From KPI to revenue

A simple way to check whether the numbers you track are connected to the business outcomes you need.

Your marketing numbers are going up. But can you explain how they contribute to revenue?

Marketing is under pressure to prove it works. Revenue is usually far removed from individual marketing activities, so teams fall back on numbers that are easy to count: impressions, followers, traffic, MQLs, meetings.

The problem starts when making those numbers grow becomes the goal, even though nobody has established how they connect to the business.

You may recognize it when

You hit the marketing target, but sales results don't improve.
Everyone knows the number, but nobody can explain why it matters.
A drop in the number triggers more activity rather than a question about what the business needs.
You report the number, but don't know what happens after it.
Questioning the metric sounds like questioning the team's performance.

Use this to find

Which KPIs are empty

Numbers you report because they're available, not because they tell you something important.

What you should be tracking

The outcomes and evidence that sit between marketing activity and revenue.

Which metrics are useful to watch but dangerous to target

Numbers that help you understand progress but create the wrong behavior when teams are asked to make them go up.

Where the connection to revenue breaks

And whether the problem is the KPI, the evidence behind it, or a decision the business never made upstream.

Start with the business outcome. Follow the chain to the numbers you report. Where you can't complete the chain, you've found something worth fixing.

Three rules

Fill it in order. If you can answer field 5 but not field 1, you have found your problem already.
Do one objective at a time. Two objectives on one sheet and both answers go vague.
Every field has answers that do not count. Those are the ones that feel finished and tell you nothing.

The example running through all eight fields is one made-up company: a scheduling software business selling to mid-size logistics firms. Hide it if you would rather start from a blank sheet.

Business outcome

What commercial result are we trying to change?

Counts

Names a commercial result: revenue, retention, win rate, market entry
Has a direction and a number, or a date
Someone in finance would recognize it as a goal
Exactly one per map

Does not count

Awareness, visibility, brand, positioning
More leads, more pipeline, growth
Anything that would fit any company in any year
A marketing goal wearing a business goal's clothes

Example

Increase new-business revenue from logistics firms running 50 to 200 vehicles, from 12 deals a year to 20.

Marketing contribution

What has to change in buyer behavior for that result to move?

Counts

Describes what buyers know, believe, or do differently
Names which buyers, specifically
Could turn out to be wrong
Written as a change, not a task

Does not count

Run campaigns, publish content, do events, launch a newsletter
Support sales, build the brand, own the funnel
Any answer about what we do rather than what changes
"Everyone in the market"

Example

Operations directors at those firms, who currently run scheduling on spreadsheets, start putting us on the shortlist without a rep contacting them first.

Intermediate outcomes

What would we see in the market if this were working, before revenue arrives?

Counts

Happens outside our building
Would move before revenue does
Two to four maximum
Each one is a change in someone else's behavior

Does not count

Posts published, emails sent, events run, pages shipped
Anything we can complete by working harder
Revenue itself, which belongs in field 1

Example

More target-band firms arrive at first calls already knowing what we do. More inbound demo requests from that band. Fewer calls that start with us explaining the category.

Evidence

What would raise or lower our confidence that this is happening?

Counts

Comes from a source outside the metric itself
At least one qualitative source: what sales hears, what buyers say
Available now, or gettable with a baseline you can set this quarter
Could contradict the hypothesis, not only confirm it

Does not count

"The metric would go up." That is circular
Evidence where no possible result would make you say this is not working
A study you would need budget and six months to commission

Example

Reps report buyers mentioning us before contact. Branded search from firms in the band. Demo request forms describing a spreadsheet problem in the buyer's own words.

Metrics

What will we monitor? For each one: whose behavior it captures, and what decision changes if it moves.

Counts

Each metric names whose behavior it captures
Each has a decision attached to it
Each traces back to a line in field 3
Five or fewer

Does not count

A metric nobody would act on if it doubled
Whatever the tool reports by default
A number with no owner behind it
Aggregate volume with no composition underneath

Example

Demo requests from the 50 to 200 vehicle band: buyer behavior, decides where next quarter's budget goes. Branded search from that band: market awareness, decides whether category content continues.

Targets

Which of those are safe to set as a target? For each, ask what a smart team would do to hit it fast.

Counts

Fewer targets than metrics
Each survives the gaming test below
At least one sits downstream of the first response
Quality condition written into the definition

Does not count

Copying all of field 5 into this box
A volume target with no quality condition
Any metric the team can move by lowering a bar

Example

Target: opportunities created from the target band. Not demo volume, because the fastest way to raise that is to accept demos from firms too small to buy.

The gaming testIf the fastest way to hit the target is to loosen a definition, widen the audience, or lower the bar, it works as a signal and fails as a target.

Downstream check

Take your primary metric and follow one cohort forward. A real number at every stage. Rename the stages to match your funnel.

Counts

Every row has a number, or the word unknown
One cohort followed forward, not six separate reports
Ends at revenue
Old enough for deals to have closed

Does not count

Leaving a row blank because the data is hard to pull
Conversion rates borrowed from a benchmark report
Stopping at the first stage that still looks good

Example

180 demo requests. 96 in the target band. 41 opportunities. 14 progressed past first stage. 6 closed. Revenue: unknown, finance reports by product line, not by source.

Stage

Number

Primary metric

Of those, in our ICP

Became opportunities

Progressed past first stage

Closed

Revenue

Read itWhere the numbers stop is where your measurement system stops telling you something about the business.

Known limitations

What can we not confidently attribute or measure here?

Counts

At least one thing you cannot attribute
Names the other causes running at the same time
Says what you would need to close the gap
Plain enough to say out loud in a board meeting

Does not count

"None"
"Attribution is hard" with nothing specific after it
A limitation that conveniently only affects the numbers you dislike

Example

Two conference sponsorships ran in the same quarter and we cannot separate their effect from content. Branded search includes existing customers and we have no way to filter them out.

Reading your own map

Four patterns. Each one points somewhere different, and the first one is not a measurement problem at all.

Field 1 vague or contested

The problem sits upstream. No metric set fixes an undefined commercial outcome. Fix the foundation first.

1 to 4 solid, 5 disconnected

The chain breaks at metric selection. The metrics were picked for availability, not relevance. Redesign the chain.

Field 7 mostly blank

Nothing confirms the metrics behave as expected. The gap is evidence, not metric choice. Build downstream tracking.

Field 6 as long as field 5

The metrics have become the work. Separate what you monitor from what you incentivize.

Natalia Bochan Talk to me on LinkedIn

The rising number trap