PLAYBOOK·
Which Loops to Close First.
The interesting question isn't whether to build AI-native infrastructure into your revenue stack. It's where to start. Here's the prioritization I run.
The interesting question isn't whether to build AI-native infrastructure into your revenue stack. It's where to start. Here's the prioritization I run.
Once a team commits to building AI into the revenue stack, the next question is always the same: where do we start? The temptation is to go showy — conversational forecasting, an AI SDR, a chat assistant that answers any question. Those are real opportunities. They're also the wrong place to start. They compound slower than the loops underneath them.
Every loop worth closing does two things. It replaces operational work you used to do by hand, and it builds institutional memory as a side effect. The second one matters more than the first. A system that captures how your team makes judgment calls becomes more valuable every month. A system that just automates a task caps out at the day you ship it.
Pick the loops that turn every human correction into permanent institutional memory. Everything else is automation.
Start here. Attribution and hygiene are the cleanest places to teach a system how your team reasons. Every correction is a rule. Every rule compounds. Three months in, you have a codified version of the tribal knowledge that used to live in a single person's head. That person becomes senior-person-doing-senior-work instead of glue-person-explaining-why-this-opp-should-be-Campaign-X. This is the loop I closed first at Homebot, and it's the reason the portfolio starts with a pipeline that went from 50% to 95% accuracy.
Once hygiene is clean, pre-call brief quality jumps by default — because the data underneath is finally trustworthy. A good pre-call brief used to cost 15 minutes of a rep's time. With an agent reading across the CRM, revenue intelligence tool, engagement data, and recent community signal, it costs 90 seconds. Reps stop prepping for the wrong things. The conversations get sharper.
This is the loop most teams never close. A rep learns something material on a call — champion is leaving, a merger is pending, the exec is skeptical of the whole category — and the insight evaporates. Closing this loop means every call becomes a structured signal in your data layer. It also means next Monday's digest or next QBR starts with a more accurate picture of the book than it would have had otherwise.
Once hygiene, pre-call, and post-call are running, renewal risk gets scored for free. You already have engagement data, call sentiment, signal flags, and a history of what healthy vs. at-risk customers looked like. Turn that into a weekly watchlist and the CSM team runs five conversations earlier per quarter than they would have.
This is the loop everyone wants to close first. Don't. Forecasting accuracy is downstream of every loop above it. Deal health is downstream of call intelligence. Pipeline accuracy is downstream of data hygiene. Close the loops in order and forecasting stops being a stress test — it starts being a readout of a system that's already working.
I've seen smart teams spend two quarters building a shiny AI forecasting feature on top of a CRM with 60% attribution accuracy and 40% of the opps missing lifecycle stages. The feature ships, the demo looks great, and nobody trusts the output because the underlying data was never fixed. Six months later the feature is deprecated and the team is back to spreadsheets, a little more cynical than before.
Close the unglamorous loops first. The glamorous ones get easier, cheaper, and more accurate as a direct result.
If you want to compare notes on where your team is in this sequence — or where you got stuck — find me.