Research · claim check · August 2026
A tier list ranking 2026 SEO tactics has been doing the rounds: backlinks and exact-match domains at S, programmatic SEO and AI content at A, press releases at B, schema at C, keyword stuffing at the bottom. Tier lists compress real experience into a shareable card, and this one gets more right than most. We measure 3,091 real businesses against the AI answer engines continuously, so instead of arguing, we checked each tier against what separates the businesses AI names from the businesses it never mentions. The numbers on this page recompute from the live index.
Basis: our 3,091-business corpus, one row per domain, comparing the 713 businesses AI names consistently against the 1,108 it never names. A tactic "agrees" when the measured gap backs its tier.
| Tactic | Tier claimed | Verdict | What we measure |
|---|---|---|---|
| Backlinks & independent mentions | S | AGREES | The largest gap in our data: businesses AI names average 81/100 on off-page brand signals against 37/100 for the invisible, a 44-point gap. Being talked about elsewhere is the strongest measured predictor of being named. |
| Exact-match domains | S | UNMEASURED | We hold no measurement of EMD lift, and nobody sharing the tier list has published one either. What we do measure cuts the other way: generic-word names create entity collisions our scanner flags, and a brand the engines cannot anchor loses mentions to its namesakes. S-tier is a claim in want of data. |
| Programmatic SEO | A | AGREES | Engines retrieve pages shaped like the questions buyers type, and use-case pages at scale match that shape. Our Cal.com claim check found their 333 programmatic use-case pages were the largest content play on the site, ten times their comparison-page footprint. |
| Video SEO | A | UNMEASURED | Outside our instrument. We measure the answer engines, and video surfaces rank by their own machinery. |
| AI-written content | A | UNMEASURED | We measure outcomes, and engines do not disclose authorship detection. Nothing in our corpus separates naming by who wrote the page. What the named businesses share is independent corroboration, whoever wrote their copy. |
| Press releases | B | PARTLY | Half right. A real journalist covering you is an independent mention, and those drive the 44-point gap above. Syndicated wire copy that nobody cites is the weak form: engines cite the coverage, never the wire. |
| Schema markup | C | AGREES | C is about right. The schema gap between named and invisible businesses is 10 points (75 vs 65): schema helps engines read you correctly and does little to make them recommend you. Necessary floor, weak lever. |
| Keyword stuffing | bottom | AGREES | Agreed, and it costs more than rankings now: one measured business's keyword-stuffed homepage title corrupted how AI reads its brand name itself, so every mention-counting system, ours included, struggles to attribute it. Stuffing now damages your entity. |
Our basis: 3,091 distinct businesses (best clean measurement per domain), each measured across the answer engines on real buyer questions, multi-sampled with majority voting, scored on six readiness pillars. "Named" means visibility of 60 or more; "invisible" under 25. The comparison shows association across cohorts, and no tier list, ours included, is a per-tactic causal experiment: a business that earns backlinks usually earns reviews and press at the same time. What the data does support is the ordering: off-page corroboration first, question-shaped pages second, structured data as a floor, and nothing on the list rescues a business nobody mentions. Full breakdown on the research page; method open-source at github.com/techhorizonlabs/thl-open.
The free scan measures you the same way: who AI names for your buyers' questions, who it names instead, and which of these signals you are missing. 60 seconds, no signup.
Run your free scan →