Mention Clock

AI prompt ideas by industry

The hard part of tracking is choosing questions buyers actually ask. Generate a balanced list, then freeze the wording.

Built by StyleLine FZE in Sharjah. Email support@mentionclock.com. We reply within one business day.

Pick prompts that are too narrow and you look strong on questions nobody asks. Too broad and you never appear. This generator suggests buyer questions for your industry, grouped by intent. Then you freeze the wording so week to week scores stay comparable.

Mention Clock runs those prompts through ChatGPT, the second chat engine and Perplexity on a schedule and records whether your brand is named, in what position, which rivals appear and which pages get cited. Free covers 1 project, 5 prompts and 2 engines. Starter is $29 a month for 25 prompts and 3 engines. Prices are on the pricing page.

Five shapes

  1. Category "best of." Highest stakes. The answer is a shortlist.
  2. Problem first. The buyer has not chosen a category yet.
  3. Alternatives. "What are alternatives to [rival]?" Strong intent.
  4. Comparisons. "[You] vs [rival] for [use case]."
  5. Specific needs. Platform, budget, constraint. Common in real chats.

Why mix the five shapes

A list made only of "best X" prompts will swing hard when one assistant changes its shortlist. Problem first prompts show whether you appear before the buyer names a category. Alternatives and comparisons show whether you show up when a rival already owns the question. Specific need prompts catch the long tail that drives real trials.

If your share looks strong on category prompts and weak on alternatives, you are winning awareness and losing switchers. That is a product and content problem, not a tracking bug.

Sketch examples

B2B software: best project tool for a small construction firm; how an architecture office tracks hours; alternatives to a well-known rival.

Local service: family dentists in a city; same-day emergency care; good with anxious patients.

Retail: running shoes for flat feet under a budget; durable trail shoes for beginners; alternatives to a popular model.

Professional firm: accountant for a startup in a city; do I need an accountant for a small online shop?

Developer tools: best CI for a monorepo; how to cut flaky tests; alternatives to a popular hosted runner.

Write the way customers talk. If they say "app", do not write "platform". If they say a city name, keep the city. If they name a rival, keep the rival spelling they use in search and chat.

A 25-prompt mix

6 category, 6 problem, 5 alternatives, 4 comparisons, 4 specific-need. That mix fits Starter. On Free, pick 5 that cover at least three shapes so you do not read a false trend from one question type.

Skip branded-only lists that only ask for your name. Skip your own jargon. Skip ten near-duplicates that differ by one adjective. Review the list each quarter or when you launch a new product line.

How to use the prompt generator

  1. Enter industry, product type and audience.
  2. Add a location when you sell locally.
  3. Add one top competitor when you care about switch questions.
  4. Generate 20 prompts grouped by intent.
  5. Copy or export the list. Drop weak lines. Freeze the keepers.

Run each draft once in a couple of assistants before you lock it. Drop prompts that stay generic. Keep the ones that produce a real shortlist with named products. Then add them to a Mention Clock project and leave the wording alone for at least four weekly runs.

Freezing wording matters

Assistants are sensitive to small edits. Changing "best CRM for a 10 person sales team" to "best CRM for small sales teams" can move who gets named. Treat the prompt like a keyword you would not rename every week in Search Console. When you must edit, note the date so a drop is not mistaken for lost visibility.

What to do with the results

When a prompt never names you, check the pages the assistants cite. Those pages are the real ranking surface for that question. When a rival owns the alternatives prompt, improve the public comparison page that names both products with dated facts. When you win category and lose comparisons, your category proof is stronger than your head to head proof.

Link the frozen prompts to ChatGPT brand monitoring, Perplexity rank tracking and share of voice in AI search so each engine has a matching guide.

Industry prompt sets

The sketches above are starters. A list you can freeze looks closer to the sets below. Each line is a question a buyer would type, with a job and a constraint. Replace the city, the rival, and the limit with words from your own sales calls. Do not paste your brand name into every line.

Clinics. Best recall software for a dental office with two locations. How a clinic books same day visits without a phone queue. What to use instead of the scheduler the front desk hates. Which practice tool keeps insurance notes next to the appointment. Software for a physiotherapy room that needs evening slots visible online.

Restaurants and hotels. Best reservation book for a restaurant with a patio and a dining room. How a hotel group reads guest complaints across three properties. What to use instead of the point of sale the bar already owns. Software that shows table turns on a Friday service. Which pickup app a neighborhood cafe should pick.

Property teams. Best contact list for a small brokerage. How an agent tracks listings and the follow up after an open house. What to use instead of the portal the office pays for. Software that shows which listings got tour requests last week. Which tool a property manager uses for repair requests.

Courses and schools. Best storefront for a tutor selling a short course. How a school keeps parent messages in one thread. What to use instead of the classroom tool the district already bought. Software that shows which lesson a buyer finishes. Which webinar tool a coach should use for a paid series.

Freight and field work. Best dispatch board for a regional carrier. How a field crew schedules heating and cooling visits. What to use instead of the tracker the warehouse uses. Software that shows late drops for a dispatcher. Which phone app a plumbing company uses to quote a job on site.

Keep one industry in one project. A clinic list and a freight list in the same share number hide both.

How to judge a prompt

Run each draft in two assistants before you lock the wording. Score the answer. Ignore how clever the sentence sounds.

  1. Named options. A useful answer names products, firms, or local businesses a buyer could contact. If the reply is only general advice, the question is too wide and a mention cannot be scored.
  2. Buyer wording. The prompt uses the job, the place, and the limit customers already say. A line full of your own category jargon will miss the chat a prospect actually starts.
  3. Same shape twice. Ask again a few days later. If one answer is a shortlist and the next is an essay, weeks cannot be compared. Keep prompts whose answers stay in one shape.
  4. Pages a buyer can open. Citations should be a product site, a comparison, or a help article. A prompt that only cites a dictionary is a definition question.
  5. Rivals you actually meet. If the shortlist never includes the rival you lose deals to, the prompt aims at the wrong buyer. If it only names you, it is a cheer and it will not show a drop.

Drop a line after three runs that name nobody in the category. Keep a line when the shortlist includes products you compete with and a page you could improve. Freeze those keepers. Do not rename them until four weekly runs have finished.

Cutting that brainstorm to a frozen list of buyer questions is how to pick the 20 prompts that matter.

FAQ

How many prompts do I need?

Twenty to thirty is enough for most single brands to see a direction. A brand with many product lines may need 100, which is the Pro limit.

Should prompts include a city?

If you sell locally, yes. Local answers depend on place.

Is the prompt generator free?

Yes. The prompt generator is free, and you can track 5 prompts on a schedule free with no card. Paid plans raise the prompt cap.

Can I change prompts later?

Yes. Prefer scheduled reviews over daily edits. After a change, wait through a few runs before you judge the new baseline.

How do I judge a prompt before I freeze it?

Run it twice. Keep it when both answers name real options, cite pages a buyer can open, and stay in the same shape. Drop it when both answers are only general advice.

Which engines does tracking cover?

Paid plans query ChatGPT, the second chat engine and Perplexity through official APIs. Free covers 2 engines. Details stay on pricing.

Add them to a free project