ChatGPT Doesn't Know You Exist. Fix That.

(Alternates: "Is AI Recommending You or Your Competitor?" / "Zero Clicks, One Answer: Win the AI Recommendation")

Your buyer asks ChatGPT, "What's the best platform for X?" and gets one answer, not ten links. Either your brand is in it or it isn't, and most teams can't say which. They know their Google rank but not their AI visibility.

This guide shows you how to measure where you stand, work out why you're missing, and fix it.

First, understand what you're measuring

Three terms get mixed up constantly:

• Mention: your brand name appears in the answer.
• Citation: your URL appears as a source.
• Recommendation: you are named as a good fit, ideally first.

You can be mentioned without being cited, and cited without being recommended. Track all three, because each one points to a different fix.

Also remember that AI answers are probabilistic. Ask the same question five times and you may get three different brand lists. One screenshot proves nothing, so you need repeated sampling.

Part 1: Track your brand mentions

Step 1: Build a 40-prompt set

Pull real phrasing from sales calls, support tickets and Google's "People also ask." Split the prompts into four buckets of about ten each:

Intent	Template
Category discovery	"Best [category] for [segment/industry]"
Comparison	"[You] vs [Competitor]" and "alternatives to [Competitor]"
Problem-led	"How do I [pain point] without [constraint]"
Brand-direct	"Is [You] good for [use case]?" and "What are the downsides of [You]?"

Problem-led prompts matter most. They are where buyers meet AI before they know any brand names.

Step 2: Run, log, repeat

Run every prompt on ChatGPT, Perplexity and Gemini, plus Claude and Google AI Overviews if your buyers use them. Do it three times per prompt per engine, weekly, in a clean session so personalization doesn't skew results.

Log these columns in a sheet:

1. Prompt and engine
2. Mentioned? (Y/N)
3. Position (1st, 2nd, 3rd or later)
4. Cited? (Y/N) and which URL
5. Competitors named
6. Descriptor used ("affordable," "enterprise-grade," "limited")
7. Sources cited for competitors

Column 7 is the most valuable one, and almost nobody records it. It is your target list.

Step 3: Score it with four numbers

• Mention rate: prompts where you appear, divided by total prompts
• Citation rate: answers linking to your domain
• Share of voice: your mentions divided by all brand mentions in the category
• Sentiment mix: how you are framed

Set your baseline in week one. Improvement against your own baseline matters more than any absolute score.

Step 4: Add tools only when the sheet hurts

A spreadsheet is enough for the first month. After that, platforms like Profound, Peec AI, Otterly.AI, and the AI visibility modules in Semrush and Ahrefs can automate the sampling. Features and pricing change fast, so verify before buying. Insist on repeated sampling and visible cited sources, not just a single score.

Step 5: Catch the downstream signal

In GA4, filter referrals from chatgpt.com, perplexity.ai and gemini.google.com. This undercounts, because many AI-influenced buyers arrive later as direct or branded search. Add "How did you hear about us?" to your demo form, with a free-text field. You will see "ChatGPT" appear sooner than you expect.

Part 2: Diagnose why you're missing

Before you fix anything, find which of these four situations you are in. The tracking data will tell you.

1. Not mentioned, competitors are: an authority gap. Models haven't seen enough credible third-party evidence about you.
2. Mentioned, never cited: a content gap. Your pages aren't the easiest source to extract from.
3. Mentioned, framed badly: a positioning gap. The internet describes you differently than you describe yourself.
4. Mentioned on one engine only: a crawl or retrieval gap. Check access and freshness.

Part 3: Get cited by ChatGPT, Perplexity and Gemini

1. Write answer-first pages

Put the direct answer in the first two sentences under every heading, then add nuance. Compare:

Weak: "Choosing a telephony solution is a journey every growing business takes..."

Strong: "Cloud telephony is a phone system hosted on the internet instead of on-site hardware. Teams use it to route calls, record conversations and connect voice agents to their CRM."

Use question-style subheadings that match how people prompt. Add a short FAQ block to every money page.

2. Publish something only you can say

Engines favor sources that add something new: original benchmarks, survey data, named frameworks and first-hand case studies with real numbers. If you are the origin of a statistic, everyone else cites you, and the models follow that trail. One genuine data point beats ten "ultimate guides."

3. Win where the engines already look

Sort column 7 of your sheet by frequency. The same domains will keep showing up: G2, Capterra, Reddit, YouTube, Wikipedia, industry publications and a handful of niche blogs. Build a short list of 15 to 20 target sources and go after them:

• Collect genuine customer reviews on the review sites that appear most
• Pitch expert quotes and contributed pieces to cited publications
• Answer questions in relevant communities as a practitioner, not a promoter
• Get included in "best of" and comparison roundups

This is the highest-leverage work in the whole playbook, because it addresses the authority gap directly.

4. Make your entity unmistakable

Use one positioning sentence everywhere: site, LinkedIn, directories, press kits. Keep a clear About page, and add Organization, Product and FAQ schema. Any system should be able to answer "what does this company do, and for whom?" without guessing.

5. Remove technical blockers

Check that robots.txt isn't blocking GPTBot, PerplexityBot or Google's crawlers unless you intend it. Keep key pages server-rendered and fast, since content hidden behind heavy JavaScript can be missed. Show last-updated dates and refresh important pages quarterly, because Perplexity in particular rewards recency.

6. Own the comparison queries

Publish honest "X vs Y" and "alternatives to Z" pages, including where a competitor is genuinely stronger. Balanced pages get cited. Sales pitches get skipped. If you don't write these, third parties will, and their framing becomes yours.

The 30-day plan

• Week 1: Build the 40 prompts, run the baseline, set your four scores.
• Week 2: Find the 10 prompts where competitors win and you don't. Log their cited sources and run the diagnosis.
• Week 3: Ship answer-first pages and schema for those 10 gaps.
• Week 4: Launch outreach to your target source list. Re-run the baseline and compare.

Then repeat monthly. This compounds slowly, like brand building with a measurement layer, so don't expect an overnight jump.

Common mistakes

• Checking once and drawing conclusions
• Tracking mentions but ignoring the sentiment and descriptors
• Optimizing only your own site while ignoring third-party sources
• Chasing tricks instead of evidence
• Blocking AI crawlers by accident

The bottom line

AI search doesn't replace good marketing. It exposes it. Brands with sharp positioning, original thinking and a real footprint across trusted sources get recommended. The rest get left out of the answer.

Start measuring this week. The brands that begin tracking now will have months of trend data while everyone else is still wondering why a competitor keeps getting named.

PS: Before you do anything else, run your brand name through all three engines and ask, "Is [your brand] worth it?" The answer is usually more motivating than any statistic.