Brand report

Overview#

A brand report packages your visibility, sentiment, citation and competitor position into a narrative document you can hand to people who do not work in Seerly every day: a leadership team, a client, a board.

It exists because the dashboard is built for somebody working the problem daily, and that is a different audience from somebody who needs the position and the direction inside five minutes and will not be back for a month.

Reports are written for you on a schedule. Nobody has to remember to ask for one.

Turning reports on#

Automatic reports are off until you switch them on, per project. On the Brand Reports page, the Automatic reports card carries the switch.

Once it is on, a new report is written each month on your report date, and the card tells you when the next one is due.

What happens the moment you switch it on depends on how long the project has been running:

• If the project is at least 28 days old, your first report is written immediately, and the date you switched it on becomes your report date.
• If it is newer than that, there is not yet a full window to report on. The card shows the date your first report will arrive: your report date, on the first occurrence that falls after the project turns 28 days old. Switching reports on early is fine, and the date does not move.

Switching reports back on later keeps the date you already had, so a date you have come to expect does not move on its own.

Only the project owner can change this. Project members can see the schedule but not change it.

Choosing your report date#

The card also carries the day of the month your reports are written. Any day from 1 to 28 can be chosen, so your date exists in every month including February.

Changing the date does not trigger a report. The new date takes effect from the next time it comes round.

Turning reports off#

Switching Automatic reports off stops new reports being written. Your existing reports stay available. Nothing is deleted, and you can still open, read and filter every report you already have.

What a report covers#

Each report covers the 28 days up to its analysis date, measured back from your most recent analysis run.

This is worth reading carefully, because it is not the same as a calendar month:

• The window ends at your last analysis run, not on the last day of the month. That anchor almost never lands on a month boundary.
• Because of that, consecutive reports are not perfectly contiguous. Roughly two days a month fall outside both windows. Reports are a series of samples of your position, not an unbroken ledger of every day.

Every report states its own basis on the cover, in a line reading something like The 28 days to 31 August, across 24 analysis runs, 1,860 AI answers, followed by the engines and market it covers. Read that first: a report over a window with few runs behind it supports considerably weaker conclusions than one drawn over many, and this is where you find out which of the two you are holding.

Where nothing has run since the period ended, the report says so rather than letting the figures read as current.

Competitor sections depend on your tracked list. If that list is empty or wrong, the comparison is limited accordingly, and the report says so rather than quietly leaving the section out.

Reading and downloading a report#

Opening a report shows you the finished document itself, the same file you would send to somebody else. What is on your screen and what arrives in their inbox are one thing rather than two, so there is nothing to check between them.

The document is produced by Seerly rather than by your browser, which is what makes that true: everyone opening the same report sees the same margins, the same cover and the same page breaks. You can page through it, search it, print it and save it using the controls your browser puts around it, and Download PDF at the top of the page saves a copy directly.

The file is named for the report rather than for the day you clicked, so two people saving it a week apart end up with the same file name and a folder of them sorts by period:

AI-Visibility-Report-yourdomain.com-month-to-2026-08-28.pdf

A report is not announced until its document is ready, so there is no waiting once it appears in your list. Reports generated before this existed do not have one, and neither does a report whose document could not be produced. In both cases the report opens as a page you can read and print in the browser, with a note at the top explaining what is missing and a Rebuild PDF button that tries again. Nothing about the figures is affected.

A report is a fixed document. The PDF of a report you downloaded in August still says in September exactly what it said in August, which is what makes it safe to send to somebody and refer back to later. The same holds when you list another site under Your websites in Brand Hub: reports written from then on treat that site as yours, and the reports you already have are unchanged.

When a report does not arrive#

The Automatic reports card explains why, rather than leaving you to guess. The reasons you may see:

• No new analysis has run since your last report. A fresh report would repeat the same figures, so none is written. This is the ordinary state between analysis runs.
• Automatic analyses are switched off for this project. Without them there is never new data to report on. Turn them back on in project settings.
• Brand Reports are not included in your current plan.
• No completed analysis has produced visibility data yet. Your first report follows the first one that does.
• The project is not yet 28 days old. A report covers 28 days, so there is not yet a full window. The card names the date your first one arrives.
• The last report could not be written. We are alerted automatically and retry on your next date.

If a report cannot be written, the report list tells you so the next time you open it, rather than simply showing no new report. Most reasons are ours to fix and need nothing from you. Where the cause is something you can act on, such as no completed analysis to report on yet, the message says which. The notice appears once, and stops once a later report has been written successfully.

Interpreting results#

The report's competition section ends with a grid of your tracked topics against the brands you are measured against. Read it a row at a time: each row is one topic, and the colour of your own cell answers a question the percentage in it does not, which is whether that share is a win.

A 20 percent share leading a topic and a 20 percent share trailing the leader by forty points are the same number and opposite situations, so the cell is coloured by where you stand rather than by how large the figure is. There are five states, and the line under each topic name spells the same thing out in words so that the grid still reads in black and white:

• Leading. You are clear of the next brand by ten percentage points or more, or you are level with or ahead of it and named in at least 70 percent of answers on that topic.
• Narrow lead. You are first, but tied with another brand or ahead of it by less than ten percentage points. A lighter green than a clear lead, because a lead that small can change on a single run.
• Contested. You are not first, but the leader is within twenty five percentage points of you, or nobody is named at all. This is the band worth acting on first, because the topic is genuinely open.
• Losing. The leader is more than twenty five percentage points ahead of you. Ground to take rather than to defend.
• Not named. You do not appear on this topic at all. It is filled solid rather than left blank, because an empty cell reads as a rendering fault rather than as a nil.

That line reads as a sentence: your placing among the brands named on that topic, then who it is measured against. TIED means another brand holds exactly the same placing. A placing shown with a plus sign means the list of named leaders was capped, so you are at least that far down rather than exactly there. Open field means no brand was named on that topic, yours included, which is an opportunity rather than a result.

Rows run weakest first. The work is at the top of the grid and the topics you already win sit below it, so the first thing you read is the first thing to act on.

Your competitors' cells are deliberately grey rather than coloured, so that the only colour in a row is yours. The exception is the brand leading each topic, filled dark, which lets you read down a column to see who owns what.

The columns are your tracked competitors, so a topic can be led by a brand that has no column, and then no cell in that row is filled dark. On those rows the leader's name carries an asterisk, and a line under the grid explains it: Leader is not tracked in your competitors. Without it the row would look as though nobody leads the topic when somebody does. The dark fill always means the top rate on that topic, never just the best of the brands on screen, so a column can be read downwards and trusted.

A competitor's cell is empty, with a dashed outline, when that competitor was not named on the topic at all. Reports built before per-topic rates were kept only recorded the four most-named competitors on each topic, so a competitor outside those four shows a ceiling instead, such as ≤ 12.5%: it was named no more often than that, and possibly not at all. A line under the grid says so whenever one appears.

The grid shows at most fourteen columns, which is what a page fits while keeping every percentage legible. If your tracked list is longer, the standings above it still name every brand, and a line under the grid tells you how many are not shown.

The sentiment section, What people say, is a table rather than a grid, and its middle column is the one worth slowing down for. Each row is one thing people discussed: Aspect names it, How opinion splits draws the split as a bar, Mentions counts how often it came up, and Where names a source it came up in.

The bar splits four ways, in reading order: spoken of well, read both ways, no clear signal, spoken of badly. The line under it names the two that carry a direction, and the third only when there is one, reading something like 12 positive, 19 negative, 3 both ways.

Both ways is a reading in its own right rather than a missing one. It is what a source gets when it praises and faults the same thing in the same breath, and it is the plainest evidence in the section that an aspect is genuinely contested rather than settled. It is counted once, as itself. Adding it to each side instead would let a row report more readings than it had mentions.

Because of that, the four readings add up to the row's Mentions exactly, and the bar fills its rail.

Mentions counts observations about an aspect, not answers and not articles. One article that praises your clinical care, faults your billing and says nothing either way about your app is three observations about three things. The line under the table states the total on that basis, which is why it does not match the answer counts elsewhere in the report.

Best practices#

• A clear direction supported by several runs, rather than one flattering snapshot chosen because it looked good.
• Findings tied to actions. A report saying what changed and what was done about it is far more useful than one that only shows numbers, particularly to somebody deciding what to fund.
• Honest about gaps. A report acknowledging a weakness is more credible than one that does not, and considerably more useful for deciding where the next quarter goes.

Next steps#

1. Switch Automatic reports on for the projects that need them, so a report arrives without anyone chasing it.
2. Record what you did during the period with Tracked actions, so the narrative can connect cause to effect rather than implying it.
3. Read it before passing it on. It is a document going to people who will form judgements from it and may not ask you a follow-up question.

What to keep in mind#

A report summarises what Seerly measures: AI answer visibility, sentiment, citations and competitor position on your tracked prompts. It does not include traffic, pipeline or revenue unless somebody reads those alongside it from your own analytics.

It describes the period it covers, and it is not a forecast.