The Drift Tax

I've been tied at the hip to a CRO or CMO on a pipeline number at three companies: different stages, different products, different markets. The same problem showed up at every one of them, and I've watched it show up again at nearly every Fabricant client since.

It wasn't lead volume. It wasn't rep activity. It wasn't the tooling. Those are the things most teams reach for first, and they were rarely the thing actually breaking.

The thing breaking was quieter. The one call the entire go-to-market team is supposed to agree on, who we should be selling to, was almost always out of date. And nobody noticed until it had already cost us a quarter.

The most dynamic thing in your business, treated like the most static

Your Ideal Customer Profile is the single artifact Sales, Marketing, and RevOps all have to read from. It decides who reps prospect, who marketing spends against, how you segment, what "qualified" means, and which deals leadership leans on in the forecast.

So you'd think it would be the most carefully maintained thing in the company. In my experience, it's the opposite.

Most teams define their ICP once: at an offsite, in a slide, right after a raise. They write it down. They "enable" the team on it once. Then it gets buried in a deck or a Notion doc, and it sits there.

Meanwhile the business moves underneath it. Every week:

1. You win a deal you didn't expect to win.
2. You lose one you were sure of.
3. A deal moves twice as fast as the one next to it.
4. A customer renews and expands.
5. Another one churns six months in.

Each of those is the business telling you something about who you should actually be selling to. And in most companies I've seen, there is no systematic way to calibrate and enable the company's ICP. So by the middle of the year, a real chunk of your team is executing against a definition that no longer matches what's true.

That gap has a cost. I've started calling it the ICP drift tax.

The drift tax shows up in three places

Wasted selling capacity & inefficient marketing spend

Reps spend weeks on accounts that were never going to close. Picture a team where a third of the pipeline is bad-fit. That's a third of your most expensive capacity working deals that shouldn't exist, and it compounds every quarter you don't catch it.

Win rates that quietly sag

When a large share of pipeline is accounts that won't convert, your aggregate win rate drops. Now you can't forecast cleanly, and you overfill the top of the funnel to compensate, which pulls in more marginal accounts. The drift feeds itself.

Customers you regret

This is the one most teams miss, and it's the one I've felt the hardest. Some bad-fit accounts do close, often the ones that closed fastest, because they were the easiest to talk into it. And then you spend the next two years paying for it. They churn early, or you bend over backwards to keep them, building features for edge cases one unhappy account demanded, burning out CS on accounts that were never going to be happy, selling more customers just like them because they look like wins on the board. The whole company starts wasting time and energy on the wrong accounts, and your read on "who's our best customer" gets skewed by the customers you least want in the first place.

That's where it stops being a sales problem and becomes a company problem. When you target the wrong accounts, every GTM function suffers and so does the business. Growth slows way down. I've watched great companies with great products run in sand when they could have been racing toward a great exit instead.

Here's the part almost everyone gets wrong

Most tools used to tune ICP look at what deals close; win/loss at the point of sale.

That's the commodity move, and it's only a single slice of the picture, because close-time data can lie to you. It rewards the deals that were easiest to win, not the customers who were best to keep.

The version that actually tells you the truth is tuned on the full loop: how your pipeline behaves (win, loss, velocity, stage movement) and how your customers behave after they buy: renewal, expansion, upsell, churn.

Tune on what closes, and you'll keep chasing accounts that look like your fastest deals. Tune on what closes and then renews and expands, and you start chasing accounts that look like your best customers. Those are not always the same accounts. Lifecycle data is what tells you the difference.

What a living ICP actually looks like

The fix isn't to define your ICP better, once. It's to stop treating it as a document and start treating it as a system, something that recalibrates as the business produces new evidence.

In practice that takes three things, and the hard part is that most teams have the bandwidth for none of them:

Do that as the business produces new evidence and the ICP stops drifting. It stays sharp instead of going stale.

Why I built Goose

I've been quiet lately. For the last seven or eight months I've been heads-down building this product, because I got tired of watching the same problem cost the same quarters at company after company, and because keeping an ICP alive is real, grinding work that most teams will never have the headcount to do by hand.

Now it's time to stop being quiet about it.

Goose keeps your ICP alive. It grades every account in your CRM against a model that recalibrates on the full loop (pipeline behavior and post-sale customer behavior) and lands that back where your team already works, on your sign-off, with the reason attached. The leader gets the narrative. The reps get told which accounts are actually worth working. Same engine, both seats at the table.

Not that those companies can't or won't eventually get there; most of them will. It's that growth doesn't need to be as painful as it so often is. So much of that pain comes from too much of the org pointed at the wrong accounts. Keep the ICP honest and you take that tax off the top. I built Goose to do just that.

I'm not going to tell you it's magic, and I'm not going to pretend it's finished. It's early, and I'd rather be honest about that than oversell it. But the argument underneath it is one I've lived through more than once, and I haven't found a team yet that didn't recognize it.

If your ICP is a slide you update once a year, it's drifting right now. The question worth asking is whether anything is listening.