Competitors

Monitoring

Competitors

• Head to head
• Tracked competitors
• Competitor intelligence

Monitoring

Competitors

• your brand beside one competitor
• you and everyone you track in one table
• what they published each week

Overview#

Competitor analysis runs rival brands through exactly the same prompts you are measured on, so what you get back is a like-for-like comparison rather than two separate reports laid next to each other and squinted at.

The comparison carries more weight here than it would in traditional search. A results page has ten positions on it and you can comfortably occupy one of them; an AI answer typically names three brands and moves on. Fourth place is not a weaker showing, it is absence, and that is why your position relative to everyone else ends up being the number that matters.

Competitors has three pages, listed beneath it in the sidebar. Head to head puts your brand beside one competitor at a time and shows where each of you is ahead, topic by topic and source by source. Tracked competitors puts your brand and everyone you track in one table, and it is where you manage that list. Competitor intelligence, listed as Competitor intel in the sidebar, reads what they publish, week by week (see Competitor intelligence).

On Head to head and Tracked competitors, the AI Engine select at the top narrows the weekly figures to one engine's answers; anything labelled "all engines" stays across every engine. Each page keeps its own choice, and a competitor's Head to head button on Tracked competitors opens the comparison on the same engine.

Every company on these pages, and in All Competitors, is shown by its name with its domain under it. Your brand goes by its name in Brand Hub. A competitor goes by the name its own site gives it, or the name Seerly has seen it called in answers; when Seerly only knows its domain, the domain is shown once rather than twice.

Choosing who is tracked#

Select Competitors is where you manage the tracked list. Seerly proposes likely competitors based on your category, and you decide which of them actually get measured. Do this before you read anything else on these pages, because every comparison here is drawn against that list, and the wrong list will give you a beautifully precise measurement of a race you are not running. The list is long, so there is a search box, and beside it a filter that narrows it to direct or indirect competitors. Each company's type is shown in its colour, red for direct and blue for indirect.

Open it with Manage tracked competitors on the Tracked competitors page, or with Track another at the bottom of the competitor menu on Head to head. View all opens All Competitors, everything Seerly found in your category, ranked by share of voice over the latest week with finished analysis. The list names that week, and it is the same week the Tracked competitors page shows, so a company's share is the same in both places. Reset to Default puts Seerly's own selection back if you want to start again. For a read-only member, Manage tracked competitors, Track another and the Type menus are disabled: they see the list without the controls to change it.

Select Competitors

All Competitors

everything Seerly identified in your category

Excluded

not a competitor, and left out of your measurement

Indirect

competes for the same answers without selling the same thing

Reset to Default

back to Seerly's own selection

Brands marked Indirect compete for the same answers without selling quite the same thing, and they are worth keeping in. In an AI answer an indirect competitor occupies the slot every bit as effectively as a direct one does, so excluding them tends to flatter your numbers without changing your situation.

Correcting how a competitor is classified#

Seerly works out whether each brand it finds is a direct competitor, an indirect one, or not a competitor at all. It gets this wrong sometimes, usually by reading industry adjacency as competition: a hospital alongside a fitness brand, a consultancy alongside a software product. On All Competitors, the Type column is a menu, so you can say so. Each type has its own colour there, red for direct, blue for indirect and grey for the rest, so the two are easy to tell apart down a long list.

Type

Direct

competes for the same buyer with the same offering

Indirect

competes for the same answers without selling the same thing

Excluded

not a competitor, drops out of your measurement

Auto

let Seerly decide, clears the label until the next analysis

Change as many rows as you like, then Save. A bar at the top of the panel counts what is waiting, and Discard throws the lot away. Nothing is written until you save, and a save applies every change together or none of them.

A classification you set is kept. Seerly will not quietly reconsider it later, which is the point: once you have told it that a supplier is not a competitor, it stays that way through every future analysis. Rows you have set yourself read Set by you under the brand name, so you can tell your decisions from Seerly's.

Auto is how you hand one back. It clears the label, and Seerly works the row out again on its next analysis, usually within a few minutes. Use it when you change your mind, or when you set something by accident.

Two consequences worth knowing before you save:

• Marking a tracked competitor Excluded also stops tracking it, because a brand you have said is not a competitor should not be sitting in your comparisons. Seerly tells you which ones it dropped.
• Setting every competitor on the project to Auto at once leaves Seerly with nothing to go on until it has re-read them, and it asks you to confirm before it does that.

The reverse does not happen on its own. Seerly never excludes a competitor you track: if a later analysis decides a tracked brand is not a competitor after all, Seerly keeps it as Indirect and leaves it on your list. Only you can exclude it.

A second site of your own can show up here as a competitor, because answers name it separately from your brand. Confirm it under Your websites in Brand Hub and it stops being listed as a competitor: its mentions count as yours, and it leaves your tracked list if it was on it. A site on your project's own domain, such as a subdomain of it, needs no confirming: Seerly marks it Excluded itself.

If a page says No competitors tracked yet, none of what follows will mean anything until you pick some.

Panels and controls#

The scoreboard#

Head to head sets your brand against one competitor, named at the top beside yours. It opens on the competitor with the most share of voice, and the menu there lists everyone you track in that order; picking another switches the whole page. The scoreboard then sets the two of you side by side: each brand's share of voice, a count of the measures led by each side, and a one-line verdict.

The measures come in two groups with two different windows. The visibility group covers the week shown, which is the latest week with finished analysis and not necessarily the current calendar week. The citations group counts every run and every engine, because citations accumulate slowly and a single week holds too few of them to compare.

Scoreboard

Share of voice

this week's mentions and citations, against every brand named

Mention rate

the share of prompt runs that name the brand

Average position

where the brand sits in answers that name it; lower is better

Topics led

topics where the brand is ahead

Prompts won

prompts where the brand is named more often

Engines led

engines where the brand is named more often

Citations

distinct citations across all runs

Earned citations

citations from sites the brand does not own

A measure where neither side is ahead reads Even, and so does any gap in share of voice or mention rate under half a percentage point, however large it looks as a ratio: 0.8% against 1.0% is two tenths of a point and not a lead. A position that does not exist, because a brand was never named, reads Not named rather than a zero. Most measure names are also controls: clicking one takes you to the part of the page that explains it, and Prompts won opens Prompts won and lost, every prompt of the week shown grouped into You won, the competitor's wins and No winner. Your citations (all runs) and Their citations (all runs) open the citations behind the totals.

The trend and By engine#

Head to head over time follows both brands week by week, on Share of voice, Mention rate or Avg. position. Weeks with no runs are left out, so the lines join the weeks that were actually measured rather than dropping to zero or breaking off. The shading between the two lines takes the colour of whichever brand is ahead in that stretch, and changes where the lines cross. On average position, a week where one brand ran but was not named has no position to plot, so a dashed line carries that brand across it. The line above the chart says who leads now and how that compares with an earlier week.

By engine shows each brand's mention rate on every engine that ran in the week shown. Engines with no runs are not listed. When the two rates are too close to call, the row reads No clear lead. The card appears only when at least two engines ran and AI Engine is set to All: with a single engine there is nothing to compare.

Topic battleground#

Topic battleground has one row per topic that ran in the week shown, with each brand's mention rate and average position. The bar points toward whoever is named more.

A topic's lead is decided in two steps. Whoever is named more often by over 5 points leads. Closer than that, whoever is placed at least half a place higher leads, as long as both were named in at least 3 runs; the chip then reads, for example, "placed 0.8 higher". Otherwise the topic has No clear lead. A topic with fewer than 3 runs shows Too few runs and appears only under All. The filters above the rows are All, You lead, the competitor's leads and No clear lead. Topics led on the scoreboard counts the same leads, and clicking either side's number there filters the battleground to those topics.

Opening a row shows the prompts that decided the topic and Citations for this topic (all runs). Show drilldown opens the topic in full, with tabs for Share of voice, Prompts, Engines and Citations. Engines is left out when only one engine ran the topic or a single engine is selected.

Where each side stands out#

The Where you stand out card and its mirror for the competitor list the biggest gaps in each direction: topics where one side is named far more often, engines (when more than one ran), sources that cite only one of you, and average position. Each item carries its size and a next step. A topic item offers Show drilldown, an engine or source item opens its citations, and a position item offers See the trend. View all lists every qualifying lead, grouped by kind. When nothing stands out, the card says No clear edge in the week shown.

Sources the engines cite#

Sources the engines cite starts with each brand's citation total. Where a brand's own site is known, the bar splits into citations earned from other sites and citations from the brand's own site, and the key under it counts each. Below that, domains fall into three columns: Only you, Both, and the sites that cite only the competitor. That last column is the practical one, since those sites have shown they cover your category and have never cited you.

Drilldowns#

Links on Head to head and Tracked competitors open Browse Citations, a prompt or Who's Citing in a new tab, with the filters already set. A link whose window differs from the panel it sits in says so, which is why most read "(all runs)".

Tracked competitors#

Where everyone stands lists your brand and every tracked competitor on the same measures: SOV rank, share of voice with its change from the previous week under it, in points, mention rate, average position and citations. The list is ordered by share of voice, and SOV rank is each company's place among every company named in AI answers that week, not only the ones you track, so a brand at the top of your list can still be #14 overall. The arrow under it counts the places gained or lost against the previous week. A tracked company nobody named that week shows a dash. A competitor's Head to head button opens the comparison with that competitor. Share of Voice Trends below it follows everyone's share week by week, once there are a few weeks of runs.

Until you save a list of your own, Seerly tracks the five competitors with the highest share of voice over the latest week, the top five of All Competitors, and works that five out afresh every time one of these pages is drawn, so the companies being measured can change from one week to the next without anyone asking. The first Competitor intel report published for your project saves the competitors it was generated for as your tracked list, in the order it used, so the same companies are measured from then on. It only ever does that when you have no list saved: a list you chose is never changed by a report.

Once there is a saved list, it is the list everywhere, on this page, on Head to head and in every competitor intelligence report, and it changes only when you change it. Reset to Default in Select Competitors clears it and hands the choice back to Seerly, which goes back to picking the top five by share of voice over the latest week until the next report saves a list again.

Who leads each topic is the whole field on one grid: a row per topic that ran in the week, a column per brand with yours first, and in each cell how often that brand was named on that topic's prompts. A competitor leading a topic has its cell filled, so a column reads as territory. Your own cell is coloured by how the topic went rather than by your rate: green where you lead it, red where a competitor does. Clicking a competitor's name opens Head to head with them.

Under each topic name is the verdict in words. You lead, or a competitor's name followed by "leads", means that brand leads every other brand on the topic, by the rule the Topic battleground uses: named in more than 5 points more of the runs, or, when that is close, placed at least half a position higher, as long as both were named in at least 3 runs. Contested means no brand beats all the others, which can happen even where you lead one of them. No brand named means the topic ran and none of the brands on this grid appeared, which says nothing about brands you do not track. Too few runs means fewer than three runs, which is not enough to call.

Topics are ordered worst first: the ones a competitor leads, then contested, then the ones you lead, then the ones with too few runs. The work is at the top. These rates are per topic and count that topic's prompt runs, so they do not average out to the Mention rate column above, which counts your runs across every topic.

Also known as, at the bottom of the page, holds the other names each tracked competitor goes by: a short form, a former name, the name of its parent company. Competitor intelligence keeps a third-party article about a competitor only when its title uses a name it knows for them, and it searches under these names too, so an article that calls a competitor by a name it doesn't know can be missed. It already picks up the names a company gives itself on its own site, so add the ones it would not find there.

The names Seerly already knows a competitor by, from its own site and from the answers that name it, come first under Found automatically. They can't be removed here, and adding one of them changes nothing: you are told it was found automatically. None yet means there are neither found names nor names of your own.

Type a name and press Add, or Enter, and it is saved straight away. The cross on a name removes it. A competitor can have up to 10 names, each 2 to 60 characters, and a name already on its list is not added twice. Every competitor on the list can have names, whether you chose it or Seerly did, and a competitor's names stay with it if it leaves the list and later comes back. Read-only members can see the names but not change them. Your own brand's other names are not set here: they are the Brand names in Brand Hub.

Competitor intelligence#

Competitor intelligence is the third page under Competitors in the sidebar, listed there as Competitor intel, alongside Head to head and Tracked competitors. Where those two compare measured visibility, this one reads like a weekly briefing: what each tracked competitor put out, on their own site and on third-party sites, with your own brand's activity carried in the same report so you can read the two side by side.

What's scanned, and how often#

Every week Seerly checks two things for each competitor you track: their own website, for anything new, and what other sites have published about them. Your own brand goes through the identical check and lands in the same report, marked with a YOU badge, so your week sits next to theirs rather than living on a separate screen.

What other sites published is looked for in your project's country. A project set to the United Arab Emirates asks about coverage in the United Arab Emirates, so a competitor that trades under the same name somewhere else doesn't bring that country's coverage with it. A project set to worldwide asks without naming a country.

Reports cover whole calendar weeks, Monday to Sunday in your project's time zone. Your first report covers the 4 complete weeks before the week it runs in. After that, a new report arrives each week on your project's set day, covering the Monday to Sunday just gone; the header shows when the next one is due. Anything published during the current week waits for next week's report, so no week is counted twice or cut short. A first scan usually finishes in under an hour. While a project is waiting on its first one, the page reads Scans run weekly. Your first one hasn't started yet., and once a scan is underway it shows Your first scan is running, with a bar that counts the checks done ("2 of 6 checked") and one row per competitor: its name, and its coverage line on the right. A row reads Waiting until its check starts and Checking…, with a spinner, while it runs, then shows how the check went (see Coverage) with a check mark, or an amber mark when we could only partly check it.

There is no need to reload while a scan runs: the page updates itself as each competitor's check finishes, and the bar fills with them. Once every check is done the heading reads Writing your first report, the bar reads full, and the report shows as soon as it's ready. A later scan works the same way. While it runs, the latest report carries A new scan is running, and when the new report is ready the page moves to it and says Your new report is ready. An earlier report you picked from the week picker stays on screen instead: the new report joins the week picker, and Your new report is ready comes with a View button that opens it.

When a report is ready, a message also says so on Telegram, if your project has Telegram connected (see Settings).

Your first report reaches back 4 weeks rather than one, so there is something to read straight away. Its four weeks show as four bars in each competitor's history, and each later report adds one bar of its own, every bar labelled with the Monday to Sunday it covers: "Sep 7–13", or "Aug 31 – Sep 6" when it spans a month end. A competitor you start tracking later begins with the week of the report it first appears in. The first report itself has nothing earlier to compare with, so its comparison reads First scan (4 weeks): not compared yet.

To find what a competitor published in those weeks, a first report reads their site section by section: their blog, their announcements, their case studies, and so on. A section counts whatever they call it. One named for its posts, such as "blog" or "news-and-insights", is read as it is. One with a name of their own is read once a few of its pages show that they are dated articles, and a part of their site whose pages carry no publish date, such as product or company profile pages, is left out. Posts published at the top of their site, under no section at all, are found the same way. A section with a feed that already reaches back before the report's first week is read from that feed. Otherwise we read the section's own page listing its posts, newest first, and stop once we're past the report's weeks, since everything further down is older. When that page also lists posts they file somewhere else on their site, those are opened first too. Only as a last resort, when a section has no such page or one that doesn't put its newest posts first, do we open its pages in the order their sitemap lists them, up to a limit on how many one check can open. That limit is shared between the sections read this way, so a large section never crowds out a small one, and a section whose pages are numbered is read from its highest number down, which is its newest. We also stop reading a section once a long run of its posts has shown no date at all (see Coverage). After the first report, each check also notes which posts are new on their site since the one before, so a post that shows no date is still counted, in the week it first appeared.

Each page is dated by what it says about itself: the publish date it declares, in whichever form their site writes one, and failing that the date shown beside its title. A date marked as the day the page was last updated or reviewed is not its publish date, and neither is a date their site prints on every page, so a page that offers only those counts as showing no date. A site that publishes the same article once for each country or language it serves has it counted once, in the edition for your project's country when there is one, and otherwise the English one.

Numbers on a report show its own week, and the bars in Recent weeks carry the weeks before it. The week picker sits at the top of the page, labelled Week and marked with a calendar icon, right beside the Viewing picker (see Focus mode); when the report was scanned and when the next scan is due sit to their right, and the line naming the competitors being watched, with Edit list, sits just below. A first report reaches back 4 weeks, so the week picker lists it week by week: its four weeks, newest first, from "Sep 21–27, 2026" back to "Aug 31 – Sep 6, 2026", then the whole first scan, "Aug 31 – Sep 27, 2026 · all 4 weeks". Each later report is one entry of its own, above those, and every entry carries its year. A first report opens on all 4 weeks; each later report opens on its own week. A later report that covers more than one week, after a gap, lists the same way, week by week and then all of them together, and opens on its latest week. Every week is dated in your project's own time zone. The week picked sets what the whole report shows: the headline tiles, whose subtitles say which week they count, every column of Activity by competitor, the week's own read, Needs your attention, the feed with its chips, and in focus mode the profile and the day-by-day timeline. The "all 4 weeks" entry shows the totals for the whole first scan, the read of all four weeks and its full feed. The week stays in the page address, so a reload or a shared link keeps it. The bars in Recent weeks keep showing the history, with the picked week's bar drawn in full colour. Focus on a competitor on its own first scan in a later report and the numbers show its latest week, as its row in Activity by competitor does.

An update counts in every week it published in. A run of look-alike pages with pages in two of the weeks counts as an update in each of them, while "all 4 weeks" counts it once, so the four weeks can add up to more than the total. Hovering the split under Unique updates, or the info icon on the Own site heading in the activity table, says so.

The headline numbers#

Up to four tiles sit at the top of a report.

Headline numbers

Unique updates

everything new in the report's week, or on a first report the week picked, each counted once and split into own site and third-party

On your tracked topics

how many of those updates cover a topic you track, discovery or sentiment

Coverage on sites AI cites

stories about them on third-party sites that already appear as sources in your AI answers

Updates naming you

updates about a competitor that also mention your brand, shown only when there is at least one

A unique update is one post, or one run of look-alike pages (a series, or a batch published together), counted once; a story on third-party sites counts once however many sites ran it. The info icon beside Unique updates says so.

Unfocused, the first three tiles also carry a You: line underneath, so your own figure for the same measure sits right there instead of requiring a separate lookup. Focus on your own brand (see Focus mode, below) and the tiles switch to your own frozen numbers for the week, compared against your own recent average.

Coverage on sites AI cites is not a count of your articles being cited. It counts what others wrote about your competitors, published on sites that the AI engines already cite when they answer your prompts: coverage that is well placed to shape those answers. When there is any, the tile names those sites, up to three and then how many more, and clicking it narrows the feed below (or, in focus mode, the day-by-day timeline) to exactly those stories: the On AI-cited sites chip. When there is none it says so: "None of their coverage reached sites your AI answers cite." Its You line is the same measure for your own brand, stories about you on those sites.

On your tracked topics counts the updates that cover at least one topic you track, of either kind. Discovery topics are subjects people search for; sentiment topics are about how your brand is perceived, such as customer reviews, pricing and value, or service quality. A competitor's update counts toward a sentiment topic when it covers that subject, even though the topic is named after your brand: their stories about customer results sit on a topic like customer reviews of your programs. A passing mention isn't enough. Under the count, the tile splits it into discovery and sentiment. An update on both kinds counts once in the total but in both halves of the split, so the two can add up to more than the total, and hovering the split says so. Clicking the tile narrows the feed (or, in focus mode, the day-by-day timeline) to exactly those updates: the On your topics chip, which always carries the same count as the tile.

Updates naming you only appears when at least one update in view mentions your brand. Without one, the tile is left out and the other three share the row.

Activity by competitor#

Right under the tiles, Activity by competitor is one row per tracked competitor, with your own brand pinned to the top. Competitors are ranked by the week you picked: most updates first (own site and third-party together), then more third-party stories, then more own-site updates, then more activity across recent weeks. A competitor we couldn't check comes after one we could.

Activity by competitor

Competitor

the tracked competitor, or your own brand labelled You

Updates

under Own site: what they published on their own website in the report's week, or on a first report the week picked, each post or run of look-alike pages counted once

Pages

under Own site: the pages behind those updates, every page of a run included

Stories

under Third-party: new stories about them on other sites in the same week, each counted once however many sites ran it

Articles

under Third-party: the articles behind those stories, one for every site that ran one

Publishers

under Third-party: how many different sites those articles came from, each site counted once however many articles it ran

Recent weeks

a small bar per week, newest on the right, for up to eight recent weeks, with a trend line above the tops

Change

the week's count against the week before, as a percentage

vs average

how the report compares with their own recent average

Main themes

the topics their new updates cluster around, in the week shown

Own site is two columns under one heading, Updates and Pages, and Third-party beside it is three more, Stories, Articles and Publishers. All five count the week the picker shows, as their headings say: the report's week, or on a first report the week picked, or all four together. Updates counts each post once and a run of near-identical pages once, and Pages counts every page behind them, so a run of 300 pages reads 1 update and 300 pages. Stories counts each story once however many sites ran it, and Articles counts every article behind them, so a story five sites ran reads 1 story and 5 articles. Publishers counts the different sites those articles came from, each site once however many of its articles are in the week, and two addresses of one publication count as one site. The info icon on each heading says what it counts. A competitor we couldn't check on one side reads Couldn't check once across the whole of that side. In a week of a first report, the pages and articles are the ones published that week.

Publishers is counted afresh for whichever week, or set of weeks, you are looking at, so it is never the weekly figures added together: one site that ran an article in each of two weeks is one publisher in each of those weeks and still one publisher across the pair.

Each number opens a side panel of what it counts. The panel's heading names the row, the side, the week and the same number the cell shows, for example "Acme · own site · all 4 weeks · 35 updates". Under Updates it lists each update with its date, and a run of look-alike pages with its page count, which opens that run's pages. Under Pages it lists every page: a single page as it is, and each run as a group carrying its count, which opens to page through its pages 20 at a time. Under Stories it lists each story with the sites that ran it and its number of articles, and under Articles every article, grouped under its story. Under Publishers it lists each site once with how many of its articles are in the week, most first, and opening one shows those articles. A 0, or Couldn't check, opens nothing.

An opened row lists that same week's updates in the same table as the feed below, under a line counting its runs of look-alike pages, for example "Own site: 2 updates made of 47 look-alike pages, plus 3 single pages.", with a link to show the updates from its other weeks. It lists them 10 rows at a time, with the same page controls as the feed under its table on the left, reading "Page 1 of 4". Picking another week, or showing the other weeks, goes back to the first page. While the next page loads, the rows already there stay on screen, faded, and if a page can't be loaded the row says so, with Try again. Open profile, under the list, focuses on that competitor, or on you from your own row. Main themes are that week's own themes on a first report, or those of all four weeks together, and a dash means no theme stood out for them that week.

In Recent weeks, each bar is one week, and hovering it shows the week it covers and how many updates were found. The week the numbers show is drawn in full colour and the others fainter: the newest, or on a first report the week picked, or all four for "all 4 weeks". A week with nothing in it is a faint mark on the baseline rather than a bar, and a dashed outline marks a week that couldn't be checked, which is not the same as a week with nothing in it. Only real weeks are drawn: a competitor with less history than the others has empty space before its first week, and every row's newest week lines up down the column. A smooth line runs just above the bars, oldest to newest, green where a week rose, red where it fell and grey where it held, so the direction reads at a glance, and the dot at its end takes the colour of the latest week against the one before. It breaks at a week that couldn't be checked rather than guessing across it. How often a competitor comes up in your AI answers isn't a column here; it appears once, on that competitor's profile (see Focus mode, below).

Change compares the week shown with the one before it. A rise reads as a green percentage, such as +25%, a fall as a red one, such as −40%, and a week that held level as 0%. When the week before had nothing at all, there is no percentage to give, so a rise from zero reads New. On a first report, which reaches back 4 weeks, it is the week picked against the week before it, the fourth against the third to begin with; after that, each new week against the last. "All 4 weeks" is no one week, so the column is left out there. A dash means there is nothing fair to compare, and hovering it says why: No earlier week to compare for a first report's first week, Not checked that week when either week couldn't be checked, or Not enough weeks yet for someone with only one. Change and vs average answer different questions: Change is one week against the week before, while vs average sets the report against their own recent run, so a competitor can be down on last week and still above their usual pace.

This week's read and signals#

Below the activity table (or straight under the tiles in focus mode, which has no table) is a short set of bullets the agent wrote about what happened: This week's read on a weekly report. A first report has a read for each of its four weeks and one for all four together. A week's read is titled with its dates, such as "Sep 21–27, 2026", and opens with a line summing the week up; a quiet week can have that line alone. "All 4 weeks" shows the read of the whole month, These 4 weeks. It only appears once there is something to say, and in focus mode its title changes to name whoever is focused instead, with the read written about them for the week shown. Focused on a competitor that had nothing in the week picked and no read for it, the card says so in one line, such as "Quiet this week: nothing new from Acme.", or "Quiet in Sep 7–13: nothing new from Acme." for an earlier week. On a wide screen it sits side by side with Needs your attention, each taking half the width; on a narrower one the two stack, and whichever appears on its own takes the full width.

The read aims to be complete rather than long. It has up to five bullets, grouped by competitor, and covers every major move of the period, whether it appeared on the competitor's own site or on a third-party site: funding, acquisitions and mergers, launches (a product, a service, or a new location or centre), pricing changes, and large pushes on your tracked topics. Minor updates are left out, and nothing is said that the updates themselves don't support. Each bullet is a sentence or two, with its key facts in bold: an amount raised, the name of a launch, or a count of pages.

Under each bullet, on a line of their own, come its sources: small pills showing the site's icon and the publisher's name (or a short title), one for each site the bullet draws on. Selecting a pill opens that article in a new tab. When a bullet rests on several updates from the same site, they share one pill with the count, such as kolors.example · 3, and selecting it lists those updates to open. A bullet that only notes which competitors were quiet has no pills. Focused on one company, the read is written the same way, with bold facts and pills, though a bullet comparing your overall numbers with a competitor's may have none. Reports published before sources were added show their bullets without pills.

Needs your attention lists anything worth a look in the report, each tagged by kind, for example Names you or Launch or pricing. On a first report it follows the week picked, listing the items from updates that published that week, or reading Nothing needed your attention this week. when there were none. That line names the picked week itself, so an earlier week reads, for example, Nothing needed your attention in Sep 7–13.; "all 4 weeks" lists them all. Every item carries a Dismiss button; dismissing collapses the row to a one-line note with an Undo link, so clearing the list is reversible. Where there is a useful next step it appears as its own button, such as Create response content or See in PR suggestions, alongside View page.

Create response content opens a Content Creator chat with its first message already written from the report: what the competitor published (the title, whether on its own site or on third-party sites, how many pages, the dates, the publishers and up to three sample titles with their links), why the signal was flagged, which of your tracked topics it touches, and whether it names you. The message asks for a channel, a content type and two or three title options for your response, and from there the chat runs its usual course: it checks what you have already published, picks a focus keyword, settles the title and shows you the brief before anything is written. The response is built from your own expertise and services. It doesn't copy the competitor's headline or structure, and it leaves the competitor unnamed and unlinked unless you ask otherwise.

When what the competitor published is customer stories (testimonials, transformation stories or success stories), the button reads Collect customer stories instead, because writing stories like those for you would mean inventing customers and results. It adds a Collect customer stories action to Tracked Actions and opens it there, where you can draft a kit for asking your own customers for their stories: see Tracked actions. Selecting it again for the same signal opens the same action rather than adding another.

Bulk content push marks a batch of at least 50 look-alike pages that a competitor published on its own site in one short burst, with a headline such as "Acme published 120 look-alike Blog pages within 16 minutes on Sep 7." It carries a neutral grey tag, and a report has at most one per competitor and two in all. Under the headline, a short note says why a batch like this can matter; for location or service pages, it says instead that such page sets are a common SEO pattern. A row of chips then gives the evidence as measured: how many pages, how long they took, the share that follow one title template and, when enough of the pages were checked, their median length and the share that are near-duplicates of another page in the batch (shown only at half or more). When your AI answers were measured, a last chip says how many of the pages they cite, even when none do. The signal states what was measured and never judges why the pages went up: a batch can be a site move, a new section or a backlog of real stories. Selecting the headline lists the batch's pages in a side panel, for the week picked. Its button follows the rule above: Collect customer stories for a batch of customer stories, and Create response content for any other, with the evidence written into the chat's first message.

The feed#

Below the read and signals, Everything your competitors published lists the same activity as a table, newest first, and never your own brand's updates: those sit under the You row and You focus instead. Chips above it narrow the list to Own site, Third-party, On your topics or On AI-cited sites, each carrying its own count, and All puts everything back. Beside the chips, a competitor filter set to All competitors narrows the list to one competitor you track. The list comes 20 rows at a time, with the page controls under the table on the left, reading "Page 1 of 3". It is a page count, never a row count: the list takes in the Not counted rows and a story listed under two competitors, so a row count would run higher than the chip's own count of updates just above it and the two would disagree. Picking a chip, another week or another competitor goes back to the first page. While the next page loads, the rows already there stay on screen, faded, and if a page can't be loaded the list says so, with Try again.

Everything your competitors published

Date

when the update came out; for a run of pages, the newest of them in the week shown

Update

the title alone, or a story's headline, with any tag such as Names you or Not counted under it

Source

Own site, or the sites that ran a story: the first named and how many more

Pages

the pages of a run or the articles of a story in the week shown, the number itself opening the list; 1 for a single page

Type

what kind of update it is, such as a blog post, a comparison or a press release

Contributes to tracked topic

the topics you track that the update covers

In Contributes to tracked topic, each topic is a pill carrying its name in capitals and a small DISCOVERY or SENTIMENT tag, and the two kinds are coloured differently so they read apart at a glance; a long name wraps onto a second line rather than being cut off. A row shows up to two pills, then how many more, such as +2; hovering that count lists the rest. The rows with pills are exactly the updates the On your topics chip and the headline tile count. These are the topic names the report was published with, not your currently tracked topics: renaming or retiring a topic afterwards does not change how an already-published report reads. The day-by-day timeline and an opened row in the activity table show the same pills at the same size, in the same table.

Most rows are a single dated update from a competitor's own site, reading Own site under Source and 1 under Pages. A cluster of near-identical own-site pages, a series or a batch published together, collapses into one row with its page count under Pages; hovering its date shows the days it spans. Its titles never stack under its own: clicking its page count opens the whole list in a side panel, 20 pages at a time, newest first, each title opening the page itself, with a search box to find one by title. In a week of a first report, a run that published in several weeks shows only that week's pages: its date is the newest of them, its count reads that week's out of its total, such as "5 of 12", and the side panel lists that week's pages, headed with both counts, such as "5 pages this week · 12 in total" for the latest week, or "5 pages in Sep 7–13 · 12 in total" for an earlier one. A story on third-party sites shows as one row: its headline under Update, the sites that ran it under Source, the first named and how many more, such as "Livemint +2", and, when it also sits under another competitor, also under and their name. Its article count sits under Pages, and clicking it lists the articles in a row right under it, each with its site and date. In a week of a first report, a story that ran in several weeks is dated by its newest article that week, and counts and lists only the articles filed under that week, with their sites, reading, for example, "2 of 5". A long title is cut at two lines and a long list of sites at one; hovering either shows the whole of it. The same rows appear under an opened row in the activity table and in the day-by-day timeline. An update where a competitor cites your brand, rather than the other way round, is tagged Brand as a source and Not counted under its title: it stays in the list, after the updates that do count, but out of every count on the page. When one of a row's own pages is cited in your AI answers during the report, the row carries a violet tag saying how many, for example "Cited in 3 of your AI answers"; rows no answer cites carry no tag.

An update dated in a week an earlier report already covered, but found only after that report came out, is counted in the report that found it rather than changing the published one. Its row shows its real date with found late under it. A page of a run found late reads found late under its date in the side panel, and an article of a story after its date in the story's list.

Coverage#

Not every competitor gets a full read every week. A short line under each name, in the activity table, on the focus profile and in the first-scan list, says how much of them was actually checked. It reads the same way everywhere: the status first, then, after a dot, a short reason when there is one, for example "Checked by search · their sitemap is out of date".

On the focus profile, a second line under it covers what other sites published about them, when that check missed something. It starts with "Third-party:", such as "Third-party: Partly checked · 1 of 12 third-party articles couldn't be checked this time". When nothing was missed there, the profile shows no second line.

Coverage

Checked

the full check went through

Checked by search

their content was found through search rather than read directly from their site

Partly checked

some but not all of what they may have published was seen

Couldn't check

no read at all this time; the profile falls back to the last report where they were checked

The reason after the dot says why. The ones you are most likely to see:

• their site blocks automated checks: their site refused our checks outright or put up a bot challenge. When search still found their content, the line reads Checked by search with this reason.
• we couldn't reach their site: their site didn't answer at all, even after we retried. Search can still cover them the same way.
• no new posts since a date, such as "no new posts since Feb 4, 2026": we read their site, and their newest post, from that date, is older than the weeks this report covers, so they published nothing new there. Without a date it reads no new posts in these weeks.
• their sitemap is out of date: their sitemap hasn't been updated recently, so newer pages may be missing from it.
• the rest next time, such as "dated 300 of 338 pages, the rest next time": their own site had more pages to date than one check can open, so we dated as many as it could and the rest are checked first next time. Without those numbers it reads not every page could be checked this time: there were more pages than one check can open, or a site slowed us down, so some weren't read. Either way it only ever means our own limit, never a stop we made on purpose: when we stop reading a section because we're past the report's weeks, or because its posts show no dates, the line doesn't say this.
• their posts don't show a publish date: their posts, or those in one section of their site, carry no publish date anywhere, so the first report that checks them can't count their earlier weeks there. A post that search dates inside the weeks still counts. From the next week on, every new post on their site is counted by when it first appeared, so the line doesn't come back. When their posts came from a different host, the line names it, such as "Partly checked · northwind.example · their posts don't show a publish date".
• couldn't be checked this time, after a count of third-party articles, such as "1 of 12 third-party articles couldn't be checked this time", in the first-scan list and on the profile's third-party line: we open each article about them that we find on other sites, to confirm its date and that it's about them, and there were more than one check could get through. Without those numbers it reads "not every third-party article could be checked this time".
• using an earlier check: this report's check didn't finish in time, so the last completed check is shown instead.

Not everyone publishes on their main site. Some keep their posts on a blog subdomain, such as blog.example.com, or under another section of their site, such as example.com/about/blog/. We find those from the links on their own homepage and read their posts where they publish them. Only a link to where they publish counts, one labelled "Blog", "Newsroom" or "Case studies", for example, and never a button such as "Learn more" or "Read more". When the posts came from a different host than their main site, the line names it between the status and the reason, such as "Checked · blog.example.com", or "Checked · blog.example.com · no new posts since Feb 4, 2026". A blog under a section of their main site is read the same way, without a name on the line.

A site that has moved is followed. When a competitor's homepage now always sends visitors on to a different domain, as after a rebrand, we read the site it leads to as theirs, and the line names that site, such as "Checked · northwind.example", or "Checked by search · northwind.example" when search had to fill in. Their old address still counts as theirs: a sitemap that still lists pages under the old domain is read, each page at the same path on the new one. A redirect that's only temporary, or one to a sign-in page, isn't followed.

Some sites show no publish date anywhere: not in their sitemap, not in a feed, not on the post itself. A post like that is still counted when search dates it inside the weeks shown. From a competitor's second week on, a post that appeared on their site since the previous check is also counted, dated to the middle of the time between the two checks. That doesn't happen on a first report, which has no earlier check to compare with, or when far more pages appear at once than that site usually adds in a week, as when a site adds its archive: those are never dated by when they appeared. Each is opened instead, and counted only when its own page shows a date inside the weeks shown.

When the site it leads to belongs to another competitor you track, their posts there are that competitor's, so they're counted once, under that competitor, and the line says so:

• their site is part of a competitor you already track, such as "their site is part of Northwind Clinic, which you track". Without the name it reads their site is part of another company you track. Both entries now lead to the same place, so you may want to stop tracking the one that moved, with Manage tracked competitors on the Tracked competitors page (see Choosing who is tracked).

A report from before we followed these moves may still read their site now redirects to a domain, such as "Checked by search · their site now redirects to northwind.example", or their site now redirects to another domain without it: that site wasn't read then, and search covered them instead.

On your own row, the reason says "your" rather than "their": "your site blocks automated checks", or "your posts don't show a publish date".

This matters for the numbers above it. A headline tile reading a dash, or a table cell reading Couldn't check, rather than zero means the competitor was not checked in this report, not that nothing happened, and every total on the report only counts competitors that were.

Focus mode#

The Viewing picker at the top of the report, beside the week picker, sets whose activity you are reading. Open it and pick from All competitors, which is the full report, then your own brand with its YOU badge, then each competitor in the report. Every entry carries a count of its updates, for the same period as the headline tiles, or reads not checked for a competitor this report couldn't check, and typing in the search box finds one by name.

Whichever way you focus, from this picker, from Open profile under an opened row in the activity table, or from a competitor's name in the feed or in Needs your attention, the page scrolls back up to the Viewing picker so you see the choice change; Back to all competitors does the same. With your device set to reduce motion, it moves there at once instead of scrolling. Only a click of yours scrolls: a link that already names a competitor, or your browser's back and forward buttons, change the focus without it.

Focusing replaces the activity table with a profile card, above the read and signals, so its numbers still come before the commentary on them, with a bar chart of up to eight recent weeks (each bar labelled by the week it covers), a line on how often the company comes up in your AI answers, such as "Acme is mentioned in 27 of your AI answers. acme.com is cited as a source in 0. Your site is cited in 8 of your AI answers.", and a Back to all competitors button.

That line says two different things, which is why it says both. A mention means the answer named the company. A citation means the answer used their website as a source. A company can be named in answer after answer while its own site is never cited, because the answers lean on write-ups about it elsewhere, so a citation count of 0 next to a healthy mention count is normal rather than a sign that they never come up. If their site has permanently moved and we followed the move, both the old and the new domain count as theirs, and the line names the site we actually read, the same one the coverage line shows. That line, like the You in your AI answers tile when you focus on your own brand, comes from the latest check of your AI answers: the answers to your discovery prompts across every completed onboarding and analysis run, taken when the report was built rather than week by week, so it stays the same whichever week you pick. The info icon beside it says so. Below the read and signals, it also swaps the feed for a day-by-day timeline, titled You, day by day for your own brand or the competitor's name followed by day by day otherwise, so a busy stretch reads in order rather than as one flattened list. It is the same table as the feed, the same Date, Update, Source, Pages, Type and Contributes to tracked topic columns, with a day label spanning the row between each day's updates, and no competitor named above each source, since every row already belongs to the one entity in focus. Only days with something published are listed, and the timeline carries the same filter chips as the feed, but no competitor filter, since it is already one competitor's or your own. In a week of a first report, an update that ran in several weeks sits on the day of its newest page that week, and when none of its pages that week carries a date it sits under "Found this week, date unknown", or "Found in Sep 7–13, date unknown" for an earlier week, instead.

Report history#

A published report always shows the competitors it was generated for, even after you change who you track. Add or drop a competitor and the report you are already looking at does not change retroactively; a banner on the latest report says so, names what was added and removed, and explains when the new list takes effect, whether that is the scan already running, the one right after it, or the next scheduled scan. Edit list takes you to the Tracked competitors page, where the tracked list is managed, and Got it dismisses the notice. The Edit list beside the tracked count at the top of every report goes to the same place, as does Track another when nothing is tracked yet.

A dropdown at the top of the page steps back through earlier reports, and through a first report one week at a time, each one still showing the list it was generated against.

A competitor you add joins the next report that starts, and its earlier weeks are filled in rather than left blank. The first scan of a company reads about four weeks back, so the weeks between where that reading starts and the report's own week are counted from what it found, on their own site and on third-party sites alike, by the same rules the report counts its own week by. Those bars are drawn like any other, and hovering one says it was filled in when that competitor was added. Weeks before the first scan could reach stay empty, and a competitor you dropped and later added back keeps the weeks the reports that tracked it measured. The filling in is history only: it never changes the report's own counts, its headline numbers, Needs your attention or its read, and it never changes a report already published.

Availability#

Competitor intelligence is an add-on, available on request rather than included on every plan. The page is always listed in the sidebar, with a padlock beside it until it is active for your project. Open it and, if it is not active yet, you will see a faded Example report behind a card headed See what your competitors publish, every week. The example is made up, with invented companies, to show the shape of a real first report, on all 4 weeks; it is illustrative only, not an interactive report, so its tiles, feed and drilldowns cannot be clicked or filtered.

While it is not active and you have not requested it yet, Head to head and Tracked competitors show a small banner above their content: See what your competitors publish, every week, with Request access opening the Competitor intelligence page, where you can see the example report and send the request. Once access has been requested the banner no longer shows. Close it with the × and it does not come back: the choice is saved to your account, on every project and device. Head to head and Tracked competitors themselves are included on every plan.

The card sums up what the page covers in three short lines, then says who it would watch: how many competitors you already track, with up to five of them as logo chips (six when that is all of them) and how many more, and a reminder that the report covers exactly this list. Change beside them takes you to the Tracked competitors page, where you manage that list and can correct how each one is classified (see Correcting how a competitor is classified, above). With nothing tracked yet, the same place reads Add competitors and takes you to that page.

Request access sends the request off; an admin reviews it and turns Competitor intelligence on for your project, at which point it starts running like any other, with a first report that covers the 4 complete weeks before the week it runs in, so there is a month of competitor moves to read from the start. While a request is waiting, the card confirms it has been received.

Interpreting results#

Read this per topic and not as a single verdict. An overall deficit is almost never spread evenly, and what the Topic battleground usually shows underneath it is a handful of topics where you compete perfectly well and a handful where you simply are not there. The second group is the work.

Watch for concentration while you are in there. When one competitor seems to own a topic, that ownership usually rests on a small number of pages that keep getting cited, not on an enormous content operation you could never match. A small number of pages is a beatable thing.

It is also worth checking whether either of you is winning. Look down the column of sites that cite only the competitor: when a review site or a directory sits there, getting yourself listed is a faster route than trying to out-publish anybody.

Finally, names you do not recognise are useful information rather than noise. If brands you have never heard of are dominating your topics, either the topics have drifted away from what you sell or there is a competitor you have not been paying attention to, and you would want to know about either one.

Best practices#

• Strength where you actually sell. Being genuinely present in the topics that produce revenue beats being mediocre across all of them.
• Share of voice moving your way over several runs, which tells you far more than any single reading.
• A competitor list you recognise, revisited every so often rather than set once and forgotten.

Next steps#

1. Confirm the tracked list, since everything downstream depends on it.
2. Find the topics a competitor leads in the Topic battleground and open them up.
3. Start from where the competitor stands out. Those items are the largest gaps, and each one comes with its next step.
4. Check whether the winner is even a competitor. If the sites citing only them are directories or review sites, getting listed is the cheaper move by a distance.
5. Follow change rather than position. Share moving steadily toward you across several runs is the sign the work is landing.

What to keep in mind#

Everything here is measured against the competitors you selected, so the picture is only as good as that list. A rival nobody added is invisible, and looks exactly like a rival who is not competing with you.

The panels do not all cover the same window. Visibility figures cover the week shown, while citations count all runs, so a citation lead can come from months ago. A topic's mention rate is measured within that topic, so the topic rates do not average out to the overall mention rate.