You Can't Produce Love by Rewarding Its Simulation

Scroll through the current crop of VP Marketing job advertisements and you will find a line that sounds unremarkable: accountable for Brand Affinity score, target +15% year on year, bonus tied to achievement. It sits there between the stock options and the benefits package, and most hiring managers read past it without pausing.

That one line does something almost nobody in the hiring process notices. It hands a single person personal accountability for a property that no single person, and no single function, can actually produce.

Picture the VP on her first day. She inherits the number. Her bonus depends on moving it. She cannot, by any act of her own will or craft, make customers feel affinity. So she reaches for the two levers she actually has.

Downward, she allocates rewards. A team bonus pool for the marketing team. A kicker on the agency's retainer. A gift-with-purchase for customers who fill in the affinity survey. Within a quarter, every person in the chain is being paid to produce the number she needs.

Upward, she runs the whole thing as a program. Call it the Brand Affinity Initiative. It has a steering committee, a named program lead, a quarterly roadmap, and a monthly status report in traffic-light colors. Every review cycle she walks into her own performance conversation with a deck that shows progress against her goal: workstreams in flight, milestones delivered, investments on track.

The money flows down the chain. The evidence flows back up it.

Even in the quarters when the number refuses to move, the program itself is proof that her goal was being pursued with discipline.

Nobody in any of those conversations asks whether this produces affinity. They already know the answer. They have watched it happen to NPS. And to CSAT before that.

The move nobody names

Here is what most conversations about metrics skip. Between "the VP's bonus depends on affinity" and "customers are gaming the survey" sits a quiet pair of decisions nobody says out loud. Call the first one the cascade and the second one the program. I will use those two words for the rest of the piece, and they do different work.

The cascade runs downward. Other people's rewards now depend on her number moving, and each downstream reward is a small side-contract paid in service of the upstream bonus.

The program runs upward. She is standing up an initiative whose only job is to make her pursuit of the number legible to the people who grant her the bonus. It converts her private accountability into an official organizational activity. The cascade produces the signal. The program produces the evidence of effort.

This is what rewarding others for your own success looks like in practice. It is not corruption. It is not laziness. It is the only rational response available to a manager who has been handed personal accountability for something her own work cannot directly produce. The chain below buys her the number when the number will come. The program above protects her when it will not.

The deeper reason for both levers is the one most organizations avoid naming. The VP's department, running at full capacity with her best people, cannot produce the number she is accountable for. Not because the team is weak. Because the thing is not made in marketing.

Affinity is produced by the product: whether it works, whether it lasts, whether it solves the problem the customer bought it to solve. It is produced by pricing, in the sense of whether the customer feels fairly charged or quietly resented. By support, meaning whether a complaint gets resolved in three minutes or three weeks. By delivery, meaning whether the package arrives on the day it was promised. And by how the company treats its own frontline staff, because unhappy staff produce unhappy customers. Affinity, in the end, is produced by how the organization behaves across every surface a customer touches.

Marketing runs almost none of those surfaces. Marketing runs campaigns, messaging, content, and surveys. It can name affinity, pursue affinity, measure affinity, and report affinity. It cannot make it.

Now the two levers stop looking like a choice of incentive design. The cascade reaches outside her department to do work her headcount cannot. The program reaches across her department and attaches her name to whatever improvements product, support and delivery produce. Without the program, any favorable move in the number belongs to whoever actually made the product better. With the program, it belongs to her initiative.

By the time the number moves, the number is measuring something real. Just not what it claims to measure. It is measuring the depth of the chain she has had to build around a department that cannot make the thing.

One lineage among many

The customer-love lineage is the most visible example of this pattern, and it has been running for thirty years. CSAT in the nineties was decomposed into call-center bonuses and collapsed under survey fatigue. NPS in the two thousands was tied to executive bonuses and sales commissions, collapsed into score-begging at car dealerships, and was eventually declared broken by its own inventor, who built a successor nobody has attached to bonuses because it cannot be faked. Brand Affinity and its siblings are now further along the same arc, wrapped in CX programs and steering committees that will outlast them exactly as the NPS programs outlasted NPS.

The metric rotates. The program through which it rotates is permanent.

Awesome signal, terrible goal

Every proxy metric starts life the same way. Somebody notices that a hard-to-measure thing the company cares about correlates, roughly, with something easier to measure. The easier thing becomes a proxy, and used as a signal, without consequences attached, the proxy is genuinely useful. Story points point at delivery throughput. Engagement survey scores point at attrition risk. Hospital readmission rates point at care quality. Police stop counts point at active patrolling. Teacher test scores point at whether students learned the material.

Each is worth looking at. A drop in velocity is worth a conversation with the team. A collapse in engagement is worth investigating. A spike in readmissions is worth a clinical review. This is the proxy doing its real work: telling you where in the system to look. Every one of these is an awesome signal.

Then each one gets promoted. The proxy becomes a goal. A bonus is attached. A quarterly target is set. A program is stood up around it. The same number that was being read to understand the work is now being used to measure the people doing the work. At that moment the proxy stops being an awesome signal and becomes a terrible goal.

Velocity becomes story-point inflation: teams estimate higher to show a bigger number while actual throughput flattens. Engagement scores become manager-coached ritual: "anything below a four is a failure for our team, please keep that in mind." Hospital readmission rates become patient-turfing: discharge a fragile patient earlier and let them surface at a different hospital's emergency room instead of yours. Police stop quotas become stops-for-the-sake-of-stops: any reason will do, clock the number. Teacher test scores become teaching to the test: the curriculum narrows to exactly what will be measured and the broader learning the test was supposed to approximate quietly disappears. In every case the number improves and the underlying reality gets worse. The loose correlation that made the proxy useful in the first place snaps under the optimization pressure the promotion created.

Every proxy metric inside an organization is an awesome signal and a terrible goal, because the thing that makes it a useful signal, a loose correlation with something you care about, is exactly the thing that makes it corrupt as a target. The moment somebody is paid to move the proxy, the proxy starts measuring their ability to move it instead of the thing it was meant to point at.

Two classes, one escape hatch

Not every terrible goal is terrible in the same way. Look back at the list of proxies from the last section. A fault line runs through it. Some of those metrics are produced by a function that, if you drew team boundaries carefully, could actually own them. Others measure something no function owns at all.

Start with the first class. Lead time is the example I reach for first. The clock starts when work enters the team's system and stops when the customer receives it. If a cross-functional team controls all the inputs to its own delivery, it can own its lead time as a goal without the pattern triggering, because the team is the integration layer for the thing being asked of it, and because the measurement is a clock the team cannot fill in itself. Time to value works the same way. Call these operational proxies: numbers produced by a function that, with the right team boundaries, actually makes the thing the number is pointing at. The fix for an operational proxy is structural: redraw team boundaries so the unit that carries the metric is the unit that produces the underlying thing. Lead time has a fix available. It is almost never taken, but it exists.

Now the second class. These measure a different kind of thing, and the difference is worth naming before going further. Call it an emergent property: something produced by how the whole organization behaves across every surface it touches, not by any one function inside it. Customer affinity is one. It is the joint output of product, pricing, support, delivery, and how frontline staff get treated. Employee engagement is another. It comes out of how work is designed, how managers behave, how decisions are made, how information flows, how mistakes get handled. Trust works the same way. So does innovation capacity. None of these belong to any department inside the organization. They belong to the whole.

Once you hear the category, the list of metrics that belong to it almost writes itself. Employee engagement. eNPS. Brand affinity. Psychological safety. Trust. Innovation capacity. Adaptability. Hospital readmission rates. Teacher test scores.

That is a serious list. Most of what HR spends its career trying to move is on it. So is most of what clinical quality offices and school accountability regimes chase. No unit below the whole system can own any of them. There is no engagement department below the level of the whole company. eNPS and psychological-safety scores are the same underlying property seen through different survey questions: each compresses a structural condition no single function controls into a number somebody is then asked to move.

For this second class there is no team-design fix. The only non-corrupting use is to treat the number as a reading, not a target: look at it, treat it as a diagnostic signal about the system, act on the structural conditions it is pointing at, and never attach it to a bonus. Every one of these metrics, given to a manager or a department as a personal goal, triggers the cascade and the program on day one. The manager cannot produce the number, so she runs the cascade to buy other people's help producing the signal, and she runs the program to claim credit for whatever moves. This is not a failure of execution. It is what happens, structurally, whenever an emergent property is turned into an individual target.

The two classes even fail for different mechanical reasons. An operational proxy fails through gaming: under optimization pressure the proxy diverges from the thing it was approximating, and the cure is to return the proxy to a unit that controls the thing. An emergent-property metric fails more fundamentally, because of what a self-report actually is. A count of delivered stories is a measurement of something that happened in the world. The stories either shipped or they didn't, and the number is a record of that event. An engagement score is not like that. It is the employee's own description of how the organization feels to work in, handed back to you through a survey. Reward the description and you change what the employee is willing to describe. The measurement and the thing being measured stop being separate. For this class, the damage goes deeper than the signal. The underlying thing itself changes.

Customer love metrics sit firmly in the second class. So does almost every metric the HR department has been asked to own in the last twenty years.

The acute case

CSAT asked whether a customer was satisfied. NPS asked whether they would recommend you. Brand Affinity, Salience, Emotional Resonance claim to measure how a customer actually feels about the brand. All of them are reaching for the same underlying feeling, and that feeling is where the second class fails fastest.

Feelings are easier to simulate in a survey and harder to verify against behavior. A customer can truthfully report high affinity today and quietly churn next month, and nothing in the two data points contradicts the other.

The thing being measured, genuine emotional connection, is specifically destroyed by the act of incentivizing its production. You cannot pay someone into authentic affection. The moment they know the reward is for performing it, the affection is no longer what it claims to be. It becomes a transaction everyone has agreed to call by a different name. You can't produce love by rewarding its simulation.

Why nobody stops it

The pattern persists because the question "who owns this metric?" still sounds responsible inside most organizations, while the structural alternative sounds like an excuse. "Affinity isn't something I can own personally. It is produced by the whole customer experience, and we would need to redesign the experience" is a career-limiting sentence. "I own affinity and I'm going to move it fifteen points this year" is a promotion.

The first sentence ends a conversation. The second sentence starts one.

There is another reason the chain holds, and it sits on the upward side. Once the VP has allocated a team bonus pool, signed an agency retainer with an affinity kicker, and funded a customer gift-with-survey program, she has created a constituency of people who benefit from the metric staying in place. And once she has stood up the program around it, she has created a second constituency: a program lead whose role exists only while the program exists, a steering committee whose members draw organizational importance from attending it, a vendor whose contract is renewed by its continuation, a quarterly slide that senior leadership is now used to seeing on the agenda. Proposing to retire the metric threatens the pay of everyone she recruited to produce it and the standing of everyone she recruited to report on it. The ratchet clicks forward, and each new reward and each new role is another tooth on the wheel.

This is why three decades and at least four metrics have produced the same cycle. Retire the metric, keep the practice of handing emergent properties to individuals as personal targets and then wrapping them in a program to prove effort, and the next metric arrives already corrupted the day it is introduced.

Three years from now

The VP will have hit her affinity target in some quarters and missed it in others. It will not matter much. The program deck will have shown consistent workstream delivery throughout: milestones landed, investments on track, initiatives in flight. The team bonus will have been paid. The agency will have been renewed on the strength of the number it helped create on its best days. The customers who rated high will have received their gift cards and stopped thinking about it. Churn will be unchanged. Revenue per customer will be unchanged. The product will be roughly as good as it was before the program started.

Somebody at the next offsite will announce that affinity has stopped correlating with anything meaningful. The person making the announcement will, with reasonable probability, be the program lead. Retiring the old metric is part of his job now, and recommending its successor is his next career move. A new metric will be proposed, Emotional Resonance Index perhaps, or Brand Intimacy Score. The same program will absorb it without restructuring. Same steering committee. Same quarterly cadence. Same slide template with the header changed. A fresh set of side-contracts will start flowing down the chain. The signal will begin to move. The thing the signal was supposed to measure will stay exactly where it was.

The question worth asking

When your organization switched from CSAT to NPS to whatever it currently treats as real, what changed besides the label? Ask one more question:

Is the department that carries the metric the department that makes the thing?

If it is, you are holding an operational proxy with a structural fix available. Redraw the team boundaries so the unit that owns the number also controls the inputs, and the metric can sometimes be saved. Almost nobody takes this exit, but it exists.

If it isn't, what you are holding is an emergent-property metric attached to somebody who cannot produce it, and you already know what happens next. The cascade. The program. The rotation. The slide template with the header changed.

"Is the department that carries the metric the department that makes the thing?" That is the question every program standing up around an emergent-property metric is structurally unable to ask, because asking it honestly would end the program. Nobody has to forbid it. The incentives of the program lead, the steering committee, the agency, and the VP herself all already point away from it. It is not suppressed. It is simply unprofitable for anyone in the room to raise.