Skip to content
Help Centre
zicy.com
Try for free

Brand Intelligence

What this is for, in one sentence: Brand Intelligence checks what AI platforms say about your brand, compares it with what your brand profile and website say, and gives you a short list of fixes — so the story AI tells your buyers is one you’ve verified.

Each check-up asks four AI platforms: ChatGPT, Gemini, Google AI Mode and Google AI Overviews. It does not include Perplexity, which is covered by prompt tracking instead.

When to come here:

  • Early in your setup, before you invest in new content — fix what AI already believes first
  • After any major change to pricing, positioning, offerings or company facts
  • On a regular cadence, because AI answers drift as models and sources update

Priya Nair’s job is MenuPilot’s story, and the most influential re-teller of that story answers thousands of buyer questions a day without ever calling her for comment. This is the screen where she reads what it’s been saying — and finds out that some of it is confidently, sourcedly wrong.


  • AI engines that describe you correctly. How many of the 4 AI engines describe your brand correctly overall, shown as a percentage and “x of 4”. Aim to keep it at 4 of 4.
  • Facts AI gets right. We check 5 things on each of the 4 engines, so there are 20 checks in total. This is how many came out correct, shown as “x of 20”.
  • Fixes done. How many of the fixes from your latest check-up are now fixed, shown as “x of N”. Rerun the check-up to confirm.
  • Share of information from your own site. The part of the information AI uses about you that comes from your own pages.
  • Check result. Each cell of the grid is Correct, Partly, Wrong, Mixed (the two runs disagreed) or No data.

There is no setup form. When you create a brand profile, Zicy starts a check-up from it automatically. The profile’s country is required, because the check-up needs to know where to look. On the Brand Intelligence page, Run analysis starts a new check-up from your current profile, and you do not fill anything in.

A check-up asks 6 questions in the profile’s language (English, Malay or Chinese). Every question goes to 4 engines, 2 runs each, so you get a fair read rather than one lucky or unlucky answer. While it runs, the page shows progress for each engine and an estimate of the time left. If something fails, you can retry from the same screen.

Run for another country is next to Run analysis. It asks “Which country is this report for?” Choose it when you want the report for a different market. Those runs appear in your history alongside the others, and you find them with the location filter.

The history page shows three cards for your latest check-up (engines correct, facts right, fixes done, each with its change since the last one), and a table of every earlier check-up with its date, location, language and results. Click a row to open that report. There is no trend chart.

The report is one scrolling page, in this order:

  1. Summary. Two or three sentences, for example: “AI knows MenuPilot and mostly describes it correctly. The main gap is pricing: Google AI Mode leaves out your Agency plans.” Above it, a row shows the date, location, language and engines (“4 engines × 2 runs”).
  2. At a glance. Four cards: AI engines that describe you correctly (x of 4), facts AI gets right (x of 20), fixes done (x of N), and the share of information that comes from your own site. Each shows its change since the last check-up, and clicking one jumps to its section.
  3. What AI says about you. One card per engine with its verdict and a short quote, then two lists: what AI says you do well, and where AI has doubts.
  4. What AI gets right and wrong. A 5×4 grid: five checks (Who you are, Location, Offerings, Pricing, Reviews) against the four engines, 20 cells in all. Reviews asks whether the reviews AI reports are real: an honest “few or no reviews found” counts as correct, and invented reviews, or another company’s, count as wrong.
  5. What to fix. The list of fixes, covered below.
  6. Where AI gets its information. The sites AI drew on, with your own pages marked as yours.

Click any grid cell or engine card to open the answer pop-up for that check and engine. It shows the exact question asked, the answer from each of the two runs, a verdict line and the sources the answer used, as links to the exact page. For pricing, it compares what AI said with your pricing page, plan by plan. Use View all answers to browse all six questions, and Show full answer to expand a long one. If an engine did not recognise your brand in a run, the pop-up says so instead of showing an answer.

If your profile and website disagree on a fact, a notice at the top of What to fix says so, with an Edit brand profile link. It is a prompt to check your own facts, not a fix, and it is not counted in Fixes done.

Mixed is not a failure. It means the two runs on that engine disagreed. Open the cell and read both answers: an engine that sometimes gets your pricing right and sometimes gets it wrong is still a source worth fixing.

Each fix says what is wrong, how serious it is (high, medium or low), and the steps to correct it (How to fix). See answers opens the answer that showed the problem. Filter the list by All, High, Medium, Low, New or Not on fix list.

  • + Fix list on a card sends that fix to your Opportunities backlog (Visibility Gaps); it then shows On fix list. Add all to fix list sends every fix that is not on it yet. Each issue becomes one item, however many times it appears.
  • Dismiss is for fixes you do not agree with. Pick a reason (Not true, Already fixed, Not important or Other), add an optional note, and confirm. The fix moves to Dismissed, and Zicy will not flag it again in your next check-up. Undo puts it back.
  • Fixed since is a group of problems from your previous check-up that the latest one no longer found.

Execution lives one module over: the Take Action tools generate the schema (Schema Generator) and the llms.txt (LLMs Generator), and the Content tools produce the corrected pages (Content Optimizer) that state your verified facts where machines read them.

Fixes take time to reach AI answers. Correcting your site changes the sources; the answers follow as AI systems re-crawl and update — the lag varies by engine and can run weeks. Rerun the check-up after shipping fixes, and judge progress over several runs rather than expecting next-day corrections.

Run analysis on the history page, or Rerun in a report, starts a fresh check-up. The new report shows what was fixed, what is still open and what is new since the last one, so you can see whether your work changed what AI says. Fixes you dismissed stay dismissed.

The report text follows the language you view Zicy in, whatever language the check-up itself ran in.

The Export menu in the report header offers PDF, Markdown (.md) and PDF + Markdown. Both follow the report order: summary, at a glance, what AI says, right and wrong, what to fix (with steps) and sources. The PDF leaves out the raw AI answers.

Share the PDF to share the report. New reports have no public share link. Reports from before the redesign still open in their old layout, with their original scores and tabs.

  • Who can use it. Brand Intelligence is hidden from viewers in the sidebar, and starting a check-up is restricted to owners and members.
  • What spends allowance. A standard check-up uses 1 Brand Intelligence run. These runs are separate from your Analysis quota. A run for another country is started from the Brand Intelligence page, and it uses a run too. When your runs are used up, the page offers to buy a runs pack or upgrade your plan. A runs pack never expires and is used only after your monthly runs are used up.
  • What does not. Reading a report, opening answers, filtering, dismissing and exporting do not use a run.

How often should I rerun it? After any batch of shipped fixes, and on a regular cadence otherwise — monthly is a sensible default for most brands. Check how many Brand Intelligence runs you have left this period before you plan a cadence.

What happened to the Perception, Coverage and Alignment scores? They are gone. The report now counts concrete things: engines that describe you correctly, checks AI gets right, and fixes you have done. Reports from before the redesign keep their old scores.

A fix is wrong, or already done. What do I do? Dismiss it with a reason. It leaves your list and Zicy will not raise it again in the next check-up. If you dismissed it by mistake, Undo it.

Can I change which questions are asked? No. Every check-up asks the same six questions, which is what makes one report comparable with the next. To track your own questions, use prompt tracking.

Do the engines matter individually? Yes — the grid shows each engine separately, and that is targeting information. A wrong fact confined to one engine usually traces to a source that engine favours.