LLMrefs

1. What LLMrefs is and where it fits in GEO

LLMrefs is a keyword-first AI-search analytics platform. Its homepage and About page position it for brands, agencies, and SEO practitioners that want to understand how often a brand appears in generative answers, how it compares with competitors, and which pages or domains are cited. The product sits in the measurement layer of GEO: it collects observations and turns them into monitoring evidence rather than publishing website changes on a buyer’s behalf.

The keyword-first model is the clearest distinction. Teams enter the familiar SEO topics they care about; according to the vendor, LLMrefs generates prompt variations and fan-out queries, runs them across selected answer engines, and aggregates responses, citations, and brand data. This can lower the setup burden for an SEO team moving into GEO. It can also hide important prompt-level choices unless the team inspects the underlying responses, so procurement should test both the summary view and the evidence behind it.

2. Core product areas

These are vendor-described capabilities, not independently benchmarked outcomes. The public API reference is useful evidence for the data boundary: keyword-detail responses can include rankings, share of voice, average position, sources, raw responses, and brands, while authenticated endpoints are limited by default to ten requests per minute. It does not establish completeness, uptime, or historical retention for a particular plan.

3. Measurement model and how teams use it

The vendor says the platform repeatedly converts a keyword into relevant prompt variations, collects answers from multiple engines, and weights the results into comparable brand rankings. Its methodology description says responses are aggregated across prompts and engines and normalized to reduce outlier effects. The homepage also presents share of voice and position rather than relying only on one composite score.

A practical workflow is to begin with a small keyword set tied to a real buying or research question, inspect the generated prompts, name relevant competitors, and then review rankings alongside the captured answers and sources. Teams can use changes as investigation triggers: a declining share of voice can lead to a prompt review; a competitor-only citation can lead to source analysis; an engine-specific gap can lead to market or crawlability checks. CSV or API access can then move dated observations into reporting.

The public documentation does not disclose the full sampling design, weighting formula, retry policy, or per-engine collection method. “Statistical significance” is therefore a vendor claim, not a result GEO Wiki can reproduce from public material. A third-party Reboot Online experiment independently shows LLMrefs being used to run consistent prompts across 11 models and store historical answers, but that experiment validates a workflow example—not the product’s accuracy across all engines.

4. Engine coverage and pricing

As checked on 13 August 2026, the official homepage names ChatGPT, ChatGPT Search, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Claude, Grok, Microsoft Copilot, Meta AI, and DeepSeek. The vendor says engine access is included without additional engine fees and that every keyword is refreshed at least weekly. Because answer-engine interfaces and plan coverage change quickly, buyers should confirm the exact engine, country, language, and collection surface they need.

The live offer is an All in One plan at $79/month, labeled “limited time only,” with a seven-day free trial. The plan block advertises 500 prompts, all engines, weekly reports, citation tracking, geo-targeting in 50+ countries and 20+ languages, unlimited projects and team members, CSV export, API access, and supporting utilities. The same page offers a no-credit-card free account path, but it does not clearly state the durable free allowance in the pricing block.

One limit needs explicit reconciliation: the plan block foregrounds 500 prompts, while the FAQ asks whether customers can buy more than 50 keywords. Those may be complementary units, but the public page does not show the conversion between them. Buyers should request a worked example using the intended number of keywords, engines, countries, languages, prompt variations, and refreshes before comparing effective cost with another vendor.

5. Fit and buying considerations

LLMrefs appears best aligned with SEO-led teams that value a familiar keyword workflow and broad self-serve coverage. Independent buyer guides from Trakkr and Generate More also characterize it as accessible to smaller agencies and teams, while raising questions about enterprise depth and diagnostic granularity. These sources sell adjacent or competing services, so their judgments are useful questions for evaluation, not neutral performance findings.

6. Company, founder, and dates

LLMrefs publicly identifies James Berry as founder and CEO on its homepage and in his official author profile. The About page gives a London contact address, and the privacy policy says the service is based in the UK. The public legal pages reviewed do not name a legal operating entity, so this profile does not invent a company name or treat the contact address as independently verified incorporation evidence.

The authoritative Verisign RDAP record records the current llmrefs.com domain’s registration on 13 January 2025. The company-controlled LinkedIn profile lists a 2025 founding year. These facts concern different events and neither proves the product’s launch date.

Public launch references conflict. Aitoolnet labels the product as launched in January 2025 and says its editorial team first featured it on 1 May 2025; Dealroom lists an April 2025 launch. Both are aggregators, and no official page reviewed gives a precise launch date. The structured launch field therefore remains unknown, while 1 May 2025 is recorded only as the earliest independently dated public product evidence found in this research pass.

7. Editorial assessment

LLMrefs offers a clear entry point for teams that think in SEO keywords but need visibility and citation evidence from answer engines. Broad named engine coverage, unlimited projects and seats in the public offer, source inspection, CSV export, and a documented API make it plausible for agency research and in-house monitoring. The strongest evaluation path is to trace every summary metric back to raw responses and cited URLs, then test how the same keyword behaves across prompts, engines, markets, and dates.

The main risks are evidence and procurement maturity. Public pages make strong data-quality claims but do not publish enough sampling detail for independent reproduction; plan language mixes prompt and keyword units; and the legal and security surface omits information many enterprise buyers expect. A May 2026 Analyze AI comparison also questions the link between visibility and traffic or conversion outcomes. That criticism comes from a competitor, but the underlying buyer question is sound: decide whether the program needs visibility evidence alone or must connect exposure to site behavior and revenue.

Treat LLMrefs as an observable measurement system, not as proof that an optimization caused an engine response to change. Before buying, run a fixed manual sample, compare it with the platform’s raw evidence, document expected refresh cadence, and confirm the contract and security terms for the intended reporting workflow.