AI answers are assembled from sources. See yours.
A web-grounded answer engine reads before it writes. The set of documents it read is the closest thing to an explanation you can get from the outside — so AnswerRank records it, canonicalises it, classifies it, and shows you where you are missing from it.
- Provider-attested
- Citations are the sources the provider itself returned, not sources we guessed at.
- Canonicalised
- One page is one source: hosts normalised, tracking parameters and fragments removed.
- Observed only
- Influence is scored from citations seen in our own scans. It is not a training-data claim.
From a cited URL to something you can act on.
Four steps sit between a raw citation and a number on a chart. Each one exists because skipping it produces a figure that looks precise and is not.
- 01
Collect what the provider attested
Citations come from the provider’s own response — the sources the API attributed to the answer, or the grounding metadata it returned. We do not infer a citation that was not given to us.
- 02
Canonicalise the URL
Host lowercased, www stripped, tracking parameters removed, fragment dropped, trailing slash normalised. Without this, one page appears as five sources and every influence figure is inflated.
- 03
Deduplicate within the answer
A source cited three times in one answer is one citation for that check. Repetition inside a single response is a formatting artefact, not extra evidence.
- 04
Classify the domain
Owned, competitor-owned, review platform, community, publication or third party. A classification we cannot make confidently resolves to unknown rather than to a guess.
Owned, earned, or someone else’s property.
Every cited domain is placed in one of seven classes. The classes are not cosmetic — they decide which lever, if any, you actually have.
| Class | What it covers | What you can do |
|---|---|---|
| Owned | A domain you control — your marketing site, docs, changelog, help centre. | Fully in your hands. The fastest thing to change, and usually the thinnest. |
| Competitor-owned | A rival’s own property, including their comparison and alternatives pages. | You cannot edit it, but you can publish the counterpart the engine currently has no access to. |
| Review platform | Directories and review marketplaces that aggregate category listings. | Earned. Profile completeness and recency are inside your control. |
| Community | Forums, Q&A sites and discussion threads where practitioners compare notes. | Earned and slow. Participation is legitimate; manufacturing threads is not, and we will not recommend it. |
| Publication | Editorial outlets, trade press and independent analyst writing. | Earned through relationships and something genuinely worth covering. |
| Third party | Other sites that cite you or your category without falling into the classes above. | Mixed. Worth reading individually before deciding anything. |
| Unknown | The classifier was not confident. Deliberately preserved as unknown rather than forced into a bucket. | Reported honestly, so aggregate class percentages are not quietly fabricated. |
Two different problems that look like one number.
A thin owned-citation share is a publishing problem you can fix this quarter. A thin earned share is a credibility problem that takes longer and cannot be shortcut with more pages.
Citation mix · Acme · 240 valid checks
- Owned (acme.com)11%
- Review platforms26%
- Communities19%
- Publications16%
- Competitor-owned21%
- Unknown7%
Unknown is shown rather than redistributed. Hiding it would make the other six figures look more certain than they are.
Reading the split
- Owned share tells you whether the engine can find your own account of what you do — and whether it considers that account worth citing.
- Earned share across review platforms, communities and publications tells you whether anyone independent is describing you at all.
- A large competitor-owned share means rivals are supplying the comparison material the engine reaches for, and yours is absent from the pool.
- The split is available per engine, because grounding indexes differ and a source that dominates on one surface can be invisible on another.
Sources that cite your competitors, and never you.
This is the single most actionable view in the product. These domains already shape answers in your category and already influence buyers — the engine simply has no reason to name you when it reads them.
| Source | Class | Prompts | Brands named | Status |
|---|---|---|---|---|
| saasreviewhub.com | Review platform | 14 | Northwind, Initech | Cites rivals, not you |
| devforum.io | Community | 11 | Northwind | Cites rivals, not you |
| northwind.com/compare | Competitor-owned | 8 | Northwind | Cites rivals, not you |
| martechweekly.com | Publication | 9 | Northwind, Acme | Cites you |
| acme.com/pricing | Owned | 3 | Acme | Cites you |
Each row opens onto the prompts and answers where the source appeared, so you can read what the engine actually did with it before deciding whether the gap is worth closing.
A score built only from what we saw.
Observed Source Influence ranks domains by how often and how prominently they appeared in the citation sets behind your category’s answers, weighted by the intent of the prompts they showed up on.
- Derived purely from citations recorded during AnswerRank scans of your prompt set.
- Weighted toward high-intent prompts, because a source that shapes purchase answers matters more than one shaping definitions.
- Scoped to your category and market — a domain influential in one category can be irrelevant in another.
- Reported with the number of observations behind it, so a source seen twice is never presented like a source seen two hundred times.
Explicitly not a training-data claim
Observed Source Influence is a measure of what appeared in the citation sets we collected. It is not a statement about what any model was trained on, what is in a provider’s index, or how a provider weights a domain internally. None of that is visible to any third party, and a tool that implies otherwise is selling you a guess.
It is also scoped in time. Influence is measured over the scans you ran, across the prompts you tracked. It is not a permanent property of a domain, and it says nothing about sources we never observed.
Getting cited by a high-influence source is associated with appearing more often in the answers that source informs. It is not a guarantee of anything, and the product will never phrase it as one.
Prompts, sources and brands are one connected record.
Because every citation is stored against the check it came from, the relationships are traversable in any direction — which is what turns a list of domains into an argument.
Start from a prompt
Open any tracked question and see every source the engines cited when answering it, across engines and across scan runs.
Start from a source
Open a domain and see which prompts it influenced, which brands it named, and how consistently it reappears.
Start from a competitor
Open a rival and see the source set their presence travels with — the shortest path to a target list for earned coverage.
Evidence suggests that closing a citation gap on a source that repeatedly shapes high-intent answers is a likely contributing factor to appearing in those answers. Whether it worked in your case is a question for an experiment, not for a marketing page.
See which sources shape answers about you.
One free check across the answer engines your buyers actually use.
No signup required for your first visibility check.
Every figure AnswerRank reports carries its sample size and how it was collected.
Read the methodology