Filters & time periods
How the period and filters scope every number, and how they follow you between pages
Every number in GEOflux is computed over a scope: a period, some engines, some prompts. The filters in the top bar set that scope for the whole page at once, every lead, chart and table. Two people looking at the "same" visibility score with different filters are looking at two different numbers, so set the scope first, then read.
The filter bar has two parts: the period, and the Filters button, which opens a panel with everything else. On narrow screens the period moves inside the panel.
Time period
The period sets the date window. Presets:
- Last 7 days (the default)
- Last 14 days
- Last 30 days
- Last 60 days
- Last 90 days
- Custom range…, which opens a calendar to pick exact start and end dates. On the Overview, the calendar starts at your project's first prompt date, since there is nothing before it.
A preset covers today and the days before it, so it keeps rolling: "Last 30 days" next week is a different window. A custom range stays fixed, which is what you want for a baseline or a monthly report.
Every change is compared with the period of the same length just before it. That is where the +2.3% next to a number comes from.
Filters on each page
| Page | Filters |
|---|---|
| Overview | Period, Brand, Engine, Topic, Tag, Buyer Journey, Branded vs Non-branded, Persona |
| Answers | Period, Brand, Engine, Prompt, Sentiment, Topic, Tag, Buyer Journey, Branded vs Non-branded, Persona |
| Prompts (Active tab) | Period, Brand, Engine, Tag, Buyer Journey, Branded vs Non-branded, Persona |
| Prompt detail | Period, Brand, Engine, Persona |
| Brand and source detail | Period, Brand, Engine, Persona |
On Answers and Prompts, the Topic, Tag and Persona filters appear once your project has topics, tags or personas.
What each one does:
- Brand: the brand the page is about. On the Overview, Prompts and prompt detail it starts on your own brand; the Overview and Prompts also offer All brands (see below). On Answers you can pick several brands, and the list shows answers that mention any of them. Brand and source detail pages start on All brands and also offer Unattributed, for citations not credited to any brand.
- Engine: ChatGPT, Gemini or Perplexity, one or several. On Prompts, pick one engine at a time: with several selected, the table shows all engines.
- Topic: one or more topics.
- Tag: one or more of your tags.
- Buyer Journey: Discover, Evaluate or Buy. See Organize your prompts.
- Branded vs Non-branded: Branded (names your brand or a competitor) or Non-branded.
- Persona: Base only (answers with no persona), one or more personas, or a mix. With nothing selected, the base answers and the answers for all your active personas count together.
- Prompt (Answers only): one or more of your active prompts.
- Sentiment (Answers only): positive (7 and above), neutral (4 up to 7) or negative (below 4), on the score shown on each answer. Answers with no score drop out once this filter is set. See Sentiment.
How filters combine
- Within one filter, values add up (OR). Evaluate + Buy is every prompt in either step.
- Across filters, they narrow (AND). Evaluate + Non-branded + ChatGPT is ChatGPT's answers to Evaluate prompts that name no brand.
- For Buyer Journey and Branded vs Non-branded, selecting every option is the same as selecting none.
- Filters include; they never exclude. A prompt with no Buyer Journey step, no topic or no tag drops out as soon as you filter by that field.
The Filters button shows how many filters inside the panel are set. The clear button next to it (Clear all filters) resets every filter to its default but keeps the period.
The All brands view
On the Overview and Prompts, the Brand filter starts on your own brand. Pick All brands to look at the whole market instead of one subject:
- Counts are totals. Mentions add up across every tracked brand.
- Percentages lead with the market leader. Visibility and share of voice have no meaningful total, so the lead shows the leading brand and your gap to it, or your margin over the runner-up if you lead.
- Sentiment is a market average, weighted by how often each brand is mentioned, so the most mentioned brands move it most.
- Some views are about one brand. The Personas tab and the topic competitor breakdown ask you to pick one; Brand Perception shows your own brand.
Filters follow you between pages
Filters you set are remembered for each project while your browser tab is open. When you open another page, it starts with the filters you last used that also exist on that page, so moving from the Overview to Answers keeps your period and engines.
- A link always wins. A URL that carries filters, such as a link a teammate sent you or a drill-down, opens with exactly those filters.
- Only matching filters carry over. Brand is a multi-select on Answers and a single choice elsewhere, so it does not travel between them.
- Each project has its own. Switching project brings back the filters you last used on that project.
Filters live in the page address
Every filter, including the period and the open Overview tab, is written to the page address:
- Shareable and bookmarkable. Copy the URL and a teammate opens the same view.
- Carried through drill-downs. Clicking into a brand or a source keeps your period, brand, engine and persona. Topic, tag, Buyer Journey and Branded vs Non-branded do not apply on those pages, so they are left behind.
Good practice
- Decide your headline scope once, for example last 30 days, all engines, Non-branded, and use it every time you report. Comparable numbers matter more than any single number.
- Use a custom range for anything you save: a baseline, a monthly report, a before-and-after.
- Change one filter at a time when a number surprises you. Whichever filter moves it is where the story is.
- Check the filter count before you screenshot. A forgotten tag or persona filter is the most common reason two people's numbers disagree.