Answers

Every number in GEOflux is counted from answers, and the Answers page is where you read them: the exact text ChatGPT, Gemini and Perplexity returned for your prompts. Metrics tell you how often you are named. The answers tell you how: as the budget pick or the premium one, with an outdated price, next to which competitors, on the strength of which sources.

Browsing answers

Open Answers from the sidebar. Every captured answer is listed, newest first, 25 per page, in one of two views:

Grid: cards with the engine, the prompt, a short preview, whether your brand is mentioned and its sentiment, the competitors mentioned, and the top cited domains.
Table: a denser list with the prompt, your brand (with sentiment and rank), competitors, sources and the date.

An answer that does not name you is marked Not mentioned. That is a finding in itself, especially when a competitor is named instead.

Searching

Search answers finds every answer containing a word or phrase: a product name, a price, a claim you want to check. It ignores case, highlights the matches in the previews, and combines with your filters and your export.

Filtering

The filter bar narrows the list by:

Period
Brand: one or several of your brand and competitors. The list shows answers that mention any of them.
Engine
Prompt: one or more active prompts.
Sentiment: positive, neutral or negative, on the score shown on each card: your brand's, or the selected brand's when you pick exactly one.
Topic, Tag, Buyer Journey and Branded vs Non-branded
Persona, including Base (no persona) for answers asked without one.

Filters and search are saved in the page address, so a link opens the same slice of answers for a teammate.

Inside an answer

Click any answer to open it. The main panel shows the prompt, the engine, the date it was captured, the persona it was asked for (if any), and the full response as the engine rendered it, with its tables, images, product carousels, place lists and inline citation markers.

The side panel shows what GEOflux read from it:

Your brand, highlighted, with its sentiment score out of 10, or a note that it was not mentioned.
Competitors in this answer, each with its own sentiment.
Other brands that came up.
Sources: every page the answer relies on, grouped by domain. The engine's own citation list comes first, then any other page linked in the answer text. Each page opens in a new tab; when GEOflux already tracks it, the row also opens its source page in the app, with your filters carried over, and the domain heading opens the domain's page.

Answers marked API

Now and then a Perplexity answer carries an API tag in the list and a Via API label in the answer. It means the answer was collected through Perplexity's API instead of its website, because the answer could not be collected from the website at that moment. API answers are usually shorter and can cite different sources than the website would, so keep that in mind when you compare one with its neighbours.

Exporting answers

Export downloads answers for offline analysis or your own tools. It uses your current filters and search.

Spreadsheet (XLSX): pick the columns (Date, Engine, Prompt, Persona, Answer, Brands with sentiment, Citations) and how many rows (50, 100, 250 or 500).
Raw HTML (ZIP): tick Also download raw HTML (.zip) to get each answer's full HTML as its own file. It keeps tables, images and product blocks that do not fit in a spreadsheet cell, which is what you want when feeding complete answers into your own AI. Untick every column to download only the ZIP.

If more answers match than the rows you picked, the newest are exported and the export window tells you so. Narrow the period or filters to export a specific slice.

What to use it for

Catch outdated or wrong claims. Search for your product, your pricing or a feature name, and find where an engine describes something you changed long ago. The sources on that answer show which page to correct.
Understand the framing. Are you recommended for small teams or enterprises, as the cheap option or the premium one? The context of a mention matters as much as the mention.
Learn what the engines prefer. Notice the shape of the answers (lists, comparisons, step-by-steps) and the pages they cite, and shape your own content the same way.
Investigate a number. When a metric moves, set the same filters here and read the answers from those days.
Show the evidence. Export a filtered set for a stakeholder or a client, or read a few answers aloud in a meeting. Nothing convinces like the actual words.