Three-hour reporting nights. Screenshots pasted into Google Docs. A prospect nodding politely while you explain why they should pay you €3,000/month to "fix their AI visibility." And a nagging feeling that the whole thing would land better if you had actual data.
If you sell GEO to clients, the pitch is easy. The proof is hard. The Audit feature is there to close that gap.
What you actually deliver to the client
An audit is not a dashboard. It is a report. A single, shareable page that your client can open, read, and forward to their CMO. It contains six sections, and each one answers a question the client is already asking.
Competitor podium. The top 8 brands ranked by how often they get mentioned across LLM responses. Your client sees exactly who AI is recommending instead of them, and by how much. No more "you're invisible on ChatGPT" hand-waving. You have a number.
Prompt coverage heatmap. Every prompt you scanned, cross-referenced with every LLM. Green where your client shows up, red where they do not. If you used personas, the heatmap is split by persona so your client sees which segments they are ghosted in.
Top citation sources. The websites LLMs quote most often when answering the prompts you scanned. This is your outreach hit list. Some of those sites are paid directories. Some are industry blogs. All of them are already feeding AI the answers your client wants to show up in.
Quality backlinks. The citation sources cross-referenced against a real backlink database. This filters the noise (random Reddit threads, low-authority blogs) and surfaces the domains actually worth pursuing.
AI-generated diagnostic. Claude reads the raw data and writes a plain-English explanation of what is going on. Why the client is invisible on Perplexity but strong on Gemini. Which personas are underserved. What content gaps are blocking visibility. It reads like a senior consultant wrote it, because that is what it was trained on.
Recommended content plan. Six to twelve specific content pieces the client could produce to close the gaps. Each one tied to a prompt that is currently driving competitor mentions. The client does not have to guess what to write next. You hand them the brief.
How an audit runs
The wizard has five steps and the whole setup takes about 10 minutes.
- URL. Enter the domain. Mentionable scans the website to extract the brand name, core description, and language. No manual form filling.
- Personas. The system generates 3 to 5 buyer personas automatically based on the site content. Refine them, swap them out, or skip the step if you just want unflavored prompts.
- Prompts. Either write your own prompts or let Claude generate 15 to 20 based on the personas and the website. You pick the ones worth scanning (typically 5 to 20).
- Configuration. Select the LLMs to scan (ChatGPT, Perplexity, Gemini, Copilot), the target country (affects localized results), and how many times to repeat each query (1 to 10 times per LLM for consistency checks).
- Preview. See the exact credit cost before launching. Confirm. Go.
The scan runs asynchronously. You close your laptop. An hour or two later, the report is ready.
Audit vs. tracking: which one do I need?
Both. They solve different problems.
| Audit | Daily tracking | |
|---|---|---|
| When to use | Pitch, proposal, one-off reporting | Ongoing monitoring of your own brand |
| Output | Shareable white-label report | Internal dashboard + email digests |
| Domain | Any (client, prospect, competitor) | Your own project |
| Frequency | One-shot (re-runnable) | Daily |
| Personas | Per-audit choice | Tied to project setup |
| Deliverable for client | Yes | No |
| Cost model | Credit-based, pay per audit | Included in subscription |
| AI diagnostic + content plan | Yes | No |
Agencies use both. Tracking is what keeps you informed week to week. Audits are what you ship to clients.
Credit-based pricing, no waste
Audits are billed in scan credits. One credit equals one prompt × one LLM × one persona × one occurrence. If you run 10 prompts across 3 LLMs with 2 personas once, that is 60 credits. If you only want a quick look, run 5 prompts on 2 LLMs with no personas and pay 10 credits.
The Agency plan uses custom pricing based on volume (contact sales). Credits, projects, and LLM coverage are tailored to your needs. At typical audit sizes (60 to 150 credits each), a single Agency bundle can cover hundreds of audits per month. You can buy extra credit packs if you run more.
This matters for two reasons. First, you never pay for unused capacity. Second, you can show clients exactly what their audit cost you, which makes pricing conversations clean.
Three ways agencies are using it
Pitching prospects. Run an audit on the prospect's domain before the discovery call. Walk into the meeting with the report already open. Skip the "let me explain what GEO is" slides and go straight to "here is what AI is saying about you right now, and here is what your top three competitors are doing differently." Close rate goes up. Call length goes down.
Monthly client reporting. Re-run the same audit every month. Send the new report with a one-paragraph summary of what changed. Clients see progress (or the lack of it) in numbers, not vibes. Renewals get easier because the value is visible.
Competitive intelligence. Audit your client's top competitors. Find the prompts where they are winning, the sources citing them, the personas they are targeting. Use that to build a content plan that systematically takes back the visibility. Clients love this because no one else is doing it.
White-label, not co-branded
On the Agency plan, the report is yours. Your logo. Your colors. Your domain (if you set up a custom domain for the white-label feature). The client never sees the word "Mentionable" unless you tell them. This is the difference between looking like a reseller and looking like an agency that built a proprietary tool.
Try it yourself
Run your first audit and see what a client-ready report actually looks like. Pick a prospect you have been meaning to pitch, or a competitor of a current client. Twenty minutes from now, you will have something worth sending.