For marketing agencies
AEO and GEO audits for marketing agencies
Picture a skincare client. The founder opens the review with one question.
When someone asks ChatGPT for a retinol, are we on the list?
Search rankings and ad reports do not show that. Run an AskLLM audit, then use optimization to refresh which sources matter and to prove movement on the next review.
What changed for buyers
Google still matters for research. The shortlist often gets picked in chat first.
They ask AI first
The buyer describes their skin and budget in chat. ChatGPT or Gemini returns a short list. They often never open a results page.
It reads like advice
The answer feels personal. There is no sponsored tag, so the names on that list feel chosen, not bought.
Search ads arrive late
If the assistant already named two other brands, your client’s Google spend reaches the buyer after the shortlist feels settled.
That is why you need an AskLLM report built for ChatGPT and Gemini, not a screenshot from one chat.
How agencies use AskLLM
- 1
Capture the question
Does ChatGPT recommend us for retinol?
- 2
Run one audit
That brand and one competitor. Same prompts every time.
- 3
Read the report
Score, both models, and where the competitor wins.
- 4
Ship one slide
Pitch, quarterly review, or renewal conversation.
Questions buyers ask first
Skincare buyers ask like this before they click. Copy a line into the deck, or run the audit and we ask it on ChatGPT 5 and Gemini 2.5 Flash-Lite.
- ChatGPT 5Gemini 2.5 Flash-Lite
“Best retinol for sensitive skin. Which brands does ChatGPT shortlist?”
- ChatGPT 5Gemini 2.5 Flash-Lite
“We sell clean beauty. Does ChatGPT name us or our competitor first?”
- ChatGPT 5Gemini 2.5 Flash-Lite
“When Gemini recommends a skincare brand, which sites does it cite?”
Why only these two
- ChatGPTMost visits
- GeminiSecond
- ClaudeNot in the audit
- PerplexityNot in the audit
We audit ChatGPT and Gemini. Together they are 81.1% of those visits. Website traffic only, not phone apps. Similarweb, August 2026.
After the slide: audit plus optimization
AskLLM is not a static recipe anyone can lift from a blog. Paid plans connect measurement to a living playbook: which citation surfaces matter in that category right now, forensics on the pages the models named or skipped, and a re-audit interval that shows when change is real.
Not a blog you bookmark once
Generic GEO lists go stale. AskLLM treats the per-category citation map as something we refresh from new measurement, not a PDF you buy once and file away.
Which sources matter now
Each month the category mix can shift. The report shows which publishers, reviews, and retailers models actually cite for that category today, not what an article recommended last year.
Forensics on real URLs
Optimization ties gaps to specific pages from the audit: where the competitor appeared, which category sources your client missed, and what to fix before the next run.
Proof on a schedule
Re-audit on a monthly or quarterly rhythm. Drift settings help you show the client when LVI moved for real, not because someone ran one optimistic chat.
Run one client brand. Name one competitor. Bring one slide to the review.
Questions agencies ask
- Can we run this for every client on the roster?
- Yes, but run them separately. Each brand gets its own audit, its own competitor, and its own score. A pet food client never shares prompts or numbers with a SaaS client. That is how you reuse the same slide shape in every pitch and every quarterly review without mixing accounts.
- What exactly goes on the slide?
- Start with the visibility score. Add one line on how ChatGPT describes the brand and one on Gemini. Then show the gap: where the competitor you named shows up in the answer and your client does not. You do not need to paste the full report. Most agencies pull three bullets and one chart.
- Does this replace our SEO audit or Search Console work?
- No, and your SEO team should keep their work. Rankings and clicks still matter. This answers a different question: when someone asks ChatGPT or Gemini for a shortlist, is your client on it? Use both. When the founder asks about AI, this is the page you show. When they ask about Google, stay on search.
- Why only ChatGPT and Gemini? Our client keeps asking about Claude.
- Gemini is in the audit. Claude is not. Similarweb, August 2026, counted visits to generative AI websites worldwide: ChatGPT 55.5%, Gemini 25.6%, 81.1% together. Claude 9.3%. Perplexity 0.9%. That is website traffic, not phone apps. We focus on the two assistants that take most of those visits so the slide matches where buyers actually ask, without paying for four models on every client.
- We still run Google Ads. Is this saying search ads do not work?
- No. Many buyers still search, click, and convert from ads. The shift is earlier in the journey. Someone asks AI for two or three brand names, trusts that list, and only then might search. If your client is missing from the AI answer, the ad often fights for a buyer who already has a shortlist. The audit shows that miss. It does not tell you to turn off paid search.
- The founder already opened ChatGPT and said the brand looked fine. Why run this?
- One chat is not a record. They may have asked a soft question, skipped the competitor, or hit a model update a week later. The audit runs the same category buyer prompts on ChatGPT and Gemini, with the competitor you name, and saves a private report link. You can rerun it before the next review and show change over time, instead of arguing from a screenshot.
- Can we just follow a GEO checklist from a blog?
- A blog post is frozen. AskLLM refreshes which citation surfaces matter in your client’s category as models shift, maps gaps to specific URLs from the audit, and uses scheduled re-audits so you can show when visibility actually changed. The per-category source list is output we regenerate, not an asset you buy once and photocopy.