Set up AI Radar in six steps

WhyIQ AI Radar asks five AI engines (ChatGPT, Perplexity, Claude, Gemini, and Google AI Mode) the questions your buyers ask, every week, and reads back which sources each engine actually cited. Setup is one honest conversation with the product: tell it whose radar it is, where your buyers search, and which questions matter. Everything is editable later, so nothing here is a one-way door except the full prompt-bank rebuild, and Radar warns you before that one.

The whole flow at a glance

1. 1

   Add your domain at checkout. Radar generates your question bank automatically.

2. 2

   Check the target market. Purchases default to the United States.

3. 3

   Review the generated prompts: edit, add, archive, reorder.

4. 4

   Set the awareness stage and tidy the business profile.

5. 5

   Pick the weekday and hour your weekly scan fires.

6. 6

   Read your first This Week screen, then act on the trend.

Pick a plan on the Radar pricing page and pay by card. The checkout form has one field that matters more than the card number: Domain to track. The domain you type there is the one Radar scans every week, so use the domain your buyers actually land on, without www or paths (for example acme.com, not https://www.acme.com/pricing).

The moment payment clears, you land in your workspace and Radar starts working: it reads your homepage, extracts a business profile, and generates 30 to 50 buyer-intent questions tailored to what you sell. That takes 15 to 25 seconds, and the Prompts page shows a live skeleton until the questions land. You do not need to write anything yourself to get started.

How to read it

Your workspace opens on the client list. Click your domain to enter its workspace: the left rail holds everything this guide walks through, with This week at the top and Client settings near the bottom. The sidebar also shows your prompt slots and the exact time of the next scheduled scan.

Do this before judging any results. AI engines answer differently by country and language, so Radar localises every engine it queries, and the vocabulary of your generated questions, to one target market. New purchases default to the United States. If your buyers are in the UK, Germany, Australia, or anywhere else, set it now: a UK business measured against US answers is reading the wrong data.

1. 1
   What it controls. The country and language for the AI engine queries and for the buyer vocabulary in generated questions.
2. 2
   The picker. 30 markets grouped by region. Pick where your buyers search from, not where your office is.
3. 3
   No data loss. Changing market never resets your history. Your existing questions stay; append new market-aligned ones after switching.

How to use it

1. 1

   Open Client settings in the left rail.

2. 2

   Scroll to Target market and pick your country from the grouped list.

3. 3

   If you changed it, open the Prompts page and use Append new queries (Step 3) so the bank gains questions written the way buyers in that market actually ask.

Prompts are the questions Radar asks the engines every week, and they are the single biggest lever you control. The generated bank is a strong starting point, but you know your buyers better than any model does. Spend ten minutes here: read every question and ask "would a real buyer of ours type this into ChatGPT?" Edit or archive the ones that miss, and add the ones only you would know.

1. 1
   Your questions. Click any prompt to read the full AI answers behind it, word for word.
2. 2
   Google AI answer. Whether Google shows an AI answer for this query in your market, and how many sources it cites. It describes the query's demand, not your performance.
3. 3
   One column per engine. Filled dot: cited you. Half dot: named you without a link. Hollow: neither. Two dashes: the engine returned no measurable answer.
4. 4
   Manage prompts. Turns on editing: drag to reorder, edit, archive, and add. The page opens read-only so you cannot reorder by accident.

Your plan scans a fixed number of questions each week (30 on SMB, 40 per client on Agency), but the bank can hold more. Everything above the blue line runs weekly; everything below is held in your bank, keeping its history, until you drag it up.

1. 1
   Drag handles. In Manage mode, drag any row above the line to include it in the weekly scan, or below to bench it.
2. 2
   Brand queries. Questions naming your brand are tagged BRAND and measured separately, so they never flatter your buyer-question rate.
3. 3
   The line itself. States exactly how many run this week and how many are held. Nothing is scanned or billed silently.

How to use it

1. 1

   Read every generated question. Archive any a real buyer would never ask (archived prompts keep their history and stop running).

2. 2

   Add questions you know buyers ask: the objection they raise on sales calls, the comparison they always make, the problem they describe in their own words.

3. 3

   Click Manage prompts and drag your most important questions above the scan line.

4. 4

   Use the Google AI answer column as a demand signal. “No answer” means Google runs no AI answer for that query, which is a fair reason to swap it out, though the other engines can still cite you on it.

Two settings shape how Radar thinks about the brand. Both live in Client settings, and both are honest questions, not vanity settings: answer them with how things are, not how you want them to be.

1. 1
   Pre-awareness. Nobody searches the brand name yet. Radar focuses the plan on getting the brand established before defending it.
2. 2
   Building. Some people search the name. Brand-defense questions run alongside the buyer questions.
3. 3
   Established. Brand searches reliably cite you. The dashboard tunes to defending late-funnel queries and watching displacement.

The business profile below it is what Radar extracted from your homepage: category, who buys, the value proposition, and the vocabulary notes. The vocabulary field is the one worth editing: it records how buyer language differs from your marketing language (sellers say "revenue intelligence platform", buyers say "call recording for sales teams"), and it drives every future prompt generation.

Prompt-bank operations live in one place: the Manage the bank panel at the top of the Prompts page.

1. 1
   Brand-defense queries. Questions naming your brand (pricing, reviews, comparisons). Additive: adding them never touches your buyer questions.
2. 2
   Append new queries. The safe way to grow the bank. Existing questions stay, so your week-over-week numbers stay comparable.
3. 3
   Danger zone. The full rebuild archives everything and starts over. Your trend resets, which is why it sits behind a warning. Use append instead unless the business has fundamentally repositioned.

Scans run weekly. In Client settings, pick the weekday and hour (UTC) the scan fires; Tuesday morning is the default. Radar emails you when each scan lands, so there is nothing to remember: the product comes to you once a week with what changed.

1. 1
   Day and hour. When the weekly scan fires, in UTC. Saving recomputes the next run time immediately.
2. 2
   Pause and resume. Pausing skips the schedule without deleting anything. Useful for a client offboarding or a site mid-migration.
3. 3
   Next scan. The sidebar always shows the exact next run time, so you are never guessing when fresh numbers arrive.

How to read it

Weekly is deliberate. AI answers move slowly and off-page work takes weeks to register, so a weekly read is signal and a daily read is noise. If you ship something big and want fresh numbers before the next slot, the Scans card in Client settings offers a one-time off-cycle scan for a small fee, without touching your schedule.

When the first scan lands, the This week screen opens with a plain answer: is AI recommending you, what share of your buyer questions had an answer citing you, and what changed since last week. The first scan is your baseline; movement starts from the second.

1. 1
   The verdict. One sentence plus the cited rate: the share of buyer questions where an AI answer cited you this week.
2. 2
   Brand queries, counted separately. Searches for your name are shown on their own line so they never inflate the buyer-question number.
3. 3
   What changed. How many questions were cited this week versus last, then every climbing and slipping prompt, each with its trend.
4. 4
   Query scope chips. Flip the whole screen between all queries, buyer questions only, and brand-name searches only.

The three states every answer lands in

Cited

The AI answer linked to your site as a source. This is the win: the engine pointed the reader at you.

Mentioned

The answer named your brand but did not link to your site. Better than nothing, but the reader gets no route to you.

Absent

The answer neither named you nor linked to you. The Action plan exists for these.

How to read it

Calibrate your expectations low and honest: most brands start near zero, and realistic cited rates stay in single or low double digits. Under 5 percent reads as Starting, 5 to 12 as Building, 12 to 25 as Strong, and 25 or more as Leading. Do not react to any single week; the trend over weeks is the product. When you want the Search Console style detail, it is one click away in Analytics, and the prioritised response to every absent answer lives in the Action plan.

Five habits separate accounts that get value from accounts that just collect data:

1. 1

   Read the weekly email

   It lands when the scan finishes and names your top engine and top competitor. It is the 30-second version of This Week.

2. 2

   Click into prompts that moved

   Every prompt opens into the verbatim AI answers behind it: which engine cited whom, at what position, and what the answer actually said. That verbatim text is the most persuasive artifact in the product.

3. 3

   Declare your review profiles

   In Client settings, tick the review platforms you already have (Google Business Profile, Trustpilot, and so on) and paste the links. Radar folds them into the advice instead of recommending places you are already listed.

4. 4

   Append, don't rebuild

   When buyer vocabulary drifts or a new sub-niche appears, use Append new queries. The bank grows, the history stays comparable, and the trend keeps meaning something.

5. 5

   Work the Action plan, then wait 4 to 8 weeks

   Off-page moves (reviews, communities, third-party coverage) take weeks to show up in AI answers. Do the work, keep scanning weekly, and judge it on the month, not the Tuesday.

On the Agency plan every client gets its own workspace, its own 40-question bank scanned weekly, and a white-label client report you can publish from the Reports tab. Each question also runs three times per scan and is averaged into a confidence band, so the numbers you put in front of a client are not single-run noise.

1. 1
   Domain first. Radar derives the brand name and, from the domain's ending, preselects the target market: a .co.uk domain lands on United Kingdom automatically. Change it if it guesses wrong.
2. 2
   Established awareness. Tick it when people already search the client's name; Radar then seeds brand-aware questions alongside the buyer set.
3. 3
   First scan and schedule. Pick when the first scan runs and which weekday the weekly cycle fires, per client.

How to use it

1. 1

   From the client list, click Add client and enter the client's domain.

2. 2

   Confirm the auto-detected market, set the awareness stage honestly, and let the bank generate.

3. 3

   Run Steps 2 to 5 of this guide once per client. Ten minutes each is enough.

4. 4

   After the first scan, open the Reports tab to preview and publish the white-label weekly report.