When to use this prompt
Quarterly. Also after a major company event: a rebrand, a leadership change, a product launch, a press cycle. The branded SERP is the page a buyer sees first when they verify who you are, and the same sources tend to feed what AI engines say about you. If it is weak, every other piece of marketing is downstream of it.
This is also the right prompt to run before a fundraising round or partnership pitch. Anyone serious will Google your brand name. You should know what they will find.
Capture the results page yourself before you run it. Chat models and research agents cannot reliably render Google’s knowledge panel, AI Overview, or People also ask boxes, so search your brand in a logged-out incognito window and paste the text or attach full-page screenshots. A SERP API export (for example from SerpApi or DataForSEO) works too.
On desktop, capture both columns. Since late October 2026, Google has been observed showing the AI Overview for many brand searches in the right-hand column, where the knowledge panel used to sit, sometimes with knowledge-panel facts like founders and headquarters folded in. Google has not confirmed whether this is a test or permanent, and placement can differ on mobile, so record where the AI Overview appeared rather than assuming it.
The prompt
<role>Search analyst auditing branded SERP results and entity signals. You work from the SERP capture the user provides. Do not reconstruct the results page from training data or from your own searches.</role>
<task>Using the pasted SERP capture for the brand name, catalog every surface that appears on the first results page, in the AI Overview if present, and in the knowledge panel if present. Classify ownership and flag reputation risks.</task>
<inputs>
<brand>[BRAND NAME]</brand>
<domain>[BRAND DOMAIN]</domain>
<known_aliases>[ANY OTHER NAMES THE BRAND USES, e.g., legal name, alternate spellings]</known_aliases>
<capture_context>[Date, location, device, and method, e.g., "2026-10-05, US, desktop, logged-out incognito"]</capture_context>
<serp_capture>
[PASTE THE RESULTS PAGE TEXT, ATTACH SCREENSHOTS, OR PASTE SERP API OUTPUT. Include both desktop columns: the knowledge panel, the AI Overview with its cited sources and where it appeared (main column or right-hand column), the first 10 organic results, People also ask, image results, and any review stars or review sites shown.]
</serp_capture>
</inputs>
<instructions>
1. Use only what appears in <serp_capture> to describe the results page. If you can browse, you may open listed URLs to judge their content, but never add results that are not in the capture.
2. Catalog the following surfaces:
- Knowledge panel (Yes/No, with the specific facts shown)
- AI Overview (Yes/No, its placement: top of the main column, mid-page, or right-hand column, with the cited sources and the summary given)
- Top 10 organic results (URL, page title, page type)
- Image results (top 6 image sources, if captured)
- "People also ask" boxes (every question shown)
- Reviews surface (G2, Trustpilot, Glassdoor, Yelp, Google Reviews if present)
3. For each result, classify ownership:
- OWNED: the brand controls the page
- EARNED: third-party page that is favorable or neutral (positive press, listings, partnerships)
- RISK: third-party page that is unfavorable, outdated, or low quality (negative reviews, defunct profiles, scraper sites, copycat brands)
4. Identify the top 3 reputation risks. For each, name the specific URL and the specific concern.
5. Identify the top 3 entity-authority gaps: things that should appear on the branded SERP but do not (e.g., no Wikipedia entry, no Crunchbase profile, no LinkedIn company page in top 10).
6. Do not fabricate results. If the capture shows fewer than 10 organic results, or a surface was not captured, say so rather than filling it in.
</instructions>
<output_format>
**Knowledge panel:** [Yes / No]. [Facts shown if yes.]
**AI Overview:** [Yes / No]. [Placement, cited sources, and summary if yes.]
**Top 10 organic results:**
| # | URL | Title | Type | Ownership |
|---|-----|-------|------|-----------|
**Image results dominant sources:** [comma-separated, or "not captured"]
**People also ask:**
- [Question 1]
- [Question 2]
...
**Top 3 reputation risks:**
1. [URL]: [concern]
2. ...
3. ...
**Top 3 entity-authority gaps:**
1. [missing surface, e.g., "No Wikipedia entry"]
2. ...
3. ...
**Highest-leverage fix:** [the one move that would most clean up the SERP]
</output_format>How it works
The branded SERP is the entity verification surface for humans, and a good proxy for what AI engines find about you. When an AI engine answers “who is [brand]” with web search turned on, it pulls from the same profiles, listings, and press that rank for your name. When it answers from memory, it reflects what those sources said when the model was trained. Either way, what surfaces on your branded SERP shapes what gets summarized.
The prompt works from a capture you provide, not a live search, because chat models and research agents cannot reliably see Google’s knowledge panel, AI Overview, or People also ask boxes. Asked to “search Google,” they tend to produce a believable but invented results page. Pasting the real page, or attaching screenshots, keeps the audit grounded.
The OWNED / EARNED / RISK classification forces the model to evaluate ownership rather than just listing URLs. Most branded SERP audits collapse into “here are 10 links,” which is not an audit. The classification surfaces the work that needs to happen: claim missing profiles, suppress risks, fill authority gaps.
The “do not fabricate results” line is essential because models will pad branded SERP audits with plausible-sounding entries when the real SERP is sparse. The explicit instruction keeps the audit honest.
Example output
Knowledge panel: No.
AI Overview: Yes, in the right-hand column (desktop). Cited sources: brand homepage, LinkedIn company page, TechCrunch coverage. Summary: “[Brand] is a mid-market product analytics platform founded in 2022.”
Top 10 organic results: [populated table]
Top 3 reputation risks:
- abandoned-startup-directory.com/[brand]: outdated 2023 listing with wrong pricing, ranks #6
- competitor-comparison.io/[brand]-alternatives: competitor-owned page, ranks #4
- crunchbase.com/[brand]: last updated 2023, missing recent funding
Top 3 entity-authority gaps:
- No Wikipedia entry
- No Wikidata entity (sameAs targets are limited)
- No Trustpilot profile claimed
Highest-leverage fix: Claim and refresh the Crunchbase profile. It already ranks in the top 5, so outdated funding and pricing data there is what buyers and AI engines are most likely to repeat.
Variations
- Founder SERP audit: Run on a founder’s personal name. Critical for personal brand-driven companies; the founder’s SERP is often the first verification step for buyers.
- Competitor audit: Run on a competitor’s brand to benchmark what a strong vs weak branded SERP looks like in your category.
- Pre-launch audit: Run before a brand or product launch to identify squatters and lookalikes.