What does an AI visibility provider actually do?
It does one of three jobs, and the name on the invoice rarely tells you which.
- Monitoring platforms run a fixed set of prompts against several AI assistants on a schedule and report how often your brand is named and which sources are cited. Examples that describe themselves this way include Otterly.AI, Peec AI and Profound.
- GEO and content agencies work on what you already publish: page structure, schema markup, entity data, answer-shaped content. GEO (generative engine optimization) is the AI-era branch of SEO; see AEO vs GEO vs SEO for the definitions. Names that position themselves here include WebFX, Directive, Silverback Strategies and Minuttia.
- Source-building specialists create or repair the independent material models draw on: a Wikipedia article where the company qualifies, a Wikidata record, earned media. WikiBusines works in this group.
Most buyers start with a monitoring tool because it is the easiest to buy, then add a content retainer. The third layer is the slowest and is gated by notability rather than budget, so it often gets skipped. How AI systems weigh these layers is covered in how AI decides which brands to cite.
How we compared
We did not test any product and we do not rank vendors. The comparison is by type, using five questions you can put to any provider:
- What is the deliverable? A report, a changed page, or a new independent source.
- How is the result measured? Prompt set, models and versions, number of runs per prompt, and date.
- Is pricing public? For every named vendor above, we treat pricing as "on request" unless the vendor publishes it. Check the vendor's own page; prices change.
- Is the work compliant? Paid Wikipedia work must be disclosed under the Wikimedia Terms of Use. A provider that proposes hidden editing is creating a liability for you.
- What does the contract promise? Effort and deliverables are honest. Outcomes in an AI answer are not.
Disclosure: we sell the third type. The first two are categories we do not sell, and we have no commercial relationship with the vendors named. Names are given so you can recognise the categories; there are no links to competing vendors on purpose, so verify each one yourself.
When are monitoring platforms enough?
They are enough when your goal is to know where you stand. They are not enough when your goal is to change it.
A monitoring tool reports mention rate, the sources cited and, in some products, sentiment or share of voice across competitors. That is useful as a baseline and as a regression alarm after you make changes. It creates no citable source. A dashboard is a thermometer: it can show the fever and it cannot treat it.
What to ask a monitoring vendor: which models and versions, how many runs per prompt, whether you can export the raw answers, and whether the prompt set is yours or theirs. Variation between runs is normal, so a single number without a method is not evidence.
For a tool-by-tool view, see AI brand monitoring tools.
What do GEO and content agencies change?
They change the layer you own, and that layer helps most when the facts behind it are already established elsewhere.
Typical work: restructuring pages so a retrieval system can extract a clean answer, adding FAQ and Organization schema, consolidating contradictory facts across your profiles, and writing content that answers the questions customers type into assistants. This is legitimate and often valuable, especially when your own site is thin or inconsistent.
The limit is structural. AI systems treat first-party pages as claims to be checked, and for a company that has no independent coverage there may be nothing for them to check against. If your brand is absent from a whole cluster of prompts, ask the agency what independent sources it expects to exist after the work, and who will create them.
Two cautions: a retainer should list deliverables, not "visibility points"; and no agency can promise a particular answer from a model.
What do source-building specialists do?
They work on the evidence layer: the independent, citable pages and records that models and retrieval systems rely on.
For a company this usually means three things, in this order of effort: consistent entity data and a Wikidata record where the notability criteria are met; independent media coverage; and a Wikipedia article if, and only if, the company meets Wikipedia's notability guideline for organizations. Our Notability Index shows how uneven that bar is by sector (in our own assessments of 2018 to 2026, 98% of Crypto / Web3 brands rated low notability, against 19% in Finance / Fintech / Investment).
WikiBusines, founded in 2010 in Kyiv, works in this group. Paid editing is always disclosed on Wikipedia in WikiBusines' name, and we never guarantee that an article will be published or what an AI will say. Details are on the AI visibility page.
How do the three types compare?
| Monitoring platforms | GEO and content agencies | Source-building specialists | |
|---|---|---|---|
| Example names | Otterly.AI, Peec AI, Profound | WebFX, Directive, Silverback Strategies, Minuttia | WikiBusines |
| Main deliverable | Measurement and alerts | Changes to owned content and structure | Independent sources and records |
| Pricing model | Subscription; check vendor pages | Retainer; quote on request | Project or annual; published prices at WikiBusines |
| Moves the answer? | No, it only measures | Partly, through the owned layer | Through the sources models read; no guarantee |
| Main constraint | Needs someone to act on the data | Owned content is the least independent source | Notability and reviewer approval |
| Ask before buying | Method, model versions, raw export | Deliverables list, expected independent sources | Disclosure, notability assessment, refund terms |
What should I buy first?
Buy a diagnosis first, then fund the layer it shows is broken.
- If you do not know what AI says about you: run the same 10 to 15 prompts across ChatGPT, Gemini and Perplexity yourself and log the results in a spreadsheet. Add a monitoring tool when you need a monthly trend.
- If AI describes you inaccurately: trace the wrong claim to its source and fix the source. This is the subject of ChatGPT wrong about your company.
- If AI never mentions you in your category: the likely gap is independent coverage. Use the self-serve media list and Notability Index to see where you stand before paying anyone, then consider a notability audit.
- If you are already described correctly but not recommended: this is the territory of GEO and community work, covered in how to improve brand visibility in AI search.
For how Wikipedia in particular affects AI answers, see Wikipedia AI visibility.
What are the red flags when hiring an AI visibility agency?
Walk away from five patterns:
- Guaranteed mentions or rankings in AI answers. Model output is sampled and varies between runs, so no one can guarantee it.
- A "proprietary visibility score" with no method. You should be able to see prompts, models and run counts.
- Self-ranked "best agency" lists with no disclosure that the author is on the list.
- Hidden or undisclosed Wikipedia editing. It violates the Terms of Use and risks the page being deleted.
- One vendor selling all three types equally well. Measurement, content and encyclopedic work need different skills.
More in our red-flags guide.
Frequently asked questions
There is no single best, because the providers sell different things. Choose by the gap: a monitoring platform to measure, a GEO or content agency to restructure what you publish, a source-building specialist to create independent sources. Any list that ranks all three together is comparing unlike things.
A monitoring tool reports what AI assistants say about you. A GEO agency changes the content and structure you publish. Neither creates independent sources, which is the work of a third kind of provider.
No. Model answers vary by prompt, run, model version and account. A provider can improve the inputs, such as accurate entity data and independent coverage, but cannot guarantee output. Treat a guarantee as a red flag.
Monitoring tools are subscriptions and GEO agencies work on retainers; both usually publish limited pricing, so ask for a written quote. Our own AI visibility audit is €490 and the packages run from €800 to €3,500, as listed on the packages page.
No. A page helps only if your company is notable under Wikipedia's rules, and many are not yet. Wikidata, consistent entity data, independent press and community presence all matter and are more accessible.
Ask for the prompt list, the models and versions, the number of runs per prompt, the dates and the raw answers. A real operator hands these over. Re-run a sample yourself.
Paid editing is allowed when it is disclosed, as required by the Wikimedia Terms of Use and WP:COI. Paying for undisclosed editing or a guaranteed page is not allowed.
How WikiBusines helps
We cover the source-building layer. We do not sell monitoring or content retainers.
- AI visibility audit, €490: what the main assistants say about you today, which sources they lean on and which layer to fix first. It is not credited toward packages.
- Wikidata entity, from €680, and a notability audit from €490, credited toward page creation from €2,400 (English, 4 to 5 weeks), if ordered within one month.
- Annual support, from €720/year, with up to four updates and a 12-month guarantee. Near real-time monitoring is Managed Protection at €1,500/year.
Start with the free 10-minute assessment via contact. If the case is unclear, we offer the paid audits above.
Authors and editorial policy
Bohdan DubylovskyiCo-founder and CBO, WikiBusines→
Roman MelnykCo-founder & Chief Editor, WikiBusines→The article was prepared by the WikiBusines co-founders. We sell paid Wikipedia services, so we disclose this interest: paid editing is always declared on Wikipedia in WikiBusines' name, and claims about the rules are backed by links to primary sources.
Editorial policy: payment disclosure, conflicts of interest, sources