What the CSV export includes
A CSV export for your report's underlying data, available from the Export tab in Report Settings. Running an export requires the Growth or Custom plan.
A three-step wizard (Choose data → Filters → Review) lets you download the three datasets:
Dataset | What's in it |
Brand mentions | Which companies were mentioned, where, and how often |
Domain sources | Which domains and pages were cited in AI responses |
Prompts & responses | Your tracked prompts, brands mentioned, sources and the full response text. |
The Export tab is visible on every plan. On Starter, you can see what the export offers and upgrade from there; running an export requires Growth or Custom.
Why getting your AI visibility data out of ALLMO matters
Until now there was no way to get a report's underlying data out of ALLMO. If you wanted to build a custom chart, join AI visibility data with your own analytics, run a pivot table for a client, or hand a dataset to an analyst, you were stuck. This is the most requested feature from our agency and analyst users.
The Prompts & responses export in particular is the one to know about: it includes the full text of every AI answer we collected, which means you can run your own analysis on how models actually talk about your brand: sentiment, phrasing, which claims get repeated, which competitors get named alongside you.
How the export wizard, filters, and file formatting work
The wizard has three steps.
Choose data: pick one of the three datasets.
Filters: set date range, models, language, country, tags, and (where the dataset supports it) the competitor scope.
Review: see the row count and any size or row-cap warning before you download.
Export filters are independent of the page you are on. This is deliberate and worth internalising: whatever filters you have active on Visibility or Domain Sources have no effect on your export, and the filters you set in the wizard do not change your live view. Exports are therefore predictable and repeatable: the same filter selection produces the same file, regardless of what anyone had open at the time.
File formatting is spreadsheet-safe. Files are UTF-8 with BOM and use a locale-aware separator, so umlauts and accented characters open correctly in Excel without a manual import step. Percentages are rounded consistently, and missing values are left blank rather than written as 0, so an absent data point never gets mistaken for a real zero in your averages.
Row cap: 5,000. The Prompts & responses export is capped at 5,000 rows. The review step warns you before you hit it, so you know when to narrow your date range or filters and export in slices.
How to run a CSV export
Open your report in ALLMO and click the gear icon in the navbar.
Go to the Export tab. (On Starter, you'll see what the export includes along with an upgrade option.)
Click Export data.
Choose data: Brand mentions, Domain sources, or Prompts & responses.
Set filters: date range, models, language, country, tags, and competitor scope where it's available. Marked competitors are required for competitor scoping (see the Mark Competitors article).
Review: check the row count. If you see a row-cap warning, narrow the date range and export in batches.
Download the CSV.
How to turn exported data into analysis and client reporting
Build the client deliverable you actually need. Agencies: export brand mentions per month, pivot by competitor, and drop a single chart into the client deck. No screenshotting.
Run text analysis on Prompts & responses. The full response text is the richest asset in the export. Look for the phrases models repeat about your category, which claims they attribute to you, and which competitor gets named in the same sentence. That output is a content brief.
Join it with your own data. Match domain sources against your backlink or referral data to see which cited domains also send traffic, and which are AI-only surfaces you'd otherwise never notice.
Track period over period. Export the same filter set each month and keep the files. Over a few months you have a longitudinal dataset that no dashboard view can give you.
Slice by model. Exporting per model and comparing side by side shows where a model has an outdated or incorrect picture of your brand. That's usually the highest-leverage thing you can fix.
Watch the blanks. Because missing values stay blank, a column full of blanks is a signal in itself: that dimension isn't being captured for those responses, which usually points at a prompt-coverage gap.