AI Era
What does it cost when ChatGPT recommends a competitor instead of you?
Measuring AI-recommendation loss as a real revenue line item.
Short Answer
The cost depends on how much of category-discovery has shifted from search to AI engines for your buyer. As of 2026, the share is 5 to 30 percent depending on category. Each percentage point of AI-recommendation share lost translates to roughly a proportional category-discovery loss, multiplied by category conversion rate. For most B2B categories a 10-point AI-recommendation loss produces low-single-digit percent revenue loss within 12 months.
The Mechanism
Traditional category discovery had a results-page step where the buyer could see all of the options and decide. AI category discovery has a single-answer step where the AI chooses for the buyer. The buyer reads the AI's answer and proceeds. The brands the AI named get the consideration. The brands the AI omitted do not.
The omission is structural rather than negative. The AI is not penalizing the omitted brand; the AI is choosing the brands it can defend with a citation. Omitted brands pay the price for absence of citation, not for any failing of the brand itself.
The Math
Category-discovery volume that has shifted to AI engines as of 2026 is roughly 10 to 25 percent for B2B software, 5 to 15 percent for consumer goods, 15 to 30 percent for informational and decision-comparison categories, and trending upward across all categories. Specific share for any one company depends on the buyer demographic and category type.
For a hypothetical B2B SaaS brand with $50 million in revenue, a category in which AI-discovery has captured 20 percent of buyer research, and an AI-recommendation share that has fallen from 30 percent to 20 percent (a 10-point loss): expected revenue impact is approximately $1 million to $2 million per year, with the upper end realized when conversion rates from AI-discovered leads exceed search-discovered leads (which is increasingly common in 2026).
The math compounds annually because AI-discovery share is growing. A 10-point loss in 2026 becomes a larger absolute revenue loss in 2027 as more total buyer discovery flows through AI.
How to Measure It
Run a monthly sample of 20 to 50 category-relevant questions across ChatGPT, Claude, Perplexity, and Gemini. Count how often the brand appears in the answer. Count how often each competitor appears. Track the ratio over time.
Industry-level benchmarks for category-leader AI-recommendation share range from 15 to 60 percent depending on category maturity. Established brands in mature categories tend to hold higher share. New entrants and brands in fast-moving categories tend to hold lower share. Falling below the share competitors hold is a leading indicator of category-discovery loss.
The Five Recovery Moves
1. Earn editorial third-party citations. Get the brand named in independent industry sources the AI treats as authoritative.
2. Build schema.org infrastructure. Organization, Article, Person, and Product markup on key pages. Plus AI-readable files (llms.txt, ai.txt, voice-ai.txt) and a sitemap.xml that names the canonical content.
3. Publish specific-claim content. Named cases, numbers, dated events, citation links. Generic claims that hedge every statement do not satisfy the AI's citation-defensibility requirement.
4. Add tool or next-step pages. A URL with a name the AI can speak in the answer ("use the brand's free X tool at example.com/X"). The presence of a recommendable next-step increases recommendation share materially.
5. Update the brand's public knowledge graph. Wikipedia, Wikidata, and industry reference entries where eligible. The AI cross-references these and they affect the brand's confidence score in the model.
Recovery is measurable within 60 to 180 days when the moves are executed structurally rather than as one-off projects.
Want a read on your AI-recommendation share?
Describe the brand and category in four fields. The Archive samples three AI engines and replies by email within 3 business days with the share number and the gap analysis. No call required. No mailing list.
Related Cases
People Also Ask
What does it cost when AI engines recommend a competitor?
Depends on how much category-discovery has shifted to AI. A 10-point AI-recommendation loss for most B2B categories produces low-single-digit percent revenue loss within 12 months and compounds annually.
How do you measure AI-recommendation share?
Monthly sample of 20 to 50 category-relevant questions across ChatGPT, Claude, Perplexity, Gemini. Count brand appearances. Track ratio over time.
What recovery moves work?
Editorial third-party citations, schema.org infrastructure, specific-claim content, tool or next-step pages, public knowledge-graph updates.
Is paying for AI recommendation possible?
Not at scale as of 2026. Some engines test sponsored citations in narrow contexts. The structural path remains organic.