What it is
Before you hand your CRM over to AI agents, it pays to know what they'll be building on. This checklist turns the five steps from my article You Can't Enrich Your Way Out of Bad Operations into a one-hour working session. You find out where unreliable data actually hurts the business, what AI and enrichment can realistically recover, whether the process behind your data is clear, how you'll judge each AI use case, and who keeps it all working afterwards.
Who it's for
RevOps and Sales Operations leads, CROs and founders who are about to roll out AI in their go-to-market, or who already have and aren't quite sure they can trust what comes out of it. It works for a five-person sales team just as well as for a scale-up with several GTM teams.
How to use it
Block an hour with your RevOps lead and one or two people from Sales or Customer Success; you'll get far more honest answers in the room than from any survey. Then work through the tabs in this order:
- Diagnose (15 min): answer the five questions and rate the business impact of each problem you find. Be specific, because "our data is messy" won't help anyone decide where to start.
- Data recovery (10 min): list the data areas that matter and decide what AI and enrichment can recover with reasonable confidence, and what simply doesn't exist.
- Process check (10 min): mark each building block as in place, partly there or missing. Anything marked missing is exactly where your AI will be guessing.
- AI evaluation plan (15 min): for every AI use case on your list, write down what success looks like before you switch it on, and how much control it gets on day one.
- Make it last (10 min): every process and automation gets a name and a review date. If no name fits, you've just found your first gap.
The summary tab turns your answers into a readiness score per step, so you can see at a glance where to start. Treat the score as a compass for deciding which gaps to close before you automate.
What's inside
Seven tabs: a one-page Start here, one tab for each of the five steps (Diagnose, Data recovery, Process check, AI evaluation plan and Make it last) and a Summary that turns your answers into a readiness score per step and flags the problems worth a closer look. Teal cells are for your input, grey cells calculate automatically, and dropdowns keep the answers consistent.
| Building block | What "in place" looks like | Status |
|---|---|---|
| Qualification criteria | Written criteria for when an opportunity counts as qualified, applied the same way by every team | Partly |
| Deal progression checks | Mandatory checks for every stage change and forecast category | Missing |
| Information capture | A clear owner (team, person or AI agent) for each key piece of information | Partly |
| Automation that maintains the setup | Routing, reminders and validation rules that keep the process consistent | In place |
| Exception handling | A defined path for deals and records that don't fit the standard process | Missing |
Terms of use: Free to use and adapt within your organization. Please don't resell it or republish it as your own. Provided as-is: the examples and formulas are illustrations, not legal, tax or accounting advice.
Related articleYou Can't Enrich Your Way Out of Bad Operations →Want help putting this into practice?
Book a call and we'll look at where your operations stand and what to fix first.
Iveta Kirilova is a hands-on RevOps operator and consultant based in Berlin. She has built revenue operations across 60+ B2B companies in Europe and the US, both in-house and at a specialist RevOps consultancy. More about Iveta →