Brand recommendations from AI platforms are decided at the prompt level: the exact buying question a customer types determines whether ChatGPT, Gemini, or Perplexity names you or a rival. To find the prompts that surface your competitors, run the real questions your buyers ask, log every brand each engine returns, and study the language and sources behind each mention. That pattern is what you fix.
At Grid13, we build automated blog content that improves both SEO and AI-search visibility, and we run this exact audit for clients. The frustrating part for most marketing leads is that their product is fine — it's the prompts that betray them. This is a solvable, repeatable problem, and below is the method we use, with the actual tools and prompt types involved.
Why competitor recommendations are a prompt problem, not a product problem
When an AI engine recommends a competitor over you, the instinct is to blame the product. Usually the real issue is narrower: for a specific phrasing of a specific question, the model has more structured, more cited evidence pointing at the other brand. Change the prompt slightly and the answer can flip entirely.
The stakes are rising fast. Adobe's Q1 2026 analytics showed 393% year-over-year growth in AI referral traffic to U.S. retail sites, and in B2B the shift is just as sharp — G2's late-2025 buyer research found that 51% of B2B software buyers now start research with an AI chatbot more often than with Google. If a chatbot is the first thing your buyer opens, the brand it names first has a real head start.
How do you audit the prompts that trigger rival brand recommendations?
Here is the method, built to be repeated in a single afternoon. You do not need enterprise software to start — you need discipline and a spreadsheet.
- List your buyers' real questions. Not keywords — full sentences. Cover the five buying moments: best provider ("what's the best [category] tool for a small B2B team"), head-to-head comparison ("[your brand] vs [rival]"), budget ("cheapest [category] that still does X"), local or niche ("[category] for [industry]"), and problem-first ("how do I fix [pain point]"). Aim for 20–30 prompts.
- Run each prompt across the engines that matter. Use ChatGPT, Google Gemini, and Perplexity at minimum. Run each prompt in a fresh session so past chat history doesn't skew the answer, and run it two or three times — answers vary.
- Log every brand named, in order. Record which brands appear, their position, and whether you appear at all.
- Capture the sources. Perplexity and Gemini cite links directly; in ChatGPT, ask "what sources support that recommendation?" Note the domains — review sites, comparison articles, the competitor's own pages.
- Tag the attribute. For each competitor mention, mark why the model chose it: price, reputation, a named feature, specialization, or support quality.
What not to do: don't run one prompt once, see your brand missing, and conclude you're "invisible to AI." A single run is noise. And don't audit only the flattering "best provider" prompts — the comparison and budget prompts are where competitors quietly win, and they're the ones buyers actually type.
What the patterns in your competitor data actually tell you
Once you've logged 20–30 prompts across three engines, the noise resolves into patterns. This is the part that turns an audit into a strategy.
If a rival keeps appearing on price prompts, the model has found pricing evidence for them and not for you — often because your pricing lives behind a "contact sales" wall while theirs is on a public, structured page. If they own the comparison prompts, there's likely a well-cited third-party comparison article the engine trusts. If they win specialization prompts ("[category] for [industry]"), they've published industry-specific content and you haven't.
The scale of the opportunity is bigger than most teams assume. A 2025 Botify study found that 62% of brands are invisible on at least one major AI platform. Invisibility on even one engine, on the prompts that convert, is a gap a competitor is already filling.
Which sources should you focus on first?
Follow the citations. AI engines lean on third-party evidence far more than on your marketing copy. If your rivals are recommended off the back of a review roundup or an independent comparison, that's your priority list: earn a place in those same sources through digital PR, accurate directory listings, and genuinely useful content. When buyers act on these answers — and 39% of consumers have purchased an AI-recommended product in the past six months — the citation you're missing is a sale you're missing.
How to turn brand recommendations data into positioning that wins
The audit hands you a ranked to-do list. Work it in this order:
- Fix the evidence gap the engines can see. If price prompts favor rivals, publish clear, structured pricing information. If comparison prompts do, publish an honest, specific comparison page on your own domain — one an engine can quote.
- Own the attribute you actually lead on. Pick the one thing you genuinely do better and make it unmistakable across your site, your content, and any third-party listing. Vague "we're the best" language gives an AI nothing to cite.
- Build content around the losing prompts. For every prompt where a competitor appears and you don't, there's a page you haven't written. Answer that exact question directly, with the evidence a model needs to name you.
- Feed the sources AI trusts. Reviews, directory profiles, and earned media do more for AI brand recommendations than another homepage rewrite.
This is precisely what our automated blog content system at Grid13 is designed to do: publish structured, well-sourced articles that answer buyer prompts directly, so search engines and AI models have a clear reason to cite you instead of the competition.
Why speed matters more than it used to
AI-driven discovery is compounding, not creeping. Statista data shows AI-powered search grew 1,200% in 2024, and ChatGPT referral visits grew 1,079% in 12 months according to Conductor research. Every month a competitor owns a buying prompt, the model reinforces that association. Fixing the gap early costs less than clawing back a position the AI has already learned.
Frequently Asked Questions
How often should I re-run a brand recommendations audit?
Quarterly is a sensible baseline, and monthly if you're in a fast-moving category. AI models retrain and re-crawl, so a prompt that names you today can drop you next quarter. Re-running the same prompt set also shows whether your fixes are actually moving the needle.
Do AI platforms recommend different brands for the same question?
Yes, and that's why you audit across engines. ChatGPT, Gemini, and Perplexity draw on different sources and weight them differently, so the same prompt can return a different brand order on each. A competitor may dominate one platform while you lead on another.
Why does a competitor appear when my product is clearly better?
Because the model recommends on available, citable evidence, not on real quality. If a rival has more structured pricing, stronger third-party reviews, or content that answers the exact prompt, the AI names them. It's an evidence gap, not a product gap.
Can I influence which brand an AI recommends?
You can't program the output, but you can change the inputs it reads: publish clear content answering buyer prompts, keep directory and review profiles accurate, and earn mentions on the third-party sources AI cites. Do that and your odds of being named rise measurably.
Is consumer trust in AI recommendations high enough to matter?
It is. A BCG survey of 9,000 consumers across nine countries found that over 60% express high trust in generative AI shopping results, and in B2B, 67% of buyers consult AI before contacting sales. When buyers trust and act on these answers, being recommended is a direct revenue lever.
Want the prompts that decide your brand recommendations working in your favor? See how Grid13's automated content platform builds AI-visible articles that get you cited instead of your rivals.