How to justify your SEO retainer when traffic falls for reasons you don't control
There's a conversation no agency wants to have: the monthly report where organic traffic falls for a quarter straight and the work has been done well. Positions hold, content gets published, the technical side is clean... and the graph drops. The client looks at the curve, looks at you, and thinks the inevitable: "What am I paying you for?".
The problem is structural, not yours. But "it's not my fault" isn't an argument that sustains a retainer. What sustains it is reframing what you're selling -and changing the report so it measures that. Let's take it in parts.
Why traffic falls even when you work well
The Spanish market data is unambiguous:
- Where Google shows AI Overviews, the average organic CTR falls by 61% in Spain (ismajimenez.com). The same position 1 from two years ago now hands out far fewer clicks, because the answer appears above the result.
- Zero-click searches -the ones that end without visiting any website- have gone from 56% to 69% between May 2024 and May 2025 (data cited on stucom.com). Seven out of ten searches no longer generate a single click for anyone.
- Gartner forecasts 25% less traditional search volume before the end of 2026 (press release, February 2024).
In other words: the pie of clicks is shrinking for the whole sector. Your client may be gaining share of a shrinking pie, and the sessions graph hides it. If your report only measures clicks, your report is making you look bad for something outside your control.
The reframe: from "I bring you traffic" to "I put you where it's decided"
For fifteen years, the implicit SEO contract was "I bring you visits". That contract has broken, and the way out isn't to disguise it: it's to renegotiate it out loud. The value you sell is no longer the click; it's presence in the places where the potential customer makes the decision, whether they click or not.
And note that the reframe isn't a euphemism for "fewer results": the traffic that does arrive is worth more than ever. Someone landing from an AI answer converts at 14.2% versus the 2.8% of classic organic (sector data compiled by roymo.es). Meanwhile, traffic from AI has grown by 527% in a year in Spain (ismajimenez.com). The honest and true narrative is: fewer clicks, better clicks, and a new surface where you have to be present.
The line for the meeting:
Key data
"The market has changed the rules: 69% of searches no longer end in a click for anyone. So from now on I'm not just going to report how many people reach your site, but in how many of the places where your customer decides -Google, ChatGPT, Gemini, Perplexity- you appear and your competition appears. Traffic will stay in the report, but it stops being the only mark on the exam."
The new metrics in the monthly report
This only works if the report genuinely changes. Sections to add:
1. Impressions, not just clicks. Search Console gives you the number of times the client appeared in front of a searcher even without a click. If impressions rise and clicks fall, you have visual proof that the problem is the distribution of clicks, not your work.
2. Conversions and their quality. If traffic falls by 20% but leads hold, the lost traffic was the kind that wasn't buying. Few graphs defend a retainer better.
3. AI visibility: the new section. In how many of the 4 main AIs the client appears when someone asks about their sector, with what description and alongside which competitors. Here almost everyone starts from zero: in the study we ran with 9,865 Spanish SMEs, 91% appear in only 1 of the 4 AIs (full study). A low baseline isn't bad news for you: it's demonstrable headroom.
4. Share of voice against competitors. "You appear in 6 of the 20 key questions in your sector; your main competitor, in 9." It's the metric that turns the conversation from "how much traffic have I lost" into "how much market can I gain". Which metrics in this block you can commit to in a proposal and which you can only report is detailed in what KPIs to put in a GEO proposal.
5. Work done. Corrections published, new content, sources fixed. In a volatile environment, verifiable effort is part of the deliverable.
Measuring sections 3 and 4 by hand -testing dozens of questions across 4 AIs, every week, for each client- doesn't scale past the first month. We solve it with Surfeo for agencies: each client in their own workspace, the 4 AIs queried every week with their sector's prompts and the result exportable as a PDF for your usual report.
The conversation, step by step
- Don't wait for the client to raise it. If you explain the drop before they discover it, you're the expert who gets ahead of it; if they explain it, you're the supplier making excuses.
- Bring the market data, not just theirs. The −61% CTR and the 69% zero-click turn "your agency is underperforming" into "the playing field has changed for everyone".
- Show them the part you do control: impressions, conversions, share of voice and the AI visibility baseline.
- Propose the expansion, not the discount. The classic mistake is to lower the price to make up for the graph. What's needed is to widen the scope: SEO + AI visibility, with its own price. If the client comes straight in with "I want to cancel because everyone only uses ChatGPT now", that specific script is here.
Frequently asked questions
Isn't it a trick to change the metrics just when they drop?
It would be if you hid the traffic. Don't hide it: leave it in the report, falling if it falls, and add the context and the new metrics alongside. Changing metrics in secret is make-up; changing the frame out loud, with market data, is consultancy. The difference is transparency.
What do I do if the client only wants to hear "how many visits"?
Translate the new metrics into their language: "Of every 100 people who found you, 8 used to visit and 1 bought; now 5 visit and 1 buys. You've lost visits that weren't buying and you're still selling the same." If even so they only value volume, show them the AI traffic conversion figure (14.2% vs 2.8%): less water, more fish.
Should I lower the retainer if traffic is going to keep falling?
No, because the work doesn't decrease: it shifts. The hours that used to go to chasing positions now go to sources, structured data, reviews and monitoring 4 AIs. In fact the scope grows. The market reference is still the one for SEO in Spain: 600-4,000 €/month (pacoruben.com), and the AI visibility module is added, not discounted.
How often do I report AI visibility?
Weekly measurement, monthly report. AI answers change from one week to the next, so a one-off figure misleads; the average of 4 weeks is defensible. Never promise a specific number of mentions: commit to measuring, correcting and showing the trend.
The retainer that rests only on a traffic graph is on borrowed time; the one that rests on measurable presence in every place where it's decided is not. Start with the baseline: run the free AI visibility test with your three most nervous clients and put the new section in the next report.
