Competitors

Competitors at /competitors is the record of which brands AI names alongside you. Trakkr reads the answers your prompts produce, pulls out every brand mentioned, and ranks the field by how often each one appears and where it lands.

Open it when a visibility number moves and you want to know who moved with it, or when the question is not "how am I doing" but "which questions am I losing, and to whom". Expect at least one name you did not put there. In SEO you pick your competitive set. Here the models pick it, from whatever they think fits the question, so the list follows your prompts rather than your strategy deck.

Competitors needs a paid plan. Without one the page shows a teaser instead of the leaderboard.

How the list is built

Your starting list comes from the scan Trakkr runs when you first set the brand up, and you can drop anything wrong before you finish onboarding. After that it is maintained from answers.

Trakkr extracts brands from your tracked prompt results across the available models. A name has to repeat before Trakkr tracks it: at least two mentions, spread across at least two different prompts and at least two different models. That gate is what keeps a single stray mention out of your rankings. The list holds up to 500 names, on every plan.

Trakkr will not let your own brand onto the list. A name that is already one of your brand's aliases, or a product or marque that belongs to you, is refused rather than scored as a rival. If a name you own is still on the leaderboard, add it as an alias under Groups → Your Brand and it stops competing with you.

There is no add button, and the Agent will not add one for you. A hand-added competitor would have no responses behind it, no prompts to compare on, and no head-to-head record, so the row would sit empty. If a competitor you care about is missing, that is a finding rather than a bug: the models are not naming them on the questions you ask. The lever is prompts. Add the comparison and category questions you would expect them to surface on. If they still do not appear, they are not in this competitive set, which is worth knowing on its own.

The three views

The header carries five numbers for the selected window: Your rank out of the tracked field, Visibility, Win rate, Trend, and Watching (how many competitors currently read as threats). The date range picker offers 7, 14, 30, and 90 days, with a compare toggle for holding two windows side by side, and an export.

Above the rows, a chart toggles between Trend, which plots everyone's visibility over time, and Share, which turns the same run into share of voice. Below it sit the quick slices, a search box, and filters for tags and model.

Competitors

The default view is a ranked list of you and every tracked competitor, sorted by visibility on your prompts.

The quick slices are All, Threats, Rising, and Trailing. A competitor reads as a threat when Trakkr flags them at high or critical severity, or when they beat you on more than half again as many prompts as you beat them and there are at least ten head-to-head results behind it. Rising and trailing are pure movement: more than five points of visibility gained or lost over the window.

Prompts

Flip to Prompts and each row becomes one of your tracked questions, with AI Vol, your Position, how many Models answered it, a 14d sparkline, the Leader, the Gap, and the Trend. The quick slices change to All, #1, Top 3, and Losing, and a category filter appears.

This is the view for chasing one competitor. Sort by Leader and every question a single brand is winning groups together.

Matrix

Matrix is a single grid: every tracked brand down the side, every model across the top, the visibility score in each cell. Use it when you suspect a competitor's lead is concentrated in one model rather than spread across the field. A brand that buries you on ChatGPT is often invisible on Claude.

Reading a row

Click any competitor and the row expands in place: Visibility vs You, the Head-to-Head record, a Trend line against yours, By Model, and, if the brand tracks more than one market, By Market. View full profile opens the competitor's own page with the same picture at full size plus two lists, the prompts where they win and the prompts where you win.

Expanding a prompt row gives you the full Leaderboard for that question, the per-model evidence, and the AI Response itself, with a model picker and every brand shown in the order it was named. That is the fastest way to see what actually won. From the footer you can open the leader's profile, or send the question to Site Optimization.

You do not have to win every model. The question is where your buyers actually search, and what the picture looks like there.

Grouping, hiding, and your own names

The Groups button opens the Manage brands drawer, split into Your Brand and Competitors.

Models do not write a brand the same way twice. Brooks, Brooks Running, and Brooks Ghost are one company, and left alone they show as three rows with split scores. The Competitors panel collects them into groups. Trakkr auto-detects the obvious ones and applies them, marking each as new so you can undo them; Auto-group matches runs the detection again over anything still ungrouped. You can also build a group by hand, move a member between groups, or dissolve one. A group needs two members, so removing the second one deletes it.

Hiding is for a name that does not belong to your market at all: a different industry, a person, a scrap of text the extractor misread. Hiding takes the name off the tracked list and remembers the choice, so later runs do not put it back. Hidden names sit at the bottom of the drawer and unhiding restores them. You cannot hide your own brand, and you cannot unhide a name that is one of your brand's aliases.

The Your Brand panel is the other half of the same job: the variants that should count as your visibility rather than someone else's.

When you are losing

Losing is information, and it is usually one of three things.

You are missing from the sources. They are cited on the pages AI trusts and you are not. That work belongs in Citations: find the domains behind their wins and work down the list.

The prompt favors them by design. A question about open-source alternatives will not list a closed-source product however hard you push. If a competitor owns the use case a prompt is about, the better move is defending your own territory. Filter to your wins and check they are safe.

The models have not caught up. If you are new or you repositioned recently, the fix is slower: getting written about, listed, and reviewed. Research shows where you are invisible against where you are recommended but buried, so you can pick an order.

Whichever bucket it falls in, the loop is the same. Open the prompt, read the answer that won, check who is cited, then send the gap to Actions so it does not get lost.

Plan access

Competitor analysis is included with every paid plan. Growth and Scale get the same page, the same depth, and the same 500-name ceiling; there is no per-plan cap on how many competitors you can track. Client portal users see the page read-only when their agency allows it, without the grouping and hiding controls.

Common questions

Why is a brand I have never heard of on my list?

Because the models named it in answers to your prompts, more than once, on more than one question and more than one model. That is the whole criterion. If it is genuinely irrelevant, hide it from Manage brands and the leaderboard cleans up. If it is relevant but surprising, it is usually the most valuable thing the page has told you, so go and read the prompts it is winning.

A competitor I care about is not showing up. How do I add them?

Not from this page. Their absence means the models are not recommending them on the questions you track. Add prompts they would plausibly appear on, such as comparison questions or category questions tilted to their strength, and check again after the next run.

Why do I have three rows for what looks like one competitor?

Name variants. Open Groups, go to the Competitors panel, and either keep the auto-detected group or build one yourself. Grouping consolidates the visibility into a single row.

Visibility here does not match my dashboard score. Why?

The dashboard covers your full prompt set across all models on its own window. The leaderboard recalculates for whatever date range, model filter, and tag scope you have active. Clear the filters and the numbers line up. The metrics reference has the definitions.

How fresh is the data?

It is built from your daily runs, so new prompts and new mentions take a day to land. If you have just changed your prompt set, give it a day before reading much into the movement.

Does tracking a competitor cost a prompt slot?

No. Prompts are the only thing your plan counts. Competitors come out of the answers those prompts already produce.