Most AEO platforms report share of voice: the percentage of AI answers in which your brand is mentioned. Recommendation Share™ is a different number. It weights each mention by the position the engine gave it, so the brand named first gets full credit and a brand named eighth gets little. The two metrics can move in opposite directions. This article explains the method, works through an example, and shows why the position-weighted number is the one that tracks revenue.
What share of voice measures
Share of voice in AI answers is a count. Run a prompt through an engine, note every brand named, and divide your mentions by the total. A brand that appears in nine out of ten answers has 90% share of voice, whether it was named first or last. It is a useful baseline. It answers "are we in the conversation?"
What Recommendation Share™ is
Recommendation Share™ is BrandRank.AI's primary visibility metric. For every prompt run, the engine's answer is read as a ranked list of up to ten brands. The brand in position 1 scores 10. Position 2 scores 9, down to position 10, which scores 1. A brand that is not named scores 0. Scores are averaged across every run of every prompt in the tracked set. The result is a 0 to 10 score, shown as a percentage: a Recommendation Share™ of 8.5 means 85%.
Because each brand is scored independently, the scores across competitors do not sum to 10 or to 100%. The platform also shows the rank distribution per category: how often a brand lands at each position across all answers.
A worked example
Suppose two dog food brands are tracked across ten buying prompts on one engine.
| Brand A | Brand B | |
|---|---|---|
| Prompts where the brand is named | 10 of 10 | 6 of 10 |
| Share of voice (mentions) | 100% | 60% |
| Typical position when named | 7th | 1st |
| Points per prompt (named) | 4 | 10 |
| Points per prompt (not named) | 0 | 0 |
| Recommendation Share™ | 40 ÷ 10 = 4.0 (40%) | 60 ÷ 10 = 6.0 (60%) |
Brand A wins share of voice. Brand B wins Recommendation Share™. When a shopper asks which dog food vets trust, Brand B is the one the engine names first in the answers where it appears. Brand A is on the list every time and chosen almost never.
Why the two numbers diverge
- Engines often name eight or ten brands in a "best of" answer and then recommend one or two. Mentions reward being on the list. Recommendation Share™ rewards being the pick.
- A brand with strong third-party citations tends to be named earlier. A brand with weak proof gets named late or as an afterthought.
- Prompt choice matters. Share of voice on generic prompts can look healthy while the revenue prompts ("which one should I buy for a puppy with a sensitive stomach") go to a competitor.
How BrandRank.AI moves the number
Recommendation Share™ is tracked on a repeating scoring cycle and moved deliberately through five steps: win the strategically important questions, strengthen the proof behind priority claims, close content gaps, build third-party authority, then publish, verify, and measure again. Changes are kept only if the score moves. A CPG launch gained 300% in AI search visibility after this process raised content readiness by 24%. A global beverage brand gained 40% in Gemini and 10% in ChatGPT.
How to read your score
Customers are told that a score below 7 or 8 signals trouble on a must-win prompt. The rank distribution shows whether the problem is absence (scoring 0 often) or weak position (named, but late). Absence is usually a content readiness problem. Weak position is usually a proof and third-party authority problem. The fix is different for each.
Get your Recommendation Share™ across ChatGPT, Gemini, Perplexity, Grok, Claude, Meta AI, and DeepSeek.
Frequently asked questions
What is Recommendation Share™?
Recommendation Share™ is BrandRank.AI's position-weighted measure of how often a brand is the one an AI answer engine recommends. The brand named first in an answer scores 10, the tenth scores 1, an absent brand scores 0, and the scores are averaged across all prompts and runs. A score of 8.5 is reported as 85%.
How is Recommendation Share™ different from share of voice?
Share of voice counts mentions regardless of position. Recommendation Share™ weights each mention by position. A brand mentioned in every answer but never first can have high share of voice and low Recommendation Share™.
Do Recommendation Share™ scores across competitors add up to 100%?
No. Each brand is scored independently against the same prompts, so the scores do not sum to 10 or to 100%.
What is a good Recommendation Share™ score?
BrandRank.AI tells customers that a score below 7 or 8 on a must-win prompt signals trouble. The right target depends on category and competitor density.
Which engines is Recommendation Share™ measured on?
ChatGPT, Gemini, Perplexity, Grok, Claude, Meta AI, and DeepSeek, reported per engine and as an overall score.
Is Recommendation Share™ a trademark?
Yes. Recommendation Share™ is a trademark of BrandRank.AI.
Can I improve Recommendation Share™ without changing share of voice?
Yes. Strengthening the proof behind priority claims and adding third-party citations moves a brand from a late mention to an early one without changing how often it is mentioned.
