The pattern in one line
Clients ask why an AI engine said what it said, and which sites it got the answer from.
AI answers draw on sources, and the sources an engine cites are the best available evidence of where an answer came from. Show clients the prompts, the answers, the cited sources, and who else is named. Then separate what you can influence, such as their own pages and listings, from what you cannot.
What this looks like
A client emails a screenshot. They asked an AI engine for the best provider in their area, and a competitor was named. Their business was not. The message ends with 2 questions: why, and what are you doing about it.
The honest answer starts with sources. AI engines assemble answers from material they can reach: pages on the open web, reviews, directories, and, depending on the engine, live search results. Many engines show the pages they relied on. Those cited sources are the best visible evidence of where an answer came from, and they show you what to change.
Agencies without a way to show this fall back on general statements about how AI works. That is unsatisfying to a client who wants to know about their own business.
Preparing for the question is easier than answering it live. Track the prompts your clients are likely to ask before they ask them, and keep the cited sources for each. Then you can reply the same day with something concrete.
A report that shows the prompt, each engine’s answer, the sources cited, and the brands named turns a vague worry into a short list of actions.
Common traits
- Clients paste AI answers into emails and ask about them
- They want to know why a competitor is named
- You can explain SEO results but not AI ones
- Your reporting has no standard view of cited sources
- Questions arrive after a client sees a wrong or missing mention
Why it happens
-
AI answers are new to clients
They have a clear picture of search rankings and none of AI visibility.
-
Answers are not repeatable on demand
A client sees 1 answer at 1 moment, and a screenshot cannot show the pattern.
-
Reporting does not cover sources
Without the cited sources, you cannot show where an answer came from.
-
Influence is unclear
Clients do not know which parts of an answer you can affect and which you cannot.
How to confirm it in Doily
- Open AI visibility for the client workspace and select the prompt the client asked about.
- Read each engine's response, the brands mentioned, and the sources cited.
- Compare the client's mention rate and share of voice with their competitors.
- Add this view to the client's report, or share it in the client portal.
Fixes to try
| Fix | Impact | Effort |
|---|---|---|
| Track the prompts your clients actually ask, and add new ones when they raise them | High | Low |
| Show cited sources next to every answer so the origin is clear | High | Low |
| Explain which sources the client controls and which they can only influence | Medium | Low |
| Set a monthly slot to review AI visibility with each client | Medium | Low |
| Report mention rate and share of voice over time, not 1 screenshot | High | Medium |
| Turn the sources into tasks, such as fixing a listing or improving a page | Medium | Medium |
What to watch after
- Mention rate for the prompts clients ask about
- Cited sources that carry the client's name
- Share of voice against named competitors
- Number of client questions per report
Questions
Can I tell a client exactly why an engine said something?
Not exactly. Engines do not publish how each answer was built. Cited sources and repeated checks give the best available evidence, and you can say so.
Which engines show sources?
It varies by engine and by answer. Use the cited sources that appear in your own data and avoid generalizing.
How do I handle a client who sees a wrong answer?
Check the full response, find the cited sources, and correct the pages that hold the wrong fact.
