What to collect from sales
- For every identified company contacted: reached, meeting, opportunity or no fit.
- The reason for no fit: too small, wrong industry, student, competitor.
- The questions prospects asked on the first call.
The monthly review
- List ad groups by meetings, not clicks.
- Pause ad groups with three months of no-fit companies.
- Add first-call questions as new context hints.
- Turn the best questions into pages for AI visibility.
- Adjust the fit score if sales disagrees with it.
Make it easy for sales
One field in the CRM, one dropdown, one Slack thread. If feedback takes more than a minute, it won't happen.
A one-field CRM setup
Add a dropdown on company records for AI-sourced accounts: "Reached – meeting", "Reached – later", "Reached – no fit: size", "Reached – no fit: industry", "Not reached". Reps choose one after each first contact.
Questions
How do we get reps to fill it in?
Make it one click, and show them monthly how their feedback changed the ads.
What if sales says all AI accounts are bad?
Look at the fit scores and questions together. Usually a few ad groups cause most of the poor fits.
Should sales see the buyer questions?
Yes. They make better openers than any template.
Why the loop matters more for AI ads
In ChatGPT, you don't see the buyer's words, so you can't read search terms to learn what works. Sales conversations are the closest thing you have to a search term report: they reveal what the buyer was trying to solve. Feeding that back into context hints is how campaigns get sharper over time.
A worked example (fictional)
Sales notices that companies from the ad group "close software for NetSuite" keep mentioning multi-entity consolidation on calls. Marketing creates a new ad group, "close and consolidation for multi-entity NetSuite groups", writes a page for it and moves 15% of budget. Six weeks later the new ad group produces more in-profile companies at a lower cost than the original.
Keep the loop small
One field, one monthly meeting, one change at a time. Big quarterly overhauls lose the cause and effect.
Start here: AI ads for B2B. More articles in Measurement and reporting.
