The dip is not the verdict.

Three weeks into it, Wes wanted out.

He was a senior rep, the kind whose numbers a region gets built around, and three weeks earlier he had stood up in a rollout meeting and asked the question everyone else in the room was thinking and nobody else said out loud. Now his pipeline was moving slower than it had in years. Deals were slipping a week, then two. His manager had started to notice, by name, which rep was struggling with the new system, and Wes was that rep.

He told his manager he was done. It was too hard. It did not make sense. The old way worked, and he had been doing this for years. Then the line every senior rep reaches for when a change is testing him: you can’t teach an old dog new tricks.

I want to be plain about what Wes is. He is a composite, built from more than one rep I have watched go through this exact stretch, folded into a single story so the whole arc is visible at once. What follows happened, in more than one rep, from where I sat.

His manager set up a call. I was on it. Wes laid out the case for reverting. When he finished, I asked him one question. What have you done to learn the system? He said he had done the training. “Wasn’t that enough?” I asked him what he thought.

There was a long silence, the kind that happens when a question lands somewhere it was not built to land. It was not the silence of a man preparing his next argument. He had produced four weeks of arguments already, one after another, without a pause. This time nothing came back. When he finally spoke again, the language had changed. He could ask for help. He could take the classes the company had already made available. He could ask a colleague who had figured out the system to walk him through their week.

He built that list himself. Nobody handed it to him. Everything on it had been sitting there the whole three weeks. He had not been unable to see those options. He had been protected from seeing them by a set of explanations whose entire job was to make the system the problem instead of the work.

How do you tell a refinement that is working from one that is failing?

From week three, you cannot, not by watching the outcome. A refinement that is working looks, for a stretch, exactly like one that is failing. The book’s own glossary names that stretch the friction weeks, the period after a new behavior is installed and before it produces anything, and puts a number on it: most friction weeks run two to twelve weeks. Wes was three weeks into a stretch that, by the book’s own math, was not even halfway finished.

Robert Bjork, UCLA, quoted in Daniel Coyle, The Talent Code, 2009

Bjork spent a career studying how skill is actually built, and his finding lands harder than most people expect: the gap between what you know and what you can currently do is not the obstacle. It is the location of the learning itself. Wes’s three weeks of being slow and clumsy on a system he had publicly resisted were not the price he paid before the improvement started. They were the improvement, happening, in a stretch that felt to him like proof he had made a mistake.

The two questions get confused constantly, and the confusion is what ends refinements early. Is this working is a question about outcomes: are the numbers moving, is the manager happier. Did I run it is a question about behavior: on this call, did I do the thing I said I would do. In the short run, outcomes are mostly noise. Behavior is the only variable a rep controls. Confuse the two, and a bad week of noise talks a rep out of a change that was never given a real chance.

1. Name the behavior, not the mood you are trying to fix.

Write the top line first, in mechanical terms. Not be more assertive on calls. Ask the budget question on the first call. One of those is a behavior a rep can run or not run, call by call. The other is a feeling wearing a strategy’s clothes, and nobody can grade a feeling honestly.

2. Put a start date and an end date on it before you run it once.

Most refinements never get a date, which is exactly what lets a rep quit one at week three and call it a decision. Pick a stretch inside the book’s own two-to-twelve-week window. Six weeks is the honest default: long enough to sit inside real friction, short enough that quitting early actually shows up as quitting instead of patience running out against a schedule nobody wrote down.

3. Run two columns and grade only the one on the left.

Left column, did I run it, ticked once per call. Right column, what happened. For the length of the card, the rule is the whole artifact: you are graded on the left column only, and you do not read the right column, not once, until the end date arrives. Single-deal exposure already made an adjacent case for this kind of restraint. When one deal carries more than 40 percent of the quarter, it fails to close about four times out of five, and no rep would use a single deal to call a whole territory dead. A single week of a new behavior deserves the same suspicion before it gets to talk anyone out of anything.

4. Hand your manager the other half of the card.

A card only holds if the room around the rep grades it the same way the rep does. The one-on-one opening question changes, from how is it going to how many times did you run it. For the length of the card, the manager does not ask about that behavior’s results, not once. The results are not due yet. The card has a due date on it already; nobody needs to guess one in a hallway.

The spike and the system.

Grading a refinement off the scoreboard on day three, because the dread got loud, is a spike. It feels like judgment. It is panic borrowing judgment’s clothes. Grading a refinement off the behavior, on schedule, for the length of the card, whether the week was good or bad, is a system.

Thirty-plus years carrying a number, and the pattern holds in every room I have watched it happen in: the reps who make it through a real refinement are almost never the ones who felt the least fear in week three. They are the ones who had already agreed, in writing, on what they were being graded on before the fear showed up. That is Isolate, Refine, Compound at its smallest working scale. Name the one behavior. Change it on purpose, on a schedule you set before the friction started. Let the weeks you already agreed to pay for finish paying you back.

Run this.

The Friction Card. One index card, or one line in a notebook. Nothing more than that.

1. Step 1. Write the top line. One sentence, the behavior, no adjectives. If a stranger reading the card could not watch you do it or not do it on a single call, rewrite it.
2. Step 2. Write the two dates. A start date and an end date, inside the two-to-twelve-week window the friction weeks actually run. Six weeks is the default. Defaulting to it beats debating it.
3. Step 3. Draw two columns. Left: did I run it, one tick per call. Right: what happened. Nothing else goes on the card.
4. Step 4. Grade the left column only, out loud, once a week. Same slot, same day. A number out of however many calls the behavior was in play. Do not open the right column before the end date.
5. Step 5. Hand the manager’s half over. The one-on-one opens with how many times did you run it, not how is it going, for the length of the card. Results on that behavior are off the table until the end date, because they are not yet the data that answers the question a refinement is asking.
6. Step 6. On the end date, and only then, read the right column. Compare it to the left. That comparison is the only honest evaluation a refinement gets.

What it changes.

1. Time. A dated dip is survivable in a way an indefinite one is not. Six weeks with an end date on it is a stretch a rep can hold. An open-ended bad patch is what gets a rep to quit in week three and call it a decision.
2. Opportunity. The left column is the week’s actual work, visible and countable, instead of a feeling a rep is trying to talk himself out of.
3. Compensation. This is the honest center of the card. A rep who quits a refinement at week three keeps the exact behavior that produced the number he already has, which means that number is the number he stays capped at. The card does not remove the dip. It prices it: a start date, an end date, and a known width, instead of no edges at all.

The short version.

1. A refinement that is working looks exactly like one that is failing, for two to twelve weeks. Six weeks is the honest default to plan against.
2. The only thing that tells the two apart is whether you are grading the behavior or the scoreboard.
3. The Friction Card: one behavior named with no adjectives, a start date, an end date, and two columns.
4. The rule is the whole artifact: grade the left column only, and do not read the right one until the end date.
5. The manager’s one-on-one question changes from how is it going to how many times did you run it, with no results discussion on that behavior until the date arrives.
6. Single-deal exposure already proved this at the pipeline level: one deal is too small a sample to trust. One week of a new behavior is too.
7. A rep who quits a refinement early keeps the behavior that produced his current number. The card is what makes the alternative affordable.
8. Isolate, Refine, Compound at its smallest scale: name it, run it on a schedule, let the weeks you already agreed to pay for finish paying you back.

How do you tell, from the outside, whether a rep is still running the new behavior or has quietly slipped back to the old one?

This essay ran as edition 18 of Frontline to Forecast, the weekly newsletter. One issue a week, published here on the Friday, then emailed and posted to LinkedIn the same morning. Same piece everywhere. Get the next one. Disagree with it? Reply to any issue, or find me on LinkedIn.