Hook: You have to open a cold conversation with something the prospect has not heard forty times this month. Researching one genuinely specific opener by hand costs EUR 23 to EUR 45 of SDR time, and most of it ends in a voicemail.
What You'll Be Able To Do
- Run your prospect's own market question on Gemini, ChatGPT and Grok from a public page
- Do it signed out, with no account, no card and nothing to install
- Read where the prospect sits, from Invisible through Known, not named and Mentioned, not backed, to Recommended
- See which of their rivals owns the citations behind the whole answer
- Send the prospect a link they open cold, on any device, a week later
Total Time: 5:34 | Difficulty: Advanced
The specific opener works, everyone knows it works, and getting one means researching the prospect properly. It falls apart at prospect eleven, when the research budget is gone and every opener collapses back into "I noticed you're in aviation". So the good openers go to the first ten names on the list and the rest get the generic one.
What it costs today
An SDR loaded at EUR 45 an hour. Research that produces one genuinely specific opener runs 30 to 60 minutes per prospect.
| Line | Rate | Time | Cost |
|---|---|---|---|
| One specific opener, researched by hand | EUR 45/h | 30 to 60 min | EUR 23 to EUR 45 |
The free check is one per address per week, so spend it on the prospect that matters most that week. Volume work runs on the signed-in Answer Trail, which takes one job per project at a time.
What comes back is a page you can read down the phone: three models answering your prospect's own market question, the brands each model named, and the share of citations sitting behind the whole answer, brand by brand. Their rivals' numbers are on the same screen as theirs.
What the video shows
Nine shots, in order, so a reader who cannot watch still learns the flow.
- Open the public page signed out. No login, no install, nothing that belongs to an account.
- Fill three fields with the prospect's details: their company or website, what they sell, and their city. The logo resolves as you type.
- Ask for the buyer questions their market actually asks. They come back in the market's language, so a Warsaw prospect gets Polish questions.
- Pick the question closest to what they sell, leave all three model chips on, and run the audit.
- Wait a couple of minutes. Everything moving on that screen is a placeholder, so there is nothing to read yet.
- The deck opens on their question put to three models, then the searches each model fired before it wrote.
- The pages every model read, the funnel counting who got named and who held up, and the donut showing who owns the citations.
- The cross-model view, one segment per model, then the reveal: every brand placed on the track, your prospect's chip marked.
- Copy the link from the last slide and paste it into the first message.
The track in shot 8 is the opener. A prospect sitting at Invisible gives you a first sentence nobody else on their call list has used this month. A prospect at Mentioned, not backed gives you a different one: the models name them and then fail to find anything to back it up, which is a proof problem they can act on.
The link from shot 9 does three jobs. It is the reason for the first message, the proof inside the call, and the attachment on the follow-up. The prospect opens it a week later with no account and no PIN, on a phone if that is what they have, and the run is dated on the page so nobody argues about when it was taken.
What you need
No plan and no account. The whole flow is public.
Before you start:
- The prospect's domain, what they sell and their city in front of you.
- A brand you are comfortable naming on a recording. The question is stored with the run and prints on the shared page.
- Two to five minutes for the run itself, so plan the call prep around it.
Questions people ask
Do I need a Rankfor.AI account? No. Nothing to seed, nothing to install, no card.
Which models answer? Gemini, ChatGPT and Grok. There is no fourth model on this page, and neither Claude nor Perplexity answers here.
Can I put my whole list through it? No. One free check per address per week, and this is no bulk enrichment tool. Volume work runs on the signed-in Answer Trail, which takes one job per project at a time.
Does it watch the prospect and tell me when their answer changes? No. There is no monitoring and no alert. The page is a snapshot of one run on one day, dated on the deck, and a second reading means a second run.
What if my prospect comes out looking fine? Then you know that before the call, and the donut still shows which rival owns the citations behind the answer. We measure what the models say. We do not promise a position and we cannot move one on demand.
Run it once on the name at the top of your list, and let their own answer do the opening. Start with Rankfor.AI.