You're optimising the wrong metric: the KPI that loses you money (while the dashboard is 'green')

The light is green, but the account is in the red

There’s a precise feeling, and if you’re an owner you probably know it: you look at the dashboard and it’s all green. Visits are up, messages are increasing, tickets are being closed, followers are growing, the team is happy because “the numbers are good”. Then you look at the bank account, or the month-end margin, and nothing adds up. The company is “doing well” on every screen and meanwhile you struggle to make it to the end of the month. How is that possible?

It’s possible because you’re optimising the wrong metric. Having wrong KPIs in the company is worse than having none, because they give you the confidence that you’re doing well while you’re doing badly — and that confidence is exactly what stops you from correcting course. A green dashboard on numbers that don’t count is a misleading dashboard: it isn’t neutral, it actively makes you decide worse.

I’ve seen this scene in healthy companies, with competent teams acting in good faith. It isn’t incompetence: it’s that the numbers that are easy to measure are almost never the numbers that count, and over time the dashboard fills with the former and empties of the latter. Nobody decides this on purpose: it happens by gravity, because it’s simpler to count visits than the real margin per product. The result is a dashboard that measures how much you move, not how much you advance.

This article is for anyone who already has numbers — maybe a dashboard built by someone — but suspects they’re measuring the wrong thing. Let’s see which vanity metrics fool you most, with concrete examples where the “nice” number hides the real problem; why choosing KPIs isn’t a technical job to delegate to IT; and how to redo the definitions so that the dashboard, finally, turns green when the company is actually doing well — not when it suits it.

Vanity metrics: the numbers that make you feel good and say nothing

Vanity metrics are the numbers that go up easily, look good in a meeting, and have almost no link to money. You recognise them with a simple question: if this number doubles tomorrow, do I earn more? If the answer is “not necessarily”, it’s a vanity metric.

The most common:

Website visits. They go up with a campaign, a viral article, a bit of bought traffic. But visits without contacts, and contacts without sales, are just a counter spinning. Traffic is a means, not a result.
Messages received / “engagement”. A hundred extra enquiries look like a success. But if they’re a hundred enquiries from people who don’t buy, or who clog customer care without ever becoming customers, you’ve only increased the work without increasing revenue.
Followers. The emptiest number of all for most companies. Ten thousand followers who don’t buy are worth less than ten customers who come back.
Tickets closed. Looks like efficiency. But “closed” doesn’t mean “resolved”, and a ticket closed fast with an unsatisfied customer is a customer who leaves — only the number is green.
Number of quotes issued. Doing lots of quotes looks like activity. But if the acceptance rate collapses, you’re only burning sales people’s time.

The problem with vanity metrics isn’t that they’re false: they’re true. The visits really are there. The problem is that they measure activity, not result — and activity can be inflated forever without the company being better off. In fact, it often gets worse, because the team optimises what is measured, and if you measure activity you get a lot of useless activity.

Three examples where the “nice” number hides the problem

Abstract is easy. Let’s look at three concrete cases, the kind I see often, where a green metric hides a red problem.

Average ticket going up vs margin going down

A shop (physical or online) celebrates: the average ticket (scontrino medio) has grown 15%. Seems like great news. But digging, the average ticket has gone up because you’re selling more of a high-price and low-margin product, maybe pushed by an aggressive discount. Result: you take more per ticket, you earn less at month end. If your KPI is average ticket, you’re happy while you lose money. If your KPI is margin per ticket, you see immediately that something is wrong.

Tickets closed vs customers lost

Customer care cheers: average ticket-close time halved, tickets closed at maximum. Green everywhere. Pity that, to close fast, they fob customers off with quick answers that don’t solve anything; those customers don’t reopen the ticket (the number stays nice) but they don’t buy again, and maybe they leave a negative review. The “tickets closed” KPI measures the speed of getting rid of customers, not of helping them. The right KPI is something like problems solved at first contact or the rate of customers who come back after a ticket.

Occupancy at 100% vs revenue per room (or per hour)

A hotel, a B&B, a studio, a salon: “we’re full, occupancy at 100%!”. Beautiful. But if you’re full because you’ve undersold — low rates, last-minute discounts, channels that take huge commissions — you’re full and poor. Occupancy is a classic vanity metric of the sector. The number that counts is revenue per available room (or per studio hour, per chair): how much your capacity yields, not how full it is. Better 80% full at a healthy rate than 100% full at a loss.

The common thread of all three: the wrong metric measures volume (of ticket, of tickets, of occupancy), the right one measures value (margin, resolution, revenue). Volume is easy to push up and makes you feel good; value is more uncomfortable to look at and tells you the truth.

Two more cases: the revenue that fools you and the hidden churn

The three examples above are the most visible, but there are two even more dangerous, because the green number is one of those “sacred” ones nobody questions.

Revenue growing vs cash going down

Quarterly revenue is growing: you pop the prosecco. But revenue is what you issue, not what you collect. If you’re growing by granting longer and longer payment terms, or selling to customers who pay badly and late, revenue goes up and cash goes down. You grow on paper and choke at the bank. It’s one of the most classic ways a company “in growth” ends up in a liquidity crisis — and the revenue KPI was green until the last day. The number to look at next to revenue is collected, with the average payment delay.

New customers vs customers who leave

Marketing brings 50 new customers a month: bright green, “the campaigns work”. But if in the same month you lose 55, you’re running to go backwards, while you spend on acquisition. Churn — the customers who leave — is almost always invisible on dashboards, because it’s uncomfortable and harder to calculate than “new”. And that’s why so many companies pour money into acquisition while the bucket leaks from the bottom. The green number (“new customers”) hides the red one (“net customers”), which is the only one that actually counts. If you acquire and lose at the same pace, you’re not growing: you’re paying to stand still.

Billable hours up vs profitability down

For anyone who sells time — agencies, practices, consulting — the trap is “utilisation”: how many of the team’s hours are billable. Green at 95%, the team is always busy, it looks like a company that turns. But if those hours are sold at low rates, or burned on clients who pay little and ask a lot, you’re full and you don’t earn. People’s occupancy is the twin of room occupancy: it measures how full you are, not how much it yields. The right number is margin per project or per client, not occupied hours. A team at 80% on healthy work beats a team at 100% at a loss.

Why it happens: you optimise what you measure

There’s an unwritten law of management that holds in every company: people optimise what is measured. If you put a number on a dashboard and look at it in a meeting, the team will work to make it go up — it’s natural, it’s what you implicitly asked them. Which is wonderful if the number is the right one, and disastrous if it’s the wrong one.

Put “number of calls made by sales” as a KPI? You’ll get lots of calls, even to people who will never buy, because the call is what counts. Put “quotes sent”? You’ll get a mountain of quotes fired into the pile. Put “tickets closed”? You’ll get fast closures, not satisfaction. The metric doesn’t only describe reality: it shapes it. That’s why choosing the wrong KPIs isn’t a neutral measurement error: it’s a way to steer the whole company in the wrong direction, with the best of intentions.

This is also why “adding more metrics” doesn’t help. A dashboard with forty numbers doesn’t solve the problem: it dilutes it. If everything is important, nothing is, and the team chooses on its own which one to push up — usually the easiest, that is the most vanity. A few KPIs, the right ones, looked at seriously, beat any crowded dashboard.

There’s a name for this phenomenon: when a measure becomes a target, it stops being a good measure. The moment you tell the team “this is the number that counts”, the team finds the fastest way to make it go up — and the fastest way is almost never the one you wanted. You measure calls and you get empty calls; you measure quotes and you get a burst of useless ones; you measure ticket-close speed and you get customers fobbed off. Not because the team is in bad faith: because you’re rewarding that. The KPI isn’t a passive mirror of reality, it’s a lever that moves behaviour. Choosing it badly is like setting the steering wrong: you turn convinced you’re going straight and you end up off the road.

What it costs you to optimise the wrong metric

Let’s put a number on the damage, because “optimising the wrong thing” sounds like an aesthetic problem and is actually a hole in cash. Imagine a company that sets as the sales team’s target revenue (not margin). Sales people, rightly, do what you ask: they maximise revenue. With which tool? The easiest: the discount. They close more orders, bigger ones, revenue soars, the KPI is green, bonuses fire. And the margin, silently, thins out.

Let’s do the sums, as an estimate. On revenue of 2 million with average margin of 30%, moving the target from revenue to margin — and therefore discouraging the easy discount — can be worth 2–3 points of margin recovered. That’s 40,000–60,000 € a year, on a single metric changed. You haven’t spent anything on software: you’ve only stopped measuring (and rewarding) the wrong thing.

The point is this: the wrong KPI isn’t neutral. It steers the behaviour of the whole company, and every day that passes with the wrong target is a day in which the team, diligently, optimises in the direction that costs you money. Changing the number you look at and reward is among the highest-return interventions that exist, because it costs a decision and pays on every future transaction. It’s the exact opposite of buying yet another tool: here you add nothing, you remove the wrong incentive.

Who chooses the KPIs (and it isn’t IT alone)

Here’s a point that looks technical and is actually a leadership one. Who decides which are the right numbers to put on the dashboard? The wrong answer, and unfortunately the most common, is: “we ask IT” or “we put it in the hands of whoever makes the dashboards”.

IT knows how to connect the data and build the chart, but they don’t have — and shouldn’t have — the authority to decide what counts for the business. If the KPIs are chosen by whoever builds the tool, you end up measuring what is easy to measure, not what is important to improve. And what is easy to measure is almost always a vanity metric: visits are easy, real margin per product is hard.

Choosing KPIs is a business decision, and it has to be made at the right table: leadership, who knows where the money is made; sales, who knows how you actually sell; accounts, who know the real costs behind every revenue. It’s the same table you need to agree the definitions — I’ve gone into this in depth explaining why hiring a data analyst on their own isn’t enough: the technical person can calculate any number, but they can’t decide which number should steer the company. That choice is yours.

What filters hide (the average that lies)

Even when the KPI is right, the way you show it can fool you. The two most common involuntary tricks: averages and filters.

Averages hide worlds. “The average customer spends 200 €.” Nice, but if half spend 20 € and the other half 380 €, the average describes a customer who doesn’t exist, and it makes you take decisions for a ghost. Often the useful number isn’t the average, but the distribution: who are the customers who are actually worth it, who are the ones who cost you. A healthy average can hide a business made of a few good customers and many at a loss.

Filters hide problems. A dashboard that shows “only active customers”, “only orders that went through”, “only the channel that works” is green because it has filtered out the red. It isn’t lying on purpose: it’s showing a chosen slice. But the owner looking at it doesn’t know that behind the filter is what should worry them. The question always to ask a dashboard that’s too pretty is: what aren’t you showing me? Which customers, orders, products have been excluded to arrive at this green.

An honest dashboard includes the uncomfortable numbers: lost customers, failed orders, products that lose margin. Not because it’s nice to see them, but because they’re the ones you can still act on. A dashboard that only shows the nice bits is an advertising poster of yourself to yourself.

A typical case: the green dashboard that hid churn

A typical profile, architectural, no names. Subscription services company, a polished dashboard: new customers up, traffic up, tickets closed up — all green. The owner was satisfied, the team too. And yet recurring revenue was growing with difficulty and cash was tight. Nobody understood why, because “the numbers were good”.

The problem was what the dashboard didn’t show. It measured everything that came in (new customers, new contacts) and nothing that went out (cancellations). When we added a single number — net customers, that is new minus lost — the green became dark yellow: they acquired 50 a month and lost 45. They were paying handsomely to fill a leaky bucket.

What was done: first the table, to decide that the steering number was net recurring, not new. Then half the vanities were taken off the main screen and the two uncomfortable numbers were added (churn and net recurring). Finally the product underneath was made capable of calculating them automatically, with an alert when churn passed a threshold.

After: the same company, with the same data, stopped celebrating new customers and started asking why the old ones were leaving — which had been the right question for months. The value wasn’t a prettier dashboard: it was stopping looking at the one that lied with a smile.

A KPI without an owner is already dead

There’s a last reason dashboards fill with vanity: numbers without an owner tend to multiply and rot. Every KPI should have a person who answers for that number — not who produces it, but who owns it: they look at it, they notice when it’s strange, and they have the mandate to act when it’s red. A number that belongs to nobody is looked at by nobody, and then you might as well not have it.

The test is simple: for every metric on the dashboard, ask yourself “who wakes up at night if this goes badly?”. If the answer is “nobody in particular”, that number is decoration. The numbers that count always have a name next to them. And a note: the owner of a KPI must also have the power to act on it. Giving someone responsibility for a number without the tools to move it is only an elegant way to distribute blame. A few numbers, each with an owner who can actually change them: that’s how a dashboard stops being a poster and becomes a governing tool.

The five numbers almost every company should look at

Every company is different, and the right KPIs are yours, not those of a template. That said, there’s a core that, in one form or another, almost everyone needs — and that almost always is missing or is masked by a vanity. Use it as a starting point for your workshop, not as a truth dropped from above.

Margin, not just revenue. What you actually keep, not what you turn over. Better still if you see it per product or per line, so you know where you earn and where you’re working for nothing.
Collected, not just revenue. What has entered the bank, with the average payment delay. It’s the difference between selling and surviving.
Net customers (new minus lost). The number that unmasks the leaky bucket. On its own it’s worth half the reflections on marketing.
Revenue per unit of capacity. Per room, per studio hour, per chair, per metre of shelf. It tells you how much what you have yields, not how full it is.
Real conversion, with numerator and denominator defined. From contact to quote, from quote to order: where your funnel jams.

Note what they have in common: none is a vanity metric, all are more uncomfortable to calculate than their easy cousins (revenue, occupancy, visits), and all answer a concrete decision. If your dashboard doesn’t have a version of these five, you’re probably looking at the comfortable numbers instead of the real ones.

How you redo the KPIs: the definitions workshop

The good news is that fixing the wrong KPIs doesn’t require new software: it requires a table and a few hours. I call it the definitions workshop, and it’s the cheapest and most valuable thing you can do for your numbers. Here’s how it runs, in practice.

The starting question. Not “which charts do we want?”, but “which decisions do we take with these numbers?”. A KPI that doesn’t change any decision is decoration. You start by listing the three or four recurring decisions that count: what to reorder, who to call back, where to push, what to cut.
For each decision, the number that steers it. For “what to reorder” you don’t need average ticket, you need rotation and margin per product. For “where to push” you don’t need visits, you need conversion per channel. You hook every decision to its number.
The vanity hunt. You look at the current dashboard and, number by number, you ask the question: “if this doubles, do we earn more?”. Those that don’t pass the test come off, or get moved to a corner as support metrics. The dashboard slims down, and that’s fine.
The signed definitions. For the numbers that remain you write exactly what they mean — sale, margin, active customer, conversion — with the signature of whoever decided. It’s the document that closes future fights.
The mandatory uncomfortable number. You add on purpose at least one number that can be red: lost customers, real margin, failed orders. It keeps the dashboard honest.

At the end you have a few metrics, hooked to real decisions, with written definitions and at least one number that keeps you awake. It’s the opposite of the shop-window dashboard with forty green charts.

The concrete result of the workshop isn’t a slide: it’s a one-page sheet with, for every number that remains, the exact definition, the decision it steers and the owner. That sheet is worth more than any dashboard, because it’s the contract the dashboard will be built on. Print it, hang it, and every time someone wants to add a metric ask them which decision it steers and who owns it. If they can’t answer, that metric doesn’t come in.

After the workshop: the product that respects the definitions

A workshop, though, isn’t enough if then nobody respects the definitions. The risk is deciding them at the table and then finding yourself with the old dashboard, because the tool keeps calculating the way it used to. Here the difference between tool and product comes back: a real data product takes the signed definitions and applies them always, the same, automatically — so margin is really the margin you decided, the active customer is really the one under the agreed threshold, and nobody can “tweak” the number by hand to make it look greener.

This is the piece that closes the circle: the right definitions, dropped into a system that respects them, give you a dashboard that turns green when the company is doing well for real. At that point green becomes useful information again instead of a tranquilliser. It’s the same principle as the dashboard the owner actually opens: a few numbers, trusted, that steer a decision — not lots of pretty numbers that make you feel good while the account tells another story. If you want the full picture of how you build numbers you can trust, I’ve collected it in the guide to dashboards and data products.

Change the numbers, but change the incentives too

There’s a sure way to make a good set of KPIs fail: leave the old incentives. If you decide the steering number is margin, but sales bonuses are still tied to revenue, you’ve lost from the start — people follow the money, not the slides. Coherence between what you measure, what you show and what you reward is everything. A new KPI on the dashboard and an old incentive in the payslip fight each other, and the payslip always wins.

So, when you redo the KPIs, look at the rest too: how are bonuses built? On what are people evaluated in reviews? What do you celebrate in Monday meetings? If you still celebrate “how many new customers”, the team will keep bringing you new customers who leave, however pretty your churn column is. Changing the metric without changing the incentive is like changing the road sign while leaving the satnav set to the old destination: diligent people will follow the satnav. This isn’t an HR detail: it’s the part that makes everything else real.

The same money, different views: KPIs by department

A fair objection: “but every department looks at different things”. True, and it isn’t a problem — on one condition. KPIs can and should be different by department, as long as they rest on the same base definitions. Sales looks at pipeline, close rate, margin per customer; accounts looks at collected, overdue, payment delay; marketing looks at cost per acquired customer and net customers; leadership looks at the few numbers that summarise the whole. Different views, same source of truth.

The disaster isn’t having different KPIs: it’s having different definitions. If sales and accounts mean two different things by “sale”, their KPIs will never reconcile, and every meeting will go back to being a fight. The rule is simple: common definitions underneath, specific views on top.

There’s a side effect to watch too: badly chosen department KPIs set departments against each other. If marketing is rewarded on “new customers” and customer care on “tickets closed fast”, marketing brings customers that care fobs off in a hurry, and nobody looks at whether they stay. KPIs have to be chosen also so they push departments in the same direction, not against each other.

How you explain it to the team without starting a war

Changing KPIs touches a nerve: people are attached to the numbers they’re evaluated on, especially if they were green. If you arrive and say “from tomorrow we measure something else”, resistance starts. The way that works is to show, not to impose.

Take a concrete, real case from your company — average ticket up and margin down, new customers up and net ones stuck — and put it on a sheet in front of everyone. Not “your numbers are wrong”, but “look at what this number we were celebrating was hiding”. When people see with their own eyes that the green hid a red, the change stops being a criticism and becomes a shared discovery. And involve the team in choosing the new numbers: a KPI chosen together is defended, one dropped from above is circumvented. The goal isn’t to be right: it’s that everyone finally looks at the same true thing.

It’s for you if / it isn’t for you if

It’s for you if: you have a dashboard all green but the economic results don’t follow; you measure visits, messages, followers, tickets closed and you don’t really know what they lead to; you suspect the team is working to push up numbers that don’t count; you feel that “the numbers are good” but you don’t trust that feeling.

It isn’t for you if: your few KPIs are already hooked to decisions and money, and you look at them honestly; you’re still building the base data (first you need the numbers, then the discussion of which ones count); you’re looking for reassuring confirmations more than uncomfortable truths — because a good set of KPIs will, every so often, give you bad news in time to fix it, and that’s exactly its value.

Frequently asked questions

Are vanity metrics always useless? No. Visits, messages, followers can be useful as support indicators, to understand what happens upstream. The problem is making them the target. Keep them in a corner as diagnostics, don’t put them at the centre as a measure of success.

How many KPIs should I have on the main dashboard? Few. For the screen you look at every day, four to six numbers that answer the questions that count (do we sell, do we collect, what’s changing, where) plus one or two “uncomfortable numbers”. The detail sits underneath, one click away; but the first screen has to make you decide, not contemplate.

Who decides which are the right KPIs? Leadership, with sales and accounts at the table. Not IT alone, not whoever builds the dashboard: those make the choice feasible and automatic, but the choice of what counts for the business is yours.

How do I know if a number is vanity or real? The acid test: “if this number doubles tomorrow, do I earn more?”. If the answer is “not necessarily”, it’s probably a vanity metric. Real numbers are the ones tied directly to margin, cash, customers who come back.

Do I have to redo the whole dashboard from scratch? Not necessarily. Often the definitions workshop is enough: take off the vanities, hook the remaining numbers to decisions, add the uncomfortable numbers and write the definitions. Then the product underneath has to be made capable of respecting them. It’s more a job of choice and definition than of technology.

And if the team is attached to the old numbers? It’s normal: green numbers reassure. The way through is to show, on a concrete case, how the old green KPI hid a red problem (ticket up, margin down). When they see it once, they change their mind on their own.

Are my dashboard filters fooling me? Always ask yourself a question: what was excluded to arrive at this number? If a dashboard shows only active customers, successful orders, the channel that works, it’s green because it filtered out the red. Ask for the unfiltered view at least once: that’s often where the problem you could still act on is hiding.

Is the average a good way to summarise? Often not. An average can describe a customer who doesn’t exist — half spend 20, half 380, average 200 — and make you decide for a ghost. Look at the distribution, not only the average: who are the few customers who are worth it and the many who cost you.

How often should KPIs be reviewed? Not often, but not never. Good definitions last years; but when the business changes — a new channel, a different revenue model, a different objective — a mini-workshop is worth it. The signal that it’s time: when you notice a green number no longer tells you anything useful.

Is changing the dashboard enough to change behaviour? No, and it’s the most common mistake. If you change the numbers shown but leave the old incentives — bonuses, targets, what you celebrate — the team follows the incentives. Coherence between what you measure, what you show and what you reward is what makes the change real.

Can an external consultant choose my KPIs? They can help you facilitate the workshop and avoid the classic vanities, but the final choice is yours and your leadership’s: only you know where the money is made in your company. Be wary of anyone who arrives with the “right KPIs” already ready before they’ve understood how you earn.

In one line

If your dashboard is all green but the bank account isn’t growing, you don’t need more measurement: you need the right measurement. Vanity metrics — visits, messages, tickets closed, occupancy — make you feel good and hide the problem; the numbers that count — margin, customers who come back, real revenue — are more uncomfortable and tell you the truth in time to act. You start from a table and a few hours (the definitions workshop), not from new software; then you need a product that those definitions always respect.

If you want to understand which are the few numbers that count for your company and how to build them so they’re trusted, look at the projects I’ve built or drop me a line: we start from the decisions you have to take, not from the charts that put on a show.