Learn · 2026-09-30 · 2 min read
Which prompts should a small agency track for its clients in AI search?
A small agency should give each client a fixed set of 10 to 15 prompts in four shapes, local ("best [category] in [city]"), category ("best [category] for [audience]"), alternatives ("alternatives to [the competitor the client loses to]"), and trust ("is [client] good", "what is [client]"), then hold the set steady and report share of voice against three named competitors every month. Add five prompts for the agency itself, because the same engines are recommending agencies to your prospects. The templates below scale from one client to fifty without changing shape.
Free 3-day trial · No credit card · Cancel anytime
Four shapes per client
| Shape | Why it matters for a client | Templates |
|---|---|---|
| Local | Most small-business buying is local, and engines answer local prompts with named businesses | best [category] in [city]; [category] near [neighbourhood]; top-rated [category] [city] |
| Category | Whether the client is in the set at all | best [category] for [audience]; [category] recommendations |
| Alternatives | The prompt where the client loses today | alternatives to [competitor]; [competitor] vs [client] |
| Trust | What a prospect hears when they check the client out | what is [client]; is [client] good; [client] reviews |
Ten to fifteen per client is enough. The number that matters is not how many prompts you track but whether the set stays the same long enough for share of voice to mean something.
See what the engines say about your brand.
Ten of your prompts on ChatGPT and Gemini, twice, with the actual answers and your brand highlighted.
Five prompts for the agency itself
- best [your service] agency for small businesses
- best [your service] agency in [your city]
- alternatives to [the agency you lose pitches to]
- [your service] agency that also does AI search / AEO
- what is [your agency]
The fourth one is the interesting one. Prospects are already asking the engines which agencies understand AI search. Being named there is the cheapest pitch you will ever make.
Sizing the set to a plan
| Plan | Tracked prompts | Fits |
|---|---|---|
| Lite ($29) | 15 | One client, or the agency's own five plus one client's ten |
| Studio ($59) | 50 | Three to four clients at 12 each, ten of them scanned daily |
| Pro ($179) | 150 | Ten to twelve clients, all scanned daily |
| Agency ($449) | 500 | Thirty to forty clients, all scanned daily |
Reporting share of voice to a client
4.1% → 7.2% in 28 days
- Competitor A31%
- Competitor B18%
- Your Brand7.2%
- Competitor C4%
- Pick three competitors the client actually loses to. Not the biggest names in the category; the ones named in the client's prompts.
- Report one number per engine and one overall, month over month. "On ChatGPT you moved from 4% to 7%" is a sentence a client understands.
- Attach the sentence. The answer text where the client was named, or where the competitor was named instead, is the part of the report they will forward.
- Attach the move. What you published in response and which prompt it was aimed at. That is the line item the retainer pays for.
- Export it. helloAISearch exports prompts, scans, mentions, and share of voice as CSV and JSON on every plan, so the report is yours to format.