Show Your Work is a free monthly email from Christopher Jan Benitez. Each issue takes one claim going around in SEO or AI search, tests it against live data from ChatGPT, Perplexity, Gemini and Google AI Overviews, and publishes the method, the cost in dollars and hours, the result and its limits. When a claim fails, you read that. When one of my own earlier numbers was wrong, you read the correction.
No sales sequence. One email a month. The first one brings the free Claude SEO skill pack. Unsubscribe in one click.
What do you get in each issue?
Each issue has one deep test, two or three shorter findings, a corrections line, one artifact you can reuse, and a vote on the next test. The deep test runs 700 to 1,000 words and follows the same order every month:
- The claim, and the hypothesis exactly as I wrote it before pulling any data
- The sample, the engines, the dates and the source, in plain words
- The result, then what it cannot tell you
- What it cost in DataForSEO credits and in hours
- One thing to try that month, with a time estimate
- The artifact behind the result: a prompt panel, a classifier map, a skill file or a sheet
- The vote: three candidate claims, and you click the one you want tested next
What has been tested so far?
Five studies since July 2026, built on 1,648 AI citations and 105 answers, plus a statistics hub of more than 60 checked figures. The newsletter is the backstage of this work, so start with the studies if you want to see the standard:
- July 2026. Does good SEO get you recommended by AI? Four email marketing tools, two scoreboards. They nearly inverted.
- July 2026. Where does AI get its recommendations? 79% of 1,648 citations were third-party sites. 21% were the brand’s own pages.
- August 2026. Does AI recommend the brand that ranks #1 on Google? monday.com, Casper and NordVPN each lost their own category.
- August 2026, corrected 26 October. Which sources does each engine cite? AI Overviews lean on video, Perplexity on review sites, Gemini on brand pages. The first version dropped a row from the table. The correction is dated on the page.
- October 2026. Who does AI name as the best SEO expert in the Philippines? 105 answers, two different lists, depending on whether you ask a chatbot or Google.
- Running. The AI search statistics hub, where every figure was re-checked against its source on 26 October 2026.
Who is this for?
Show Your Work is for people who run SEO with AI tools and want evidence before they change a process: in-house SEOs, agency staff, consultants, and marketers at ecommerce and B2B companies who are asked why the brand does not show up in ChatGPT.
It is not writing advice, and it is not a tool roundup. If you want tips on a freelance writing career, this is the wrong email, and I would rather you skipped it than unsubscribed in month two.
How are the tests run?
Every test runs on DataForSEO data pulled through Claude Code, with the hypothesis written down before the first API call. The whole October 2026 research month cost $34.42 in credits, and that receipt is published.
The prompt panels call ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, then log every brand named and every source cited. The classifier maps that sort those citations are the artifacts subscribers get. Each study page states its sample size and what it cannot show. Corrections get a dated note on the original page and a line in the next issue.
Why trust a newsletter that admits it was wrong?
Because the correction is on the record. On 26 October 2026 I recounted all 1,648 citations from the August source study after finding that the first classifier had left 16 to 34 percent of citations unlabeled and dropped that row from the published table.
The corrected figures replaced the old ones on the study page, the statistics hub and the two guides that quoted them. The August social posts still carry the old number, because those cannot be edited. That is the standard every issue is held to, and it is the reason the newsletter has this name.
Who decides what gets tested next?
Subscribers do. Every issue ends with three candidate claims, and the one with the most clicks becomes the next month’s deep test. The count is published in the issue that follows.
You can also reply to any issue with the SEO or AI claim you keep hearing and want checked. Replies go into the test backlog with your question quoted, and the backlog is where the vote options come from.
Get the next test
Monthly. Free. The first email brings the Claude SEO skill pack. Issue 1 goes out in December 2026.
Written by Christopher Jan Benitez, AI SEO specialist and Director of SEO & Content Strategy at AlchemyLeads. Past issues are archived at subscribe.christopherjanb.com. Links to DataForSEO on this site may be affiliate links; I may earn a commission on a sale, at no extra cost to you.