Good reads · Guide
What to report instead of rankings
In 2014 I brought Rand Fishkin a tidy click through stat and he wasn't having it. These days I'd put rankings near the back of the report.
At MozCon in 2014 I went into my interview with Rand Fishkin holding a statistic. Click through rate studies from Moz and Dr. Pete, I told him, suggested the top three spots took about 40% of the clicks.
Rand wasn't having it. He'd seen the studies too, from WordStream, Conductor and others, and what struck him was how much they disagreed:
"The standard deviation on all of those studies is so insanely high that what it's suggesting is there is no percentage you should count on for ranking number one for any given term or phrase. Not even on average."
It was a tidy number, the kind that ends up in a pitch deck. Rand's point was that it didn't mean much.
Twelve years later there often isn't a single position one to count on at all.
Position one was always a bit of a story
Rand's advice for clients who fixate on one pet keyword was to change the question. Don't promise the ranking. Ask: "how do I help you make twice as much from your search traffic next year as you made this year?" Rankings are part of that, he said, along with "more traffic, more better conversions, better user path". And if a client still judges you on whether you're number one for a term? "There's a certain kind of SEO for that."
Even then, the ranking report was a simplification. When I asked Gary Illyes about clicks, he explained that Google does use them for personalisation, because with one person the signal is clear: "if they want to mess up their own search results, it's like sure, go ahead." Your customer's results page and your rank tracker's results page were never quite the same page.
Mike King once argued with an agency CEO over this sort of thing. They wanted him to promote an internal tool that tried to reverse engineer "not provided" keyword data from old numbers, which he said "inherently doesn't make sense because you're getting new queries every day." It was "not something to like hang your hat on". A lot of ranking reports are the same: a confident number standing in for something nobody can see.
In 2026 there isn't one page to rank on
What a "number three ranking" is up against in 2026:
- AI Overviews sit on top of a big share of searches. Advanced Web Ranking's July 2026 CTR study of US desktop results found position one got about 20% of clicks overall, and under 10% when an AI Overview was on the page.
- AI Mode is mainstream. Google said at I/O in May that AI Mode passed one billion monthly users. It reached Australia in October 2025. Google's own documentation says AI Overviews and AI Mode may use "query fan-out", running several related searches to build one answer. Your page might be cited for a question nobody typed.
- Results are getting personal. The same I/O post announced Personal Intelligence in AI Mode for people "in nearly 200 countries and territories", with no subscription needed. People who opt in let AI Mode draw on things like their Gmail and Photos. Gary's "mess up their own search results" now comes with their inbox attached.
- Fewer searches send anyone anywhere. Rand's company SparkToro found 68.01% of US Google searches ended without a click in the first four months of 2026.
We wrote about the answer box years ago and asked, "if searchers get their answers easily, why would they go to your site?" That question now covers most of the page.
Search Console doesn't settle it. AI Mode clicks and impressions are counted inside the ordinary "Web" totals, and the generative AI reports Google added in June show impressions but no clicks, queries or position. So a rank tracker's position is one logged out check from a data centre, of a page your customer probably never saw. It's fine as a diagnostic. It isn't a result, and it's a poor thing to be paid on.
What our guests measured instead
Wil Reynolds put it most bluntly. People get excited that their quality scores went up, he said, "but like, it's about getting revenue in the cash register for our client." Then:
"We are not here to get them links or rankings. What we believe is that rankings are part of the steps we have to go through to help them grow their business. But you and I know it's completely possible to get somebody to rank highly for something that doesn't actually help them grow their business and brings them the wrong traffic."
That's Wil at 42:14. Wrong traffic looks brilliant on a ranking report.
When I interviewed Jeff Sauer in 2022, he said every marketer should own one KPI at most, and that it has to line up with what the business counts. Marketing KPIs that don't are "the reason why marketing teams get laid off all the time." Marketers praise each other for "impressions, engagement, reach, and all that crap", he said, but "that's not what the company cares about", and "executives just look at them like they're clowns." His full answer is in the written interview.
He was sceptical of fancy attribution too: "Probably 1% of businesses would benefit from it at most." More on that in data informed, not data driven.
And Ian Lurie described the chain that still holds in AI search. "If you have a compelling experience for people, they will link to you and they will cite you, and that will help you move up in the rankings." Rankings come last in that sentence. Being cited is the part worth measuring now.
What to put in the report instead
The outline I'd use, roughly in this order: The first page should make sense to someone who has never heard of Search Console.
| Section | What it answers | Where the numbers come from |
|---|---|---|
| 1. The business number | Did organic search bring in money or good leads? | CRM or sales data, GA4 key events, compared with last year |
| 2. Qualified visits | Are the right people arriving on the pages that sell? | Search Console clicks by page group, GA4 engaged sessions and key events by landing page |
| 3. Brand demand | Are more people searching for us by name? | Search Console branded vs non-branded queries |
| 4. AI visibility | Are we being shown, named and cited in AI answers? | Search Console generative AI impressions, a fixed set of prompts checked monthly, referrals from AI assistants in GA4 |
| 5. Search visibility (diagnostic) | Where are we gaining or losing ground? | Share of visibility across keyword groups, not single positions |
| 6. What we did and what we learned | Did last month's work do anything? | Change log, tests, annotations |
| 7. Next month | What are we doing and why? | Three priorities, tied back to section 1 |
| 8. Data health | Can we trust any of the above? | Tracking changes, consent rates, known gaps |
A few notes on making it work:
- Agree the section 1 number before any work starts. Jeff's one KPI rule. If it's leads, agree what counts as a good one with sales.
- Group pages and keywords. "Service pages up 18%" beats a table of 400 positions.
- Keep the AI prompt set stable. Use the same twenty or so questions every month, run them logged out, and note whether you're cited, named or missing. It's a sample, so treat it like one.
- Keep rankings, but move them to the back. They're still handy for spotting a broken template or a lost page. Clients who love them can have an appendix.
- Put a sentence under every chart. If you can't say what it means for the business, cut it.
Automate the numbers, talk through the story
Jon Henshaw started Raven Tools because his agency spent "the last week of a month" assembling client reports. He wanted a "magical button" that would pull it all together. He tells that story here.
The button exists now. When I interviewed Paul Shapiro in 2019, he called building a report "one of the most automatable things", but that wasn't the end of it: "we still take the time, we sit down with the client and analyse the output of our automated reports... We don't just hand them a dashboard." (His written interview has the rest.)
The machine can build the report. A person still has to explain it to someone who cares whether the phone rings, which is understanding the context, and the best inputs still come from talking to the people behind the screen, not from a rank tracker.
Rand's version from 2014 still does the job. Stop promising number one and start asking how to make twice as much from search next year.