100 Questions, Twentieth Category: SEO Tools, Where Ahrefs Is the Pick in 55 Answers and Reddit Ranks for 90 Questions but Is Never Cited

Two suites share the verdicts

Ahrefs is named in 60 of 100 answers, recommended in 59 and picked as the answer's own choice in 55. Semrush is named in 56, recommended in 55 and picked in 46. They are picked together in 37 answers, and at least one of them is picked in 64. SE Ranking is third, named in 30 and picked in 28. On the 20 direct advice questions, every answer picked at least one dictionary brand: Ahrefs in 14, Semrush in 12, both in 12. The four head-to-heads that name both picked both, and three of them draw the same line: Semrush for a broader marketing or agency toolkit, Ahrefs for deeper SEO research, content and links. The fourth, about tracking AI Overviews, leans Semrush. Across all 30 head-to-heads, 29 picked every brand named; the other was a one-sentence answer that never made the comparison.

Moz is described more than it is chosen

Among brands named in 15 or more answers, Moz has the widest gap between recommended and picked: named in 20 answers, recommended in 19, picked in 8. Five of those 8 picks came in head-to-heads that name Moz. In the 15 answers to questions that do not name it, Moz was picked 3 times; elsewhere it was the beginner-friendly or simpler option in a list of fits. Semrush has the next gap, 9 answers, and in three of them it is recommended only to buyers who already use it. Six answers picked no dictionary brand at all. Three never answered the question, one described eight popular tools without a verdict, one listed eight SE Ranking alternatives without saying who each suits, and one picked free crawlers outside the dictionary.

Specialist questions go to specialists

In the 15 rank tracking questions, AccuRanker and SE Ranking were each picked in 9, ahead of Ahrefs, Semrush, Nightwatch and Wincher (5 each). BrightLocal was picked in all three local SEO questions and Local Falcon in two. Enterprise questions went to BrightEdge and Conductor (5 of 6 each) and seoClarity (3). WordPress and Shopify questions went to Yoast SEO (4 of 5), Rank Math and All in One SEO; crawling questions to Sitebulb (4 of 5) and Screaming Frog. Content optimization went to Surfer and Clearscope (4 each), Frase and MarketMuse (3 each). Across these 39 lane questions, Ahrefs or Semrush was picked in 13, and in each of them next to a specialist pick.

AI visibility is the exception

In the five AI visibility questions that name no brand, Ahrefs was picked in all five. The dedicated tracker Otterly.AI was picked in three; Semrush, SE Ranking, Profound and Peec AI in two each. Asked which tool shows whether ChatGPT mentions a brand, the only dictionary pick was Ahrefs's free AI visibility checker, with a tracker outside the dictionary, Siftly, suggested for ongoing monitoring. The question asking for a Profound alternative for smaller brands was read as a question about wording, and the answer named no tool.

Reddit ranks, and is never cited

Google's top ten for these 100 questions is 88 percent third-party pages, and the most common site in it is Reddit: reddit.com is in the top ten for 90 of the questions and in first place for 35. Quora is in the top ten for 37 and YouTube for 24. ChatGPT cited none of them. Reddit and Wikipedia have zero citations, for the twentieth teardown running. Of the 309 picked brand mentions, 258 were for brands whose own website did not rank in Google's top ten for the question, 83 percent. Ahrefs's own site ranked for 5 of the 55 questions where it was picked, Semrush's for 6 of 46.

The vendors write the reading list

Across 100 answers there were 306 citation events to 130 domains. Vendors' own sites took 170 of them, 56 percent, and 49 answers cited nothing but vendor pages. ahrefs.com is cited in 42 answers, 31 of them to questions that never mention Ahrefs; semrush.com is cited in 38, 27 of them to questions that never mention Semrush. The two domains account for 80 of the 306 citation events, yet ahrefs.com ranks in Google's top ten for only 5 of the 100 questions and semrush.com for 6. No independent site comes close: the most-cited third party, techcognate.com, appears in 5 answers.

Forty-one of the 306 citation events involved a domain that also sat in Google's top ten for that query, 13 percent, the lowest overlap in the series so far.

Your own pages are read for other people's questions. ahrefs.com and semrush.com were cited in 31 and 27 answers to questions that do not mention them, while ranking in Google's top ten for 5 and 6 of the 100 questions. In this category the model reads vendor guides and comparison pages far more often than Google ranks them.

A lane beats a general pitch. Two suites share the general questions. The other vendors that were picked were picked for a stated job: rank tracking, local, enterprise, plugins, crawling, content. Where a suite was picked on those questions, it was picked next to a specialist. Say plainly which job you do and for whom.

Being described is not being chosen. Moz was named in 20 answers and picked in 8, mostly offered as the simpler or beginner option in a list. Whether the answer picks you is a different measure from whether it names you.

Free tools are part of the verdict. Google Search Console was picked in 22 answers, never to a question that named it, and in 9 of the 20 direct advice questions, each time alongside at least one tool that is not Google's. A paid product is weighed against a free starting point, not just against other paid products.

One run per query means answer variance is unmeasured; Benchmark 01 runs each query three times across three surfaces. The teardowns are collected on different days, so differences between categories mix the category with the date. The brand dictionary covers the general SEO tools market and deliberately excludes vertical tools, so their appearances are described in prose and not counted. The readers are AI models, not people: they follow a written protocol and agree with each other closely, but a shared blind spot would not show up as disagreement. The line between picked and recommended is a judgment, documented in the protocol with examples. Sentiment is not coded. No vendor-tool cross-check was read for this category. Google's control counts a brand as ranking only when its own domain is in the top ten; a listicle that features the brand does not count. One run per question, one day, United States, English, logged out. Answers vary between runs, so these are frequencies for this sample. Three answers did not answer their question. Shopify SEO apps, several free crawlers and some newer AI visibility trackers are not in the dictionary, so answers that pick them undercount those picks. Four Google tools share google.com, so their citations are counted together.

Every number above can be recomputed from these files. CC BY 4.0: use them, cite the page.

• queries.csv: the 100 queries with intent labels.
• brands.csv: the 54-brand dictionary with aliases and canonical domains.
• mentions.csv: 413 coded brand mentions with position, type, a 0/1 picked column and whether the brand's domain was in Google's top ten.
• citations.csv: 306 citation events with domain class and Google overlap.
• observations.csv: one row per query with brand, recommendation, citation and control counts.
• coder-notes.md: coding method and agreement, the judgment calls the AI readers and the third reader made in this category, and dictionary notes
• codebook.md: the rulebook, with dated amendments through v2.0.
• reader-protocol.md: the written protocol both readers coded against (codebook v2.0).
• picked_stats.json: the picked-level figures.

The earlier teardowns: Teardown 01, project management, Teardown 02, crm, Teardown 03, email marketing, Teardown 04, help desk, Teardown 05, accounting, Teardown 06, payment processing, Teardown 07, payroll, Teardown 08, hr software, Teardown 09, applicant tracking, Teardown 10, password managers, Teardown 11, endpoint security, Teardown 12, business intelligence, Teardown 13, data warehouse and etl, Teardown 14, ecommerce platforms, Teardown 15, website builders, Teardown 16, scheduling, Teardown 17, video conferencing, Teardown 18, e-signature, Teardown 19, marketing automation.

In SEO tools the model reads the two most-picked vendors' own websites far more often than Google ranks them, and skips the forum Google ranks most. In password managers it reads the vendors' sites; in applicant tracking, a comparison library that does not rank. Finding out what the model is reading for your category is the first thing a Category Salience Brief does, with a query set you approve first.