CASE STUDY · STRATEGY Aug 14, 2026 · 6 min read

How a 200-person company out-cites a $160bn giant in AI search

MK Mara Kovač Editor-in-Chief · 6:00 AM ET 𝕏 in ✉

Editorial illustration: one small solid shape outranking a receding row of much larger pale shapes

The most repeated hope in AI search is that it flattens the field, that a small company with better answers can out-cite an incumbent with a bigger balance sheet. It is a nice story. It is also, in specific documented cases, true. The useful question is not whether it happens but what the winners actually did, because the pattern across the published case studies is unusually consistent.

⚡TL;DRDocumented cases show a 200-person company holding its own against a $160bn competitor in AI answers, and smaller commerce brands out-citing incumbents many times their size. The common factors are not budget: category-level comparison content that names competitors honestly, a genuine refresh cadence, and structural clarity. The catch, also documented, is that being cited is not the same as being chosen.

The cases

The clearest published example is Search Engine Land's teardown of how a roughly 200-person software company competes in AI search against a competitor worth about $160 billion. Not by outspending, but by owning the category-level questions buyers actually ask. A separate Q1 2026 analysis of commerce brands found smaller merchants beating enterprise incumbents by twenty to forty points of AI citation share in head-to-head category queries. And a study of three startups out-citing their category leaders identified the same handful of shared behaviours.

This is what the 50,000-brand study predicts, incidentally. When domain authority correlates with topic ownership about as well as a coin flip, and 85% of topics have no consistent winner, size stops being destiny. The data said the door was open. These cases are companies walking through it.

THE PATTERN ACROSS DOCUMENTED WINS01Comparison pages that name rivalsThe highest-leverage asset is an honest "X vs Y" page where each option wins on specific use cases, including the ones where you lose. Comparison content carries the single highest citation rate of any format on ChatGPT.02Category-level answers, not brand-level onesEngines answer "best tool for X", not "tell me about your company". The winners publish the category page and put themselves in it as one credible option among several.03A refresh cadence that is realMonthly updates with substantive change, not date-stamp edits. Recency is a measured citation factor and it is one incumbents with heavy publishing processes are structurally slow at.04Structural clarity over authorityComparison tables with real pricing, explicit criteria, extractable facts. This is the "linkability beats authority" finding in practice.

The catch nobody mentions

Getting cited is not the same as getting picked. Trakkr's research on comparison pages makes the uncomfortable point that a page can be cited as a source while the answer built from it recommends someone else. Your honest comparison page, the one that fairly describes a competitor's strengths, can end up as the citation underneath a recommendation for that competitor.

"Your honest comparison page can end up as the citation underneath a recommendation for your competitor. Cited is not the same as chosen."

That is not an argument for dishonest comparison pages, which engines and readers both punish. It is an argument for measuring the right thing: not just whether you appear in the citation list, but whether the answer text recommends you. Those are two different metrics and most AI visibility tooling reports only the first.

What to do this week

Write the comparison page you have been avoiding, the one that names your two strongest competitors and says plainly what each is better at. Put real pricing in a table. Then track two numbers on the queries that matter: how often you are cited, and how often you are the recommendation. If the first is climbing and the second is flat, your page is doing someone else's selling, and the fix is sharpening the specific use cases where you genuinely win.

KEY TAKEAWAYS01A 200-person company competing with a $160bn giant in AI answers is documented, not hypothetical.02Smaller commerce brands beat enterprise incumbents by 20 to 40 points of citation share in category head-to-heads.03The shared pattern: honest comparison pages, category-level framing, real refresh cadence, extractable structure.04Cited is not chosen. A fair comparison page can support a rival's recommendation.05Measure citation share and recommendation share separately. Most tools only report the first.