The one number a commercial team should share.

The offsite runs from nine to five. There is a facilitator, there are two teams who have been briefing against each other for three quarters, and by four o’clock there is an agreed definition of a qualified lead on a whiteboard, photographed on the way out. Six weeks later the definition lives in a document, and the argument is back.

Nothing was wrong with the definition. The problem is that alignment was treated as a matter of agreement, when it is a matter of measurement — and nothing about how either team was judged changed at any point during the day.

A commercial team is aligned when there is one number both sides can move, both sides are measured on, and neither side can move alone. Everything else is a workshop.

First, stop trying to make them think alike

The instinct behind alignment work is that marketing and sales see the world differently and that this is the problem. The best evidence available says the second half of that is wrong.

Christian Homburg and Ove Jensen tested exactly this question — which differences between the two functions actually damage performance — and published the answer in the Journal of Marketing. Their finding is more useful than the premise it corrects:

It confirms that, in general, differences hamper the cooperation between marketing and sales, which leads to a lower market performance of the business unit. However, some facets of thought-world differences enhance the market performance of the business unit through a direct effect that outweighs the negative effect mediated by the quality of cooperation.

And then, specifically:

Market performance is enhanced if one side plays the customers’ advocate while the other plays the products’ advocate. Market performance is also enhanced if one side plays the advocate of short-term considerations while the other plays the advocate of long-term considerations. In contrast, differences between marketing and sales in regard to product knowledge and interpersonal skills are deleterious to market performance.

Read the last sentence twice. The differences in orientation are worth keeping. The differences in competence are the ones doing damage. A team where marketing argues for the long term and sales argues for this quarter is doing something the paper’s evidence says improves market performance. A team where the two sides know the product to different depths is carrying a real cost.

The alignment exercises I have sat in attacked the first and left the second untouched: a day spent talking two functions into a single point of view, which removes the productive tension, followed by a return to a training budget that was not the subject of the day.

Which reframes what a shared number is for. It is there to make the disagreement safe — to give the argument a referee that neither side appointed.

What has to be true of the number

The idea of one shared number is old, and it belongs to someone else. Philip Kotler, Neil Rackham and Suj Krishnaswamy set it out in Harvard Business Review in 2006, in the article that still frames this whole subject, and they name it exactly: “On a macro level, companies like General Electric have ‘the number’—the sales goal to which both Sales and Marketing commit.” Their prescription is shared revenue targets, shared reward systems, and what they call a “rise or fall together” culture.

The mechanism underneath it is older still. Morton Deutsch’s 1949 theory of co-operation and competition described goal structures in which the parties’ attainments are positively correlated — what David and Roger Johnson later systematised as positive interdependence and stated as the condition where individuals “can attain their goals if and only if the other individuals with whom they are cooperatively linked attain their goals.” That is condition two below, published in a social-psychology journal seventy-five years early.

So the argument is settled, and has been for twenty years. The open question is the one a commercial leader needs answered on a Monday: given a specific candidate metric, how do you tell whether it qualifies? Kotler prescribes the shared target without saying which targets can carry one. Deutsch and the Johnsons define the property without operationalising it against a compensation plan. In between sits the decision anyone actually has to make — pipeline created, or qualified pipeline that converts? — and neither answers it.

Few metrics can carry this. Five conditions, and a candidate that fails any one of them will make things worse rather than better.

1. Both sides have to be able to move it. This is the condition that eliminates the usual candidates. Marketing-qualified leads can be moved by marketing alone, which makes them a marketing target wearing a shared badge. Win rate can be moved by sales alone, and the cheapest way to move it is to disqualify anything difficult.
2. Neither side can move it alone. The mirror of the first, and the one that does the real work — positive interdependence, stated so that a compensation plan can be checked against it. A number that one function can hit without the other is a number that will be hit without the other.
3. It has to be a customer outcome. Stage counts, activity totals and CRM hygiene scores measure whether the system was fed. They can improve while nothing happens in the market.
4. Gaming it has to require doing the work. The best shared numbers have this property structurally: the shortest path to moving them is the thing you wanted anyway.
5. It has to be one number, in both rooms, on the same slide. Two numbers is two teams. A composite index that cannot be decomposed is worse than either, because when it moves there is no way to say why.

That third and fourth condition together rule out more candidates than teams expect.

The objection, which is serious

Reducing a commercial team to one number is exactly the thing that people who study measurement warn about. The concern is old enough to be foundational: the first volume of Administrative Science Quarterly, in 1956, carried a paper by V. F. Ridgway titled “Dysfunctional Consequences of Performance Measurements.” Donald Campbell later gave the general form its name — the more heavily a quantitative indicator is used to make decisions, the more pressure it comes under to be distorted, and the more it distorts what it was meant to describe.

Both are right, and neither is an argument against a shared number. They are an argument about what the shared number is for.

The one number is the only shared measure. Marketing keeps its own diagnostics, sales keeps its own, and both keep the instruments they need to run their own function. What has to be singular is the thing both are judged on together — because that is the only place where the incentive to cooperate can live. A team with fifteen shared metrics has no shared metric; it has a dashboard, and a dashboard is a place where two functions can each find evidence that they are doing fine.

The gaming risk is real, and condition four is what handles it. If the shortest path to moving the number is the work you wanted, gaming and doing converge. If it is not, no amount of supplementary measurement will save it — it will just take longer to notice.

A second objection has a real tradition behind it. The marketing–sales service-level agreement solves the same problem with two reciprocal numbers instead of one: marketing commits to volume and quality, sales commits to response time and follow-up, and each side is held to something the other can verify.

That works, and where it is working it should be left alone. But notice what it is. Two numbers, each owned by one function, with a promise attached to each. It makes the handoff auditable, which is worth having on its own terms. What it leaves out is condition two: a quantity neither side can move alone. An SLA is how two functions stop blaming each other. A shared number is how they find out which of them was right.

There is a second-order point here that connects to the difference between work that accumulates and work that resets. A number chosen well is a stock measure: it asks what the commercial system is now capable of. A number chosen badly is an activity measure with a customer-sounding name, and an activity measure resets with the quarter and takes the team’s attention with it.

Does collaboration actually pay?

Worth checking, since the whole exercise rests on it. Kenneth Le Meunier-FitzHugh and Nigel Piercy examined whether collaboration between the two functions shows up in business performance at all, and reported that “there is a direct and positive relationship between these two constructs”. They identify five antecedents, and the list is instructive because of what is on it and what is not: senior management attitude toward collaboration, reduced interdepartmental conflict, improved communications, organisational learning, and effective market intelligence systems.

Note what does not appear: a workshop. Four of the five are structural conditions that leadership sets, and the first one is simply whether the people above both functions behave as though collaboration matters. A shared number is the cheapest available way to make that behaviour legible, because it is the one thing that shows whether leadership meant it. This is also the terrain of my doctoral research, where sales–marketing integration is one of the two things under examination, so I hold the general claim more loosely than the specific ones.

Try this in your next pipeline review

The exercise takes ten minutes and it is uncomfortable, which is the point.

Ask the head of marketing and the head of sales, separately and in writing, to name the one number they believe the other is optimising for. Not the number they think the other should have. The number they believe the other is actually being judged on today.

Then compare the two answers with each other, and with what the compensation plans actually say.

Three outcomes, and each tells you something different:

• They match each other and match the plans. You are aligned, and whatever is going wrong is not an alignment problem — stop spending on it.
• They match each other but not the plans. Both leaders understand the situation and both know the incentives point elsewhere. This is the easiest of the three to fix and the one I have seen left alone longest, because fixing it means opening a compensation plan.
• They do not match each other. The two functions are running on different models of what the other one wants. No definition of a qualified lead will survive that, which is why the last workshop did not.

Whatever the answer, you now have the alignment problem stated as something with an answer, rather than as a feeling that the two teams do not get on.

The reason this is worth the discomfort is that the alternative keeps working — as a meeting. Alignment sessions produce genuine agreement in the room, and the agreement is genuine right up until the first week where one side can hit its own number by doing something that costs the other side theirs. At that point people do what they are measured on. That is the system operating exactly as designed.

Pick the number. Put it on both slides. Then find out how much of last quarter’s disagreement was ever about lead quality at all.