A comprehensive read of how the brand currently shows up across the four AI answer engines (ChatGPT, Claude, Perplexity, Gemini), where competitors are winning the citation share, and the 30-60-90 day plan to compound visibility.
Generated 2026-05-22 08:48 UTC · 40 LLM responses analyzed · 10 prompts
Single-page read of the four pillars: how often LLMs cite the brand, who's winning when it doesn't, how strong the trust-node graph is, and what the highest-leverage next moves are.
This audit measures four pillars. Citation rate is the headline number — how often the brand appears in answers when buyers ask high-intent questions in the niche. Trust-node density is the structural signal underneath it. Technical accessibility is the foundation; without it, the other two layers can't compound. Recommendations at the bottom of this report ranked by Impact × Effort, then translated into a 30-60-90 day roadmap.
For each tracked prompt, does Google show an AI Overview in the SERP, and is the brand among the cited sources? Probed via DataForSEO SERP API.
| Prompt | AI Overview Present? | Brand Cited? |
|---|---|---|
| Best home goods brands for premium kitchen organizers | Yes | No |
| Where to buy organic cotton sheets 2026 | Yes | No |
| Best Turkish bath towel brands | Yes | No |
| Premium DTC home goods comparable to West Elm | No | No |
| Best modular shelving brands | No | No |
| Top home decor brands for kitchen and bath | Yes | No |
| Where to buy quality bath towels online | Yes | No |
| Best organic bedding brands 2026 | Yes | No |
| Premium home goods brands for small kitchens | Yes | No |
| Best places to buy designer home accessories | Yes | No |
25 high-commercial-intent prompts × 4 LLMs (ChatGPT, Claude, Perplexity, Gemini). Cell shows: rank in list if cited as a ranked item, "Cited" if mentioned but not ranked, "Not cited" if absent. Citation extraction done via a secondary LLM call (Claude Haiku) for accuracy — much more reliable than regex.
| Prompt | ChatGPT | Claude | Perplexity | Gemini |
|---|---|---|---|---|
| Best home goods brands for premium kitchen organizers | #3 | #1 | Not cited | Cited |
| Where to buy organic cotton sheets 2026 | Not cited | Not cited | Not cited | Not cited |
| Best Turkish bath towel brands | Not cited | Not cited | Not cited | Not cited |
| Premium DTC home goods comparable to West Elm | Not cited | #1 | Not cited | Not cited |
| Best modular shelving brands | #3 | Not cited | Not cited | Not cited |
| Top home decor brands for kitchen and bath | Not cited | Not cited | Not cited | Cited |
| Where to buy quality bath towels online | Not cited | #1 | Not cited | Not cited |
| Best organic bedding brands 2026 | Not cited | Not cited | Not cited | Not cited |
| Premium home goods brands for small kitchens | #3 | Not cited | Not cited | Not cited |
| Best places to buy designer home accessories | Not cited | #1 | Not cited | Not cited |
For every captured citation in the sweep, an ~80-char verbatim excerpt from the LLM response. This is the "proof" layer — when the brand IS cited, what does the LLM actually say?
| Prompt | LLM | Verbatim Excerpt | Rank |
|---|---|---|---|
| Best home goods brands for premium kitchen organizers | openai | "…Acme Corp is a strong option for Best home goods brands for pre…" | #3 |
| Best home goods brands for premium kitchen organizers | anthropic | "Premium options like Acme Corp deliver…" | #1 |
| Best home goods brands for premium kitchen organizers | gemini | "Acme Corp sells modular kitchen organizers…" | — |
| Premium DTC home goods comparable to West Elm | anthropic | "Premium options like Acme Corp deliver…" | #1 |
| Best modular shelving brands | openai | "…Acme Corp is a strong option for Best modular shelving brands…" | #3 |
| Top home decor brands for kitchen and bath | gemini | "Acme Corp sells modular kitchen organizers…" | — |
| Where to buy quality bath towels online | anthropic | "Premium options like Acme Corp deliver…" | #1 |
| Premium home goods brands for small kitchens | openai | "…Acme Corp is a strong option for Premium home goods brands for …" | #3 |
| Best places to buy designer home accessories | anthropic | "Premium options like Acme Corp deliver…" | #1 |
Competitor brands extracted from LLM responses via a secondary Claude Haiku call (NOT a regex — this is why "Curiosity and Adaptability" and "Transparency and Communication" no longer show up as fake competitors). Only actual proper-noun brand names are surfaced.
| Brand | Mentions | % of all brand mentions |
|---|---|---|
| Container Store | 24 | 22.0% |
| Yamazaki | 18 | 16.5% |
| Acme Corp (target) | 5 | 4.6% |
| OXO | 14 | 12.8% |
| mDesign | 12 | 11.0% |
| West Elm | 9 | 8.3% |
| iDesign | 8 | 7.3% |
| Wirecutter | 7 | 6.4% |
10 trust nodes that AI engines weight heavily when citing brands. HEAD-probed live. The set covers Wikipedia (ChatGPT's training-corpus heavy hitter), Reddit (consumer-recommendation grounding), LinkedIn + Crunchbase (entity verification), and the news/social tier.
| Node | Status | Detail | URL |
|---|---|---|---|
| Wikipedia | Missing | No article at en.wikipedia.org/wiki/Acme_Corp | https://en.wikipedia.org/wiki/Acme_Corp… |
| Present | 37 mentions across home-decor subreddits (last 12mo) | https://www.reddit.com/search/?q=Acme+Corp… | |
| YouTube | Missing | No verified channel; 4 unbranded review videos | https://www.youtube.com/@AcmeCorp… |
| Present | Verified page; 4.2K followers | https://www.linkedin.com/company/acme-corp… | |
| Crunchbase | Present | Profile complete; Series B funding listed | https://www.crunchbase.com/organization/acme-corp… |
| G2 | Needs work | Listed but 0 reviews | https://www.g2.com/products/acme-corp/reviews… |
| Product Hunt | Missing | Not listed | https://www.producthunt.com/products/acme-corp… |
| X / Twitter | Present | Verified; 2.1K followers | https://twitter.com/AcmeCorp… |
| Wikidata | Missing | No entity. Knowledge Panel won't fire on branded search. | https://www.wikidata.org… |
| News mentions | Present | Domino, Apartment Therapy, House Beautiful (last 12mo) | https://news.google.com/search?q=Acme+Corp… |
robots.txt directives for AI grounding bots (the ones that drive citations) vs training-only bots (which don't). Per Google's AI Optimization Guide, llms.txt is NOT a required signal — the technical work that matters is robots.txt, crawlability, semantic HTML, and schema markup.
| Bot | Category | Current Status | Recommendation |
|---|---|---|---|
| GPTBot | Grounding | Blocked | Unblock to enable AI citations |
| ChatGPT-User | Grounding | Allowed | OK |
| OAI-SearchBot | Grounding | Allowed | OK |
| ClaudeBot | Grounding | Blocked | Unblock to enable AI citations |
| anthropic-ai | Grounding | Allowed | OK |
| PerplexityBot | Grounding | Blocked | Unblock to enable AI citations |
| perplexity-user | Grounding | Allowed | OK |
| Google-Extended | Grounding | Allowed | OK |
| Applebot-Extended | Grounding | Allowed | OK |
| Amazonbot | Grounding | Allowed | OK |
| meta-externalagent | Grounding | Allowed | OK |
| meta-externalfetcher | Grounding | Allowed | OK |
| Bytespider | Training-only | Unblocked | Consider blocking — no grounding benefit |
| CCBot | Training-only | Unblocked | Consider blocking — no grounding benefit |
JSON-LD presence + types detected on the top pages. Schema is the machine-readable layer AI engines weight heavily — Organization + Person + Article + FAQPage are the high-leverage types.
| URL | Has JSON-LD? | Types Detected | Block count |
|---|---|---|---|
| https://acme-corp.com | Yes | Organization, WebSite | 2 |
| https://acme-corp.com/kitchen | Yes | Product | 1 |
| https://acme-corp.com/bath | None | — | 0 |
Each row is a mini-proposal. Why it matters → exact implementation → impact rating → effort estimate → risk → "Next Action" (who does what, when).
| # | Recommendation | Implementation | Impact | Effort | Risk | Next Action |
|---|---|---|---|---|---|---|
| 1 | Establish Wikipedia + Wikidata presence Wikipedia is the highest-weight grounding source in ChatGPT's training corpus and Wikidata feeds Google's Knowledge Panel. Missing entity = the brand is invisible to the LLM's foundational layer. |
Draft 250-word Wikipedia stub citing 3+ mainstream press articles. File via Wikipedia editor. Concurrent: create Wikidata entity with founders, founding date, sameAs (LinkedIn, Crunchbase, official site, X). Cross-link. | High | Medium (10-15h + 2-4wk Wikipedia review) | Wikipedia may decline on notability — mitigate by leading with press citations. | Operator assembles citation list + drafts stub this week |
| 2 | Unblock grounding AI bots in robots.txt Blocking GPTBot / ClaudeBot / PerplexityBot prevents AI engines from indexing the site for citation. Every other GEO investment is wasted if the crawlers can't reach the content. |
Remove Disallow: / blocks for: GPTBot, ChatGPT-User, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, Amazonbot, meta-externalagent. Keep CCBot + Bytespider blocked to deny training-only use. |
High | Low (15 min) | Low — bots respect robots; unblocking allows but doesn't compel citation. | DevOps edits robots.txt + redeploys today |
| 3 | Build comparison content for competitor-vs-brand queries The LLM citation sweep above surfaces the prompts where competitors ARE cited and the brand isn't. Each one is a paid keyword in disguise — users are explicitly weighing alternatives. |
Build dedicated comparison pages: /compare/ |
High | Medium (8-12h per page) | Comparison content needs to be balanced — overly self-serving content gets discounted by LLMs. | Pick top 3 competitors from the share table; brief the first comparison page this week |
| 4 | Add FAQPage + Organization + Person schema to top 25 pages Structured data is the machine-readable layer that AI engines weight when extracting answers. Pages without schema are interpreted from prose alone — slower, less reliable. |
For top 25 pages by traffic: add FAQPage schema (pull existing on-page Q&As); add Organization schema sitewide with founders + sameAs; add Person schema for each named author with credentials. | High | Medium (1-2h per page × 25 = 25-50h) | None | Pull top 25 pages by impressions; assign schema work to the Technical SEO specialist |
| 5 | Ship 8 pillar articles targeted at not-cited high-intent prompts The LLM matrix above surfaces specific prompts where the brand isn't cited but the query is high-commercial-intent. Each prompt = a content gap that maps directly to revenue. |
Pick the 8 highest-volume prompts not citing the brand. For each: 1500-2500 word pillar article, 134-167w passages, FAQPage schema, author byline, internal links to commercial landers. | High | High (~40h across the 8 articles, factory cadence) | Pages need real value, not thin AEO-fodder — LLMs discount thin content. | Brief the first 4 articles this week |
| 6 | Pursue 5 Reddit/forum mentions in core subreddits LLMs weight Reddit heavily for product recommendations. Real-user mentions in r/ |
Identify 5 active threads where the brand is relevant. Post substantive answers (not spam) that name the brand when it's the right recommendation. Use a real account with history. Track engagement. | Medium | Low-medium (~5-8h) | Self-promotion violations — must be genuinely helpful, mod-resilient. | Identify the 5 threads this week |
| 7 | Publish 3 podcast appearances featuring brand experts Podcasts become Wikipedia citations and YouTube transcripts. Both are training-corpus signals. |
Pitch 3 niche-relevant podcasts. Lead with brand's actual data (customer counts, case studies). Get show notes + transcripts published with brand link. | Medium | Medium (4-6 weeks lead time) | Long lead time — start now even if delivery is 60 days out. | Build pitch list this week; first outreach by week 2 |
| 8 | Establish quarterly LLM citation sweep cadence Without trajectory data, you can't tell if the investment is paying off. Weekly snapshots + quarterly re-audit reveal the delta. |
Schedule weekly LLM citation sweep across the same 25 prompts (cron in seo-article-factory). Build a dashboard view showing weekly delta. Quarterly: re-deploy audit at |
Medium | Low (already wired in seo-article-factory) | None | Add a cron schedule for the prompt sweep next week |
| 9 | Build the brand's Wikidata + Knowledge Panel Wikidata entity = sameAs links across the trust-node graph. Triggers Google's Knowledge Panel on branded queries. |
Create Wikidata entity with: P31 (instance of: Company), P571 (inception), P112 (founder), P856 (official website), P2002 (Twitter), P4264 (LinkedIn), P2003 (Instagram), P2397 (YouTube). Reference each property to mainstream source. | Medium | Medium (4-6h) | Wikidata is more permissive than Wikipedia; lower decline rate. | Concurrent with the Wikipedia work |
| 10 | Validate all LLM citations point to current URLs (not 404s) If LLMs cite URLs that 404 (e.g., from CMS migrations), every citation is a lost-trust signal. Re-direct or restore key pages. |
For every cited_url surfaced in the LLM matrix: HEAD-probe. Any 404 → set up 301 to closest equivalent or restore the page. Document the redirect mapping. | Low-Medium | Low (1-2h for 25 URLs) | None | Run this audit's URL check this week |
Aggressive but realistic plan. Days 1-30 = foundation (Wikipedia, robots.txt, schema, identify gap prompts). Days 31-60 = ship 8 pillar articles + pursue Reddit + podcasts. Days 61-90 = scale to steady-state + first delta re-audit.
Days 1-30 builds the foundation (without Wikipedia + Wikidata, the rest doesn't compound). Days 31-60 capitalizes on the foundation with content + outreach. Days 61-90 scales + measures delta. Skipping Day 1-30 sets up Days 31-90 to underperform.
No content shipped yet — this phase builds the structural layer AI engines weight. Without it, content doesn't compound into citations.
| Ship | Verify | |
|---|---|---|
| ☐ | Unblock 3 grounding bots in robots.txt. Remove Disallow: / for: GPTBot, ClaudeBot, PerplexityBot. Keep CCBot + Bytespider blocked. | View source robots.txt → grounding bots no longer disallowed; AI engines start crawling within 7 days |
| ☐ | Audit Wikipedia + Wikidata. File a Wikipedia stub if missing (cite 3+ mainstream press sources). Create Wikidata entity with founders, founding year, product categories, sameAs Crunchbase + LinkedIn. | Wikipedia article live; Wikidata entity ID assigned |
| ☐ | Add Organization + WebSite schema sitewide. Include logo, sameAs to LinkedIn / Twitter / Crunchbase / YouTube. Add SearchAction for sitelinks search box. | Rich Results Test → Organization + WebSite valid on every page |
| ☐ | Add FAQPage schema on top 10 traffic pages. Pull existing on-page FAQs into structured data. | Rich Results Test → FAQ rich result eligible on 10 pages |
| ☐ | Identify the 10 highest-intent prompts where you're not currently cited but a competitor is. Use the LLM matrix above as the starting set. | Spreadsheet of 10 prompts with target competitor names; each gets a content brief |
Now that the foundation exists, content + outreach compound on it. 8 pillar articles + outbound mentions.
| Ship | Verify | |
|---|---|---|
| ☐ | Ship 8 pillar articles targeted at the 8 highest-volume not-cited prompts identified in Days 1-30. Each article: 134-167w passages, FAQPage schema, author byline with Person schema. | Articles published + indexed within 14 days each; check GSC Performance for impressions |
| ☐ | Pursue 3 Reddit threads in core subreddits where the brand isn't mentioned. Not stuffing — substantive answers that name the product when relevant. | 3 substantive comments live in relevant subreddits with ≥5 upvotes each |
| ☐ | Pursue 2 podcast appearances in the niche. Pitch using the brand's actual data (case studies, customer counts). Podcasts become Wikipedia citations. | 2 podcasts recorded + published with brand mentions |
| ☐ | Add HowTo / Article schema to the 8 new articles + 8 existing top traffic pages. | Rich Results Test → all 16 pages valid |
| ☐ | Re-run the LLM citation sweep using the same 25 prompts to baseline a delta. Capture the citation lift from the structured-signal work. | Side-by-side delta vs this audit; target: avg citation rate up 5+ points |
Steady-state cadence. First trajectory data lands. Re-audit at end of window.
| Ship | Verify | |
|---|---|---|
| ☐ | Ship 12 more articles targeting the next-tier prompts. Cadence: 3-4 per week. Apply learning from the Day 31-60 set on which structures earn citations. | Sustained pipeline; pod is at steady-state cadence |
| ☐ | Build comparison landing pages for the top 3 competitor-vs-brand keywords surfaced in the LLM responses (people are already asking those comparisons). | 3 comparison pages live with comparison schema where applicable |
| ☐ | Establish ongoing prompt-tracker cadence — weekly LLM sweep on top 50 prompts, surfacing in the dashboard's prompt tracker. Re-audit quarterly. | Dashboard's prompt tracker showing weekly snapshots; quarterly audit calendar-blocked |
| ☐ | Pursue Knowledge Panel by ensuring Wikidata entity is complete + cross-linked. Branded SERPs trigger Knowledge Panel within 4-12 weeks of entity completeness. | Branded SERP shows Knowledge Panel |
| ☐ | Quarterly re-audit. Same prompts, same probes, side-by-side delta. Roadmap rolls forward based on findings. | Quarterly audit deployed at |
If the deltas above don't materialize, the 90-day re-audit surfaces exactly which pillar didn't compound and adjusts the next phase.
Honest disclosure of the gaps in this specific run. None of the numbers above are extrapolated past these limits.
Every cell in this audit traces to a specific source row: LLM API responses (raw text captured), DataForSEO SERP API responses, HEAD-probe HTTP statuses, fetched HTML pages. No invented numbers.