by closed deals, per segment | Whether you are competitive where you claim to be | Lagging |
| Average deal size | Revenue divided by number of won deals | Whether your offering portfolio is priced and packaged | Lagging |
| Sales cycle length | Median days from first contact to signature | Whether you sell to the right level of the organisation | Lagging |
| Revenue concentration | Share of revenue from your largest three customers | How much of your company depends on three phone calls | Lagging |
Five numbers that look like KPIs and are not
Build a baseline instead of chasing benchmarks
Industry benchmarks for win rate or cycle length are published widely and vary so much by segment, deal size and cycle length that they mostly serve to reassure. Four quarters of your own history is a stricter, cheaper and more honest yardstick, and you already own the data.
Build it like this: take the last four quarters, compute each of the eight metrics per quarter, and use the median as your baseline plus the spread as your tolerance. Then a metric is only interesting when it leaves its own spread, which is a far better trigger than a round number somebody else published.
There is a size argument here too. Per KMU im Fokus 2025, 99.7 % of Austrian companies are SMEs, and the report puts average return on sales at 5.5 %. At that margin, an eight-metric dashboard someone actually maintains beats a thirty-metric one that decays, because the maintenance itself is billable time. Start with what you can keep current, and see the sales analysis guide for the structural half.
| KPI | Your baseline | Warning signal |
|---|---|---|
| First contacts per channel | Median of the last four quarters, per channel | One channel carries more than half of all first contacts |
| Qualified meetings | Meetings per person per month | Meetings rise while new pipeline value stays flat |
| Follow-up latency | Median hours, measured over a full month | The median is fine but the slowest 10 % takes days |
| Win rate by segment | Per segment, never as one company-wide figure | Two segments diverge by more than a factor of two |
| Revenue concentration | Share of the top three customers, per year | The share grows two years in a row |
Never build a baseline from a quarter you already know was unusual. A quarter with one outsized deal will set a target nobody can hit twice, and the team will learn that the dashboard is fiction. Exclude it explicitly and write down why.
Build the dashboard in five steps
Name the decision each metric serves
Write the decision next to the metric. A metric with no decision attached is decoration, and decoration is what makes dashboards get ignored.
Split the screen by rhythm, not by team
Leading metrics in the weekly block, lagging metrics in the monthly block. Same screen, visibly separated, so nobody reacts to revenue on a Monday.
Set thresholds from your own spread
Use the baseline median plus the observed spread from the last four quarters. Colour only fires when a metric leaves its own range, not when it misses a wish.
Give every metric one owner
Not a team, a person. Shared ownership of a number reliably produces a discussion about the number instead of an action on it.
Review the dashboard itself twice a year
Ask which metric never triggered an action in six months and delete it. Dashboards grow by default and shrink only on purpose. Pair this with your annual review routine so the numbers and the conversations match.
If the numbers are fine but nothing changes, measure the coaching
Leading indicators only move if someone coaches them weekly. Manager feedback shows whether that is happening, in about eight minutes.
Teamo AI: so wird aus Kennzahlen wöchentliche Führungsarbeit
Keeping eight metrics current and coached is weekly work. Teamo AI connects Pipedrive, HubSpot, Slack, Teams and your calendar into one shared intelligence that belongs to your company and stays in Europe, governs who sees what through access control, and is ready on day one. 14 days free, no credit card, your team invited in minutes.
A KPI tells you that something is off, not where
This is the ceiling every KPI dashboard hits. Win rate fell: fine, but did it fall because the segment was wrong, because the offering is not comparable, because targets never reached weekly behaviour, or because the acquisition model brings in the wrong conversations? The number cannot answer that, because all four causes produce the same drop.
That is what a structural check adds, and why it belongs next to the dashboard rather than instead of it. Nine questions across six dimensions give you the where, the dashboard gives you the how much. The sales analysis guide compares the methods in full, and people analytics covers reading structural and human data together.
What to do with eight numbers and one diagnosis
Put the eight metrics in place first, because they cost nothing but a query and they give you the baseline you will need for every later comparison. Then run the structural check with three people answering independently, and compare the field it flags against the metric that is furthest outside its own spread. When both point at the same field, you have a decision rather than an opinion, and that is the whole point of measuring.
If they point at different fields, trust the structural finding first and use the metric to size it. A number is precise about the wrong thing far more often than a structure is wrong about the right thing.
The short version
- Sort metrics by one question: can anyone change this number before Friday? Leading goes in the weekly, lagging in the monthly.
- Eight metrics is the working limit for a weekly meeting. More and people read colours instead of values.
- Dials, total pipeline, open rates, quote counts and CRM activity scores look like KPIs but measure effort or tidiness.
- Build thresholds from four quarters of your own history, not from published benchmarks that vary by segment.
- A KPI says that something is off. A structural check says where. Run both, and act when they agree.
![B2B Customer Acquisition: Why Effort Cannot Replace the Model [2026]](https://www.teamazing.com/wp-content/uploads/2026/08/manager-effectivness.jpg)
![Sales Commission Models: The Structure Decides, Not the Rate [2026]](https://www.teamazing.com/wp-content/uploads/2026/08/provisionsmodell-vertrieb.jpg)
![Sales Analysis Methods: What Actually Explains Your Numbers [2026]](https://www.teamazing.com/wp-content/uploads/2026/08/potential-analysis-for-leaders-1.jpg)