---
title: "Earning mentions in the sources AI answers draw on"
description: "Recommendations are shaped by third parties For prompts like \"best project management tool for agencies\", \"top dermatologists in Lahore\" or \"alternatives…"
url: https://optimizeall.com/learn/ai-search-optimization-geo/earning-mentions-in-trusted-sources
updated: 2026-10-05
---

AI Search Optimization: SEO for AI Overviews & Answer Engines · Entity clarity and brand consensus · lesson 8 of 17 · 13 min

# Earning mentions in the sources AI answers draw on

## Recommendations are shaped by third parties

For prompts like "best project management tool for agencies", "top dermatologists in Lahore" or "alternatives to [competitor]", AI answers typically synthesise **third-party** sources: comparison articles, review platforms, "best of" lists, forums, videos and expert roundups. If you're absent from those sources, you'll rarely be recommended — however good your own site is.

## Find the sources that shape answers in your category

1. **List your priority prompts** (recommendation, comparison, problem/solution, local).
2. **Run them** in several assistants and AI search experiences, several times each, in your target markets and languages.
3. **Record cited sources** and brands mentioned.
4. **Classify sources**: review platforms (G2, Capterra, Trustpilot, Google reviews, Tripadvisor), editorial comparisons, publisher lists, forums and communities (Reddit, Quora, specialist forums), YouTube reviews, industry associations, government or academic sources.
5. **Look for patterns**: which sources are cited repeatedly? Where are competitors present and you're not?

This becomes your **AI source gap analysis** — similar to a backlink gap analysis, but focused on citations and mentions.

## Strategies by source type

| Source type | Legitimate strategy |
|---|---|
| Review platforms | Invite genuine customers to review (no incentives where platforms prohibit them), complete your profile, respond to reviews |
| Editorial comparisons and "best of" lists | Brief journalists and reviewers with accurate facts, offer demos or trials, provide data; accept editorial independence |
| Affiliate comparison sites | Legitimate partnerships with clear disclosure; understand these are commercial listings |
| Forums and communities | Participate transparently as yourself; help answer questions; disclose affiliation |
| YouTube and podcasts | Create useful videos; work with creators (with disclosure where paid) |
| Industry bodies and awards | Membership, accreditation, credible awards (avoid pay-to-win "awards") |
| Expert roundups and media | Digital PR and expert commentary |

## The line you must not cross: astroturfing

Posting fake reviews, creating sock-puppet accounts to praise your brand in forums, paying for undisclosed "organic" mentions, or planting fake Q&A are:

- Violations of platform rules (Reddit, review sites and others ban such manipulation).
- Potentially unlawful: fake reviews and undisclosed paid endorsements are targeted by the US FTC, the UK's DMCC Act and CMA, and advertising rules in many markets.
- Increasingly detectable, and damaging when exposed — communities remember.

Genuine participation is slower but durable: a founder answering questions helpfully in a subreddit under their real name, with disclosure, builds real reputation.

## Comparison content on your own site

Publishing your own "X vs Y" and "alternatives to X" pages is legitimate and useful if they are **fair and accurate**:

- State facts about competitors correctly and date them.
- Acknowledge where competitors are stronger.
- Disclose that it's your comparison.
- Avoid disparaging claims you can't substantiate (comparative advertising rules apply in many jurisdictions).

These pages can be retrieved for comparison sub-queries, but third-party comparisons usually carry more weight for trust.

## Local businesses and AI answers

For local queries, AI features often draw on map data, Business Profile information and reviews, plus local articles. The Local SEO course's fundamentals — accurate GBP, reviews, citations, local coverage — are your AI strategy for local.

## Measuring progress

Track, for your prompt set over time: how often you are mentioned, how often cited, how you are described (accuracy and sentiment), and which sources are cited when you are mentioned. Improvements in third-party presence should precede improvements in AI mentions.

## Illustrative example

A UK SaaS company offering bookkeeping for freelancers finds that AI answers for "best bookkeeping app for freelancers UK" cite three comparison articles, a review platform and a popular forum thread. It's absent from two of the articles and has few reviews. Over two quarters it briefs the article authors with an accurate fact sheet and trial access, runs a compliant review invitation programme, and its product team answers questions in the forum under their real names. Subsequent sampling shows more frequent mentions. (Illustrative scenario; outcomes vary.)

## Hands-on: an AI source gap analysis template

Run each priority prompt 3 times per surface and log the cited or clearly used sources:

```text
prompt_id | surface | market | run | brands_mentioned | our_brand (Y/N) | cited_urls | source_type
P07 | Perplexity | UK | 1 | FreeAgent-like X, Y, Z | N | site-a.com/best-apps; forum/thread/123 | editorial; forum
```

Then pivot: **source domain × number of prompts where it was cited × whether you're present on it**. Domains cited often where you're absent are your targets. Tag each target with the legitimate strategy from the table above and an owner (PR, customer success, product, community).

A spreadsheet formula approach works fine; for larger panels, load the log into Python:

```python
import pandas as pd
log = pd.read_csv("panel_log.csv")          # one row per cited URL per run
log["domain"] = log["cited_url"].str.extract(r"https?://(?:www\.)?([^/]+)")
gap = (log.groupby("domain")
          .agg(prompts=("prompt_id", "nunique"),
               we_are_present=("our_presence_on_domain", "max"))
          .sort_values("prompts", ascending=False))
print(gap[gap.we_are_present == 0].head(20))   # top sources where we're absent
```

(`our_presence_on_domain` is a 0/1 column you fill in once per domain after checking whether the source mentions you accurately.)

## Worked example 2: a Riyadh B2B SaaS goes where the answers look

A Riyadh-based HR software company (illustrative) finds that answers to "best HR software for Saudi companies with GOSI and Mudad integration" cite a regional tech publication's comparison, two software review platforms and an Arabic-language forum thread. It is absent from the comparison and has few reviews. Over two quarters it briefs the publication's reviewer with an accurate, dated fact sheet and demo access, invites genuine customers to review (no incentives where platforms prohibit them), and has its product lead answer integration questions in the forum under her own name with her role disclosed. The team re-runs the panel monthly and reports mention rate and source coverage, not a promised ranking.

## Measuring success

Report per quarter: source coverage (share of frequently cited domains that mention you accurately), mention rate and citation rate on the prompt panel, accuracy of how you're described, and review volume and rating trends on the platforms answers cite.

## Common mistakes

- Focusing only on your own site for recommendation queries.
- Astroturfing forums or buying fake reviews.
- Unfair comparison pages that damage credibility.
- Ignoring non-English sources used in your markets.

## Video lecture: Earning mentions in the sources AI answers draw on

Lecture coming soon · 13 chapters · about 9 minutes. Read the full transcript below.

1. Earning mentions in trusted sources
2. Why third parties matter
3. Find the sources
4. Hands-on: the gap script
5. Strategies by source type
6. The line you don't cross
7. Example 1: UK bookkeeping app (illustrative)
8. Example 2: Riyadh HR SaaS (illustrative)
9. Your own comparisons and local
10. Mistakes and measures
11. Watch me do it: a one-hour gap analysis
12. Timing and expectations
13. Recap and try this now

## Lecture transcript

### Earning mentions in trusted sources

Ask any assistant, what's the best project management tool for a small agency, and look closely at where the answer comes from. It's rarely the vendors' own websites. It's comparison articles, review platforms, forums, videos and expert roundups. If you're absent from those sources, you'll rarely be recommended, no matter how good your own site is. In this lecture you'll learn how to find the sources that shape answers in your category, and how to earn a place in them legitimately.

### Why third parties matter

Why do third parties carry so much weight? Think about how you'd choose a plumber in a new city. You wouldn't just trust the plumber's own website saying they're the best. You'd check reviews, ask a neighbour, maybe read a local forum. Answer engines do something similar at scale. For recommendation, comparison and best-of questions, they synthesise what independent sources say. Your own site still matters for facts, but your reputation is written by others.

### Find the sources

So first, find the sources. List your priority prompts: recommendation, comparison, problem-solution and local. Run each one several times, in several assistants, in your target markets and languages. Record every cited source and every brand mentioned. Then classify the sources: review platforms, editorial comparisons, publisher lists, forums and communities, YouTube reviews, industry associations, government or academic pages. The patterns you'll see are your AI source gap analysis. It's like a backlink gap analysis, but focused on citations and mentions.

### Hands-on: the gap script

The lesson text gives you a logging template and a short pandas script. Each row is one cited URL from one run. The script extracts the domain, counts how many different prompts cited it, and shows the top domains where you're absent. That's your target list. Then you tag each target with a legitimate strategy and an owner. Review platforms go to customer success. Editorial comparisons go to PR. Forums go to whoever can genuinely help, under their own name.

### Strategies by source type

Now strategies by source type. For review platforms, invite genuine customers to review, complete your profile, and respond to reviews, and don't offer incentives where the platform forbids them. For editorial comparisons, brief journalists and reviewers with accurate facts, offer demos or trials, and accept their independence. For affiliate sites, use legitimate partnerships with clear disclosure. For forums, participate transparently as yourself. For YouTube and podcasts, create useful content and work with creators, with disclosure when paid. For awards, choose credible ones, not pay-to-win.

### The line you don't cross

And here's the line you must not cross: astroturfing. Fake reviews, sock-puppet accounts praising your brand in forums, undisclosed paid mentions, and planted questions and answers. These break platform rules. They can be unlawful: fake reviews and undisclosed endorsements are targeted by the US Federal Trade Commission, the UK's consumer protection regime under the DMCC Act, and advertising rules in many markets. And since May twenty twenty-six, Google's spam policies explicitly cover manufacturing signals to manipulate its AI answers. Communities remember. Being exposed costs far more than the shortcut saved.

### Example 1: UK bookkeeping app (illustrative)

Worked example one, simple. A UK bookkeeping app for freelancers finds that AI answers cite three comparison articles, a review platform and a popular forum thread. It's absent from two of the articles and has few reviews. So, over two quarters, it briefs the article authors with an accurate fact sheet and trial access, runs a compliant review invitation programme, and has its product team answer questions in the forum under their real names. Later sampling shows more frequent mentions. Illustrative, and outcomes vary, but the direction is typical.

### Example 2: Riyadh HR SaaS (illustrative)

Worked example two, with illustrative details. A Riyadh HR software company wants to appear for best HR software for Saudi companies with GOSI and Mudad integration. The answers cite a regional tech publication's comparison, two review platforms, and an Arabic-language forum thread. The company briefs the reviewer with a dated fact sheet and a demo, invites genuine customers to review, and its product lead answers integration questions in the forum under her own name, with her role disclosed. They re-run the panel monthly and report mention rate and source coverage, not a promised ranking.

### Your own comparisons and local

What about comparison pages on your own site? They're legitimate and useful if they're fair. State facts about competitors correctly and date them. Acknowledge where competitors are stronger. Make it clear it's your comparison. And avoid claims you can't substantiate, because comparative advertising rules apply in many places. For local businesses, the local SEO fundamentals are your AI strategy: an accurate Business Profile, genuine reviews, consistent citations and local coverage.

### Mistakes and measures

Common mistakes. Focusing only on your own site for recommendation prompts. Astroturfing forums or buying fake reviews. Publishing unfair comparison pages that damage credibility. And ignoring non-English sources in markets where customers search in Arabic or Urdu. How do you measure success? Source coverage, meaning the share of frequently cited domains that mention you accurately. Mention and citation rates on your prompt panel. Accuracy of how you're described. And review trends on the platforms answers cite.

### Watch me do it: a one-hour gap analysis

Watch me do it. I'll build a mini source gap analysis for a UK bookkeeping app in about an hour. Step one: I pick five prompts, like best bookkeeping app for UK freelancers, and alternatives to the market leader. Step two: I run each three times in two assistants, and paste every cited URL into the log sheet with the prompt ID and run number. That's thirty runs and about ninety URLs. Step three: I run the short pandas script from the lesson, which extracts domains and counts how many different prompts cited each. Step four: for the top ten domains, I open each page and search for our brand name. I mark present, absent, or present but wrong. Three editorial comparisons cite us nowhere. One review platform has only four reviews for us. A forum thread is cited in four of five prompts. Step five: I assign owners. PR briefs the three comparison authors with a dated fact sheet and trial access. Customer success starts a compliant review invitation for verified customers. And our product lead answers two genuine questions in the forum thread, under her own name, with her role disclosed. Everything goes in the backlog with a re-sample date.

### Timing and expectations

Let's talk timing and expectations, because this is where clients get impatient. Third-party presence is slow to build. A reviewer might update a comparison article once a quarter. Reviews accumulate over months. Forum reputation takes consistent, genuine participation. And answer engines refresh their indexes at different speeds. So set a two-quarter horizon, and report leading indicators along the way: briefings sent, reviews received, articles updated, forum answers accepted. The general pattern you should expect is that improvements in third-party presence come first, and improvements in AI mentions follow. If a vendor promises you'll be recommended by ChatGPT next month, that's a red flag, not a plan.

### Recap and try this now

Recap. For recommendation and comparison prompts, third parties write your reputation. Find the sources answers cite, identify where you're absent, and earn presence through genuine reviews, briefings, useful content and transparent participation. Never astroturf. Try this now. Take five priority prompts, run each three times in two assistants, and log every cited domain. Circle the domains that appear more than once where you're not mentioned. That's your target list for the next quarter.

## Key takeaways

- Recommendation answers are shaped by third-party reviews, comparisons, forums, videos and media.
- Run an AI source gap analysis: prompts, repeated sampling, cited sources, gaps vs competitors.
- Earn presence legitimately; astroturfing breaches platform rules and consumer law.
- Publish fair, accurate comparison pages, and apply local SEO fundamentals for local AI answers.

## Try it

Run ten recommendation prompts three times each in two assistants, record cited sources, and list the top five source gaps.

- [Previous: Entity clarity and brand consensus](https://optimizeall.com/learn/ai-search-optimization-geo/entity-clarity-and-consensus)
- [Next: AI crawlers and robots.txt: tokens and what they control](https://optimizeall.com/learn/ai-search-optimization-geo/ai-crawlers-and-robots-txt)
- [All lessons of AI Search Optimization: SEO for AI Overviews & Answer Engines](https://optimizeall.com/learn/ai-search-optimization-geo)