Churn conversations are rarely about deliverables. They are about a client who cannot tell whether the last two quarters bought them anything, deciding under uncertainty, and deciding against the thing they understand least. SEO is nearly always the thing they understand least.
Which means retention responds to the same discipline as measurement. Not more calls, not a prettier deck.
What actually moves it
- Set the pace expectation in month oneA verdict on a change made today arrives in about six weeks. Say that at the start, in writing, and you convert 'nothing is happening' into 'this is on schedule'.
- Show the comparison, not the trendA rising line invites the client to wonder what else was rising. Changed pages against untouched pages answers the question before it is asked.
- Publish the missesThe fix that did nothing, in the same table as the wins. It costs one row and buys the credibility of every other row.
- Retire what does not payIf a class of change keeps producing no effect on this site, stop recommending it and say you are stopping. Clients notice a shortening list of recommendations far more than a lengthening one.
- Keep the record where it survives peopleAccount managers change. A ledger of what was tried on this site and what it did is the thing that usually walks out of the door with someone.
The conversation this produces
The renewal meeting stops being a defence. It becomes a review of a ledger: eleven changes landed, six measurably positive, three no effect, one regression that was reverted, one still inside its window.
That is a document a client can take to their own board, which is often the actual decision-making unit and one you never meet.
What about clients whose developers never ship anything?
The ledger makes it visible without an accusation: recommendations raised, and what is still open with dates. Most of these relationships improve once the delay is a shared fact rather than a suspicion on both sides.
Does this work for a brand-new site with no history?
Partly. Measurement needs impressions to compare, so the first months are honestly foundational work. Say so rather than manufacturing a metric that looks like progress.
Try it on your own site, free
No card, and the free tier does not expire — the first verdict takes about six weeks to arrive.