Citations

Monitoring

Citations

• Overview
• New Citations
• Who's Citing You
• Browse Citations

Monitoring

Citations

• which sources shape answers in your category
• what has started citing you recently
• which sources reference your brand
• search everything, read the answers

Overview#

When an AI engine answers a question, it is drawing on sources to do it. Citations shows you which ones.

This tends to be the most immediately useful screen in Seerly, because citations sit upstream of everything else it measures. The sources an engine leans on are what shape whatever it says about you, so influencing those sources is how you influence the answer itself. Visibility tells you that you are missing from somewhere; Citations tells you what is winning in your place, which is the half you can do something about.

The Overview sub-page#

This is the starting point. Sources here are ranked by how much they shape answers in your category across your tracked prompts, rather than by general popularity or authority, which means a niche industry directory can comfortably outrank a famous publication. When that happens it is a genuine finding rather than a quirk of the data, and usually a useful one, since the directory is far easier to get onto.

Citation % shows how much of the total citation volume a source accounts for, and AI Platforms breaks that down by engine, which matters because engines favour noticeably different sources. Break down by domain groups results by source site rather than individual page, and that is usually the right view when you are deciding where to invest effort, since you are choosing a relationship rather than a URL.

Never cited you is the label worth knowing about. It marks sources that demonstrably influence your category and do not mention you anywhere at all. High influence combined with zero presence is, in effect, your priority list already sorted. On this sub-page the label appears in the Sources citing your topics list, where each source also says which engine it is Favoured by, on a line under the domain, when one can be named.

That list sits under Which sources to target and splits into Gaining ground and Losing ground: the share of all citations each site accounts for, now against before. It compares two periods of the same length and reaches back as far as your history allows: 90 days against the 90 days before, 30 days against the previous 30 while there are too few runs for that, and otherwise the latest week against the previous week. The dates at the top right of the card are the two periods being compared, so a source that has been climbing for months shows up as such rather than as one good week. Because the figures are shares, a site can gain ground without being cited more times, since the total number of citations differs between the two periods. Until the previous week has a completed run there is nothing to compare, and the list says so in place of a trend.

Never cited you, applied

Source	Citation %	Favoured by
an industry directory	14%	ChatGPT
a review site	11%	Perplexity
a comparison blog	6%	Google AI Overview

Source Balance splits your citations into your own pages against third-party ones. Both numbers matter and neither has a target: engines weight independent sources heavily, so a position resting mostly on your own domain is more fragile than a healthy-looking count suggests. Your own site appears in the source lists alongside everything else, on this screen and on the dashboard, which is what makes the split visible in the first place. It is tagged Your site rather than given a source type, because the type describes the kind of page an engine reached for and that is not the useful thing to say about your own domain. If your brand runs more than one site, the sites you list under Your websites in Brand Hub count as your own pages here and carry the same tag.

When the picker at the top of Citations is set to a competitor rather than your brand, How you compare takes the place of Risks. Its Citation Gap card compares that competitor's cited pages with yours, and its Sources you're missing list names the sources that cite that competitor and have never cited you, each with how many of its pages cite the competitor. Each row also names the engine the source is Favoured by, when one can be named, and opening a row shows what the source cites the competitor for. For the topics you are missing from on each engine, see Gap analysis.

New Citations#

This shows what started citing you in the latest week, against how many did in the previous week, and it is the natural place to confirm whether work has landed. Latest week is the default; Last 30 days widens it. A page you published six weeks ago appearing here is about as clear a piece of evidence as you will get that the loop is functioning.

A citation counts as new only if no earlier run had it: First cited in the latest week marks a genuinely new source rather than one that had gone quiet and has since come back, which is a distinction worth having when you are trying to attribute a change to something you did.

Who's Citing You#

Every source referencing your brand in the period you choose, with how often each one does it. The switch at the top right of the screen sets the period: Latest week, which is where the screen always opens, Latest month, the 30 days ending at your most recent completed run, or All time, every run you have. Each period ends at that same most recent run, and the dates it covers are shown beside the switch. A link from another screen always opens the latest week.

The period decides which sources are listed, the source type breakdown, and each source's Citations, Responses and Reach. A source that cited you before the period but not inside it is not listed. On the latest week and the latest month, a small arrow and number beside a count is the change against the previous week or the previous 30 days, and nothing is drawn when the count held or that previous period had no runs. All time is compared with nothing, so it shows no change at all. Reach is how many AI answers about your category cited the source in the period, whichever brand they named. Last cited is the date of the most recent run in which the source cited you. Influence and Favoured by are the exception: they are measured across all your runs whichever period is chosen. Your own pages appear here alongside review sites, directories, news coverage and community discussion, and it is common to find the third-party sources doing considerably more work for you than your own site is. Sites you have listed under Your websites in Brand Hub are counted as your own here, not as external sources.

Each row is one website. When any of its citations come from an address other than the site itself, such as a store or another subdomain, the addresses are listed in small type under the name, busiest first. The source type under the name is the one the website is cited through most often across answers. A single citation in the Browse list shows the type of its own address, so the two can differ for a website whose addresses serve different purposes.

Influence scores how much the engines lean on each source and bands it Dominant, High, Moderate or Minor, drawn as a bar with the score and band beneath it. Under the Reach figure sits a badge counting the answers in the chosen period where that source cited you, amber and reading Never cited you when there are none. The two answer different questions: a source can shape answers across your category heavily while mentioning you rarely, and that gap is the opening. The table is ordered by this score. Where a project has too little to compare against, such as very few runs or a single engine, the score is left out rather than shown as a low one.

The info icon beside the Influence heading says the same thing on the screen. What the score weighs, in order: how often the engines reach for the source at all, which counts for considerably more than the rest, then how many different engines cite it, how much of your topic space it covers, and how many of its individual pages get used. Breadth counts because a source one engine reaches for on one topic is a narrower bet than one the whole field uses. Two things are worth holding on to. The score is a standing within your project rather than a universal rating, so it can move when other sources move even though nothing about this one changed. And it counts every answer in your category, not only the answers that mention you, which is exactly why the cited-you badge sits beside it instead of repeating it. See Citation influence for the fuller definition.

The last column, Favoured by, names the engine that leans on each source most, measured as described under Browse below. A dash means no engine could be named, which is not a finding. The column does not change the order, which stays by Influence, and it is left out when no engine can be named for any source in the table, as on a project tracking a single engine.

Browse Citations#

The full searchable view, and the place where you can read the answers themselves rather than counting them.

In the domain tile view each tile carries the same Influence score and band as Who's Citing, with the same explainer on its info icon, and the cited-you badge in the top corner. Your own site is tagged Your site here too. When you are looking at a competitor's citations, here or on Who's Citing, the marked site is the competitor's own and reads Their site; the competitor badge would only repeat it, so that row leaves it off.

None of this disappears when you filter. Influence is measured across all your runs whatever the grid is currently narrowed to, which is deliberate: it describes the source rather than the slice, so it does not change when you pick one engine or one topic. That is also why it is worth reading next to a filtered view. You can see which of the sources answering this topic are the ones the engines lean on generally.

Each tile also shows when the source was Last seen: the most recent answer that cited it. Unlike Influence, this follows your filters, as the citation count beside it does, so narrowing to one engine shows when that engine last cited the source.

Favoured by names the engine that leans on the source most, adjusted for how many prompts each engine answered, so no engine wins just by answering more of them. Like Influence, it is measured across all your runs and does not change when you filter. It only names an engine you still track that has at least 3 answers citing the source, and a project tracking a single engine never shows it, since there is nothing to compare. Either way, a tile without Favoured by is not a finding.

A tile whose domain belongs to a tracked rival reads Direct competitor or Indirect competitor in place of the cited-you badge, because whose site it is changes what the score means for you. A Dominant score on a directory is an opportunity; the same score on a competitor's own domain is not, since there is no version of that where you get published on it. And "never cited you" says nothing there: of course they have not. Where a rival's site has cited you, that count still appears beside the citation count, because it is a genuine finding. Everything else carries no tag, which is the case worth acting on and by far the most common.

The List view shows the same Influence score and band on each row's domain, the same Favoured by line under it, and the same tag when the source is your own site or a competitor's, so a citation you are reading carries the standing of the source it sits on. That standing repeats down the rows of a domain, which is the trade for having it at all in a per-citation view. The Sentiment column appears only when sentiment prompts are in scope: under Discovery Only the runs carry no sentiment at all, so the column could only ever be empty.

You can narrow things with Filter by AI engine, Filter by topic, Filter by sentiment and Filter by type. Filtered by shows what is currently applied and Clear all resets it, which is worth knowing because filters persist between visits, and an empty result is very often a filter left on rather than a genuine absence. Limited to the selected topic appears when a topic filter is narrowing what you can see.

A link from another screen can arrive with a topic, prompt or engine the filter lists do not contain, for instance a prompt with no citations yet. The filter still applies, so its control and chip read Unavailable in this view (an engine shows its own name) rather than pretending nothing is filtered. When nothing matches, the message names every filter that is on.

Narrowing Browse Citations

Filter by AI engine

one engine's answers only

Filter by topic

one tracked topic

Filter by type

the kind of source

Open link takes you to the cited page, and it is worth doing this more often than you probably will. A citation supporting a recommendation and a citation supporting a criticism are indistinguishable in a count and are not remotely the same thing for your business.

Best practices#

• A mix of sources. If every citation traces back to your own domain, your position is more fragile than the numbers make it look, because engines weight independent sources heavily.
• Recurring rather than scattered. A handful of sources citing you consistently is worth more than a long list of sources that cited you once. Recurrence is the signal.
• Cited in the right context. Being referenced inside a recommendation is worth a great deal more than turning up in a list of things to avoid, and the count alone cannot tell them apart.
• Presence on the sources that genuinely influence your category, rather than presence everywhere for its own sake.

Next steps#

1. Start at Overview and look for Never cited you, since sources with high influence and no presence are the shortest path you have to an improvement.
2. Do the joinable ones first. Directories, listings and review sites are frequently a form submission rather than a content project, and their effects compound quietly.
3. Check your details wherever you already appear. An outdated listing works actively against you, because engines will repeat what it says without knowing it is stale.
4. Study the pages being cited instead of you, paying attention to format and specificity rather than length.
5. Make your own pages citable. Clear structure and direct, concrete claims get quoted, while general marketing prose gives an engine nothing to lift out. Smart Audit covers the mechanics of that.
6. Watch New Citations after publishing rather than watching visibility, because citations move first and visibility follows them.

If you see No citations found for this prompt or Not enough data to show trends yet, that usually means too few runs have completed so far rather than a genuine absence of coverage.

What to keep in mind#

Citations reflect what engines referenced when answering your tracked prompts. A source cited in conversations you do not track will not appear here at all, which is another reason prompt coverage matters. See Prompts.

Citations are also not links. A site can link to you without any engine ever citing it, and that is a separate measurement covered by Web Presence.

Engines expose their sources differently, so how deep the citation data goes varies between them. An engine showing fewer citations is not necessarily consulting fewer sources; it may simply be less forthcoming about them.

Earning a citation also takes time. Published content has to be discovered and indexed before it can be referenced at all, and that typically spans several analysis cycles rather than days.