Hook: See the places AI actually cited when it learned about a person, and decide where the personal-brand budget goes. One round of that work by hand costs EUR 540 to EUR 900.
What You'll Be Able To Do
- Run the same scan on a person that you run on a company, and read the score with its six weighted parts
- Open the places a model actually went to learn about them, grouped by source type, with the share each took
- Filter the same list by reputation theme and by sentiment, so the few damaging pages take minutes to find
- Export every citation as a CSV somebody else can check
- Argue with "post more" using the product's own scoring table, and point a retainer at something specific
Runtime: 6:10 | Difficulty: Advanced
Someone decides whether to shortlist you by asking a model. Three prompts, ninety seconds, and a list exists. You were not in the room and you have no idea what was said. The instinctive fix is to post more, and this playbook argues with that advice using the product's own scoring table.
What it costs today
The person or their comms lead at EUR 90 an hour. One round means running the probes by hand, then classifying and grading the results.
| Line item | Assumption | Arithmetic | Cost |
|---|---|---|---|
| Running 18 probes by hand, two languages | 36 runs at 3 to 5 min | 36 × 3 ÷ 60 and 36 × 5 ÷ 60 | 1.8 to 3.0 h |
| Classifying ~200 cited URLs by source type | 20 s each | 200 × 20 ÷ 3600 | 1.1 h |
| Grading each citation for sentiment | 20 to 30 s each | 200 × 20 ÷ 3600 and 200 × 30 ÷ 3600 | 1.1 to 1.7 h |
| Assembling the mix and the spend call | 2 to 4 h | ||
| One round, one model | 6 to 10 h at EUR 90 | 6 × 90 and 10 × 90 | EUR 540 to EUR 900 |
| Bought in as an executive reputation audit | comms agency, one-off | EUR 2,000 to EUR 6,000 | |
| The budget this steers | personal-brand retainer | flat monthly | EUR 1,500 to EUR 4,000 per month |
| Marketer plan | EUR 196 per month |
The labour line is EUR 540 to EUR 900 a round against EUR 196 a month. The bigger number is the retainer: EUR 1,500 to EUR 4,000 a month, allocated on instinct today. This is the first evidence anyone has for moving it.
What the video shows
- The person's name goes into the public Dice Roller, no account, and the same question runs several times over. The answers spread.
- Inside the product, a person is a project like any other: a name and a website, and the website is required before a scan can run at all.
- The Rankfor Index: one score built from six weighted parts, with recommendation share carrying the most and source quality ten percent.
- A part with nothing to measure is withheld, so a blank card means unmeasured.
- "Where AI gets your story": the donut, the source types, the share each one took, and the real cited pages inside every row.
- The theme filter. Trust and governance pulls a different set of pages than innovation.
- The sentiment filter. Click Negative and see what survives, which is usually very little.
- The sources CSV, exported and opened: one row per citation, with its address, title, source type, theme, sentiment and language.
- The "Counter with a placement" link, followed through to Paid Boost, stopping at the budget screen.
What you need
- Marketer, EUR 196 a month, for the Rankfor Index.
- A live domain in the person's name, before anything else. A project with no website cannot be scanned, and a domain they own carries one of the highest weights.
- The project's region and industry set; both ride into the questions the scan asks.
- A completed scan. The cap is one per model per calendar month, across Gemini, ChatGPT and Grok.
- Written consent. Every page on screen is a real page about a real person.
Questions people ask
Does posting more work? Dmitrij ran it on himself first. His score came back 58, and the three places the model went to learn about him were a research profile, an academic paper and a conference speaker page. Not LinkedIn, where he had posted for years. In the scoring table social media carries 0.25, academic 0.9, tier-one news and your own domain 0.8. Those are weights, frozen in July 2026 after an offline reclassification of 4,000 production citations, and they say nothing about what a placement costs or the traffic it sends.
So a better source mix triples my score? No. Source quality carries ten percent of the composite. Three times the weight inside that one part is the claim the data supports.
Do all seven themes work for a person? Four do: overall, what they are known for, innovation, and trust. The other three ask about an employer, so workplace, sustainability and growth come back thin for a person.
Is there a separate reputation product for people? No. A person is a project like any other, measured with the same instrument as a company. Nothing person-specific exists to buy.
Watch it, then read the source list for the person you are steering.