Five AI firms, and the distance between famous and legible
OpenAI leads Signal Strength at 9.01 and ranks second on Brand Integrity at 8.57. The strongest signal system in the set is not the clearest source story.
Evaluated
5 organizations- OpenAI Signal Strength
- 9.01
- Anthropic Brand Integrity
- 8.81
- OpenAI Signal Gap
- +0.63
- AI answer eligibility
- 8.50
1st of 5 — strongest machine-readable ecosystem.
1st of 5 — safety, research and Claude reinforce one story.
Signal outperforms the source-level brand story.
Average across eight prompt journeys for OpenAI.
Summary
Corporate AI sites now have to satisfy consumers, developers, procurement teams, journalists, regulators and AI systems from the same source material. This study scored OpenAI against Anthropic, Gemini, Microsoft Copilot and Lovable on all fifteen dimensions, using direct site evidence only — official homepages, product and pricing pages, developer documentation, trust and security materials and support surfaces. OpenAI's owned ecosystem is category-leading structurally, but Anthropic and Lovable publish clearer source-level positions. Gemini and Copilot show a parent-ecosystem pattern: very high entity authority and channel reach, with the primary product story spread across multiple official surfaces.
What was measured
- All fifteen dimensions — seven Brand, eight Signal — scored identically across five sites.
- Official homepages, product, business, pricing and developer-documentation surfaces.
- Trust, safety, security and transparency pages as corporate proof.
- Support and response pathways for consumer and enterprise users.
- Eight AI answer journeys, used to validate the Signal scores rather than replace them.
Results
| Dimension | OpenAI | Anthropic | Gemini | Copilot | Lovable |
|---|---|---|---|---|---|
| Brand Entity Presence | 9.5 | 9.1 | 9.2 | 9.0 | 7.5 |
| Page Identity | 8.0 | 8.8 | 6.6 | 6.8 | 8.5 |
| Audience Language | 8.2 | 8.5 | 7.4 | 7.8 | 9.0 |
| Visual Brand | 8.7 | 8.5 | 7.7 | 8.0 | 8.8 |
| Usefulness | 8.4 | 8.6 | 7.6 | 8.3 | 8.6 |
| Brand Voice | 8.2 | 9.2 | 6.8 | 6.8 | 9.0 |
| Content Vitality | 9.0 | 9.0 | 8.2 | 8.5 | 8.3 |
| Average | 8.57 | 8.81 | 7.64 | 7.89 | 8.53 |
| Dimension | OpenAI | Anthropic | Gemini | Copilot | Lovable |
|---|---|---|---|---|---|
| AI & Search Accessibility | 9.2 | 8.6 | 8.8 | 8.4 | 8.1 |
| Engineering Strength | 8.8 | 8.5 | 8.4 | 8.4 | 7.8 |
| Offering Architecture | 8.8 | 8.5 | 7.7 | 7.8 | 8.7 |
| Machine Content & Context | 9.3 | 8.8 | 8.2 | 7.8 | 8.3 |
| Channel Reach | 9.2 | 8.4 | 9.3 | 9.2 | 8.2 |
| Third-Party Verification | 9.5 | 9.0 | 9.4 | 9.2 | 7.7 |
| User Action Pathways | 8.7 | 8.4 | 7.8 | 8.3 | 8.8 |
| Response Infrastructure | 8.6 | 8.7 | 8.3 | 8.8 | 8.4 |
| Average | 9.01 | 8.61 | 8.49 | 8.49 | 8.25 |
Structural presence only. Signal scores are provisional pending crawler tests and structured-data verification.
| Rank | Brand Integrity | Signal Strength |
|---|---|---|
| 1 | Anthropic 8.81 | OpenAI 9.01 |
| 2 | OpenAI 8.57 | Anthropic 8.61 |
| 3 | Lovable 8.53 | Gemini 8.49 |
| 4 | Copilot 7.89 | Copilot 8.49 |
| 5 | Gemini 7.64 | Lovable 8.25 |
Eight prompt journeys, used as a diagnostic layer on the Signal scores.
| Journey | Intent | Result | Score |
|---|---|---|---|
| Category discovery | Best AI assistant or platform for work | Appears, but Anthropic and Microsoft often also qualify. | 8.7 |
| Comparison | OpenAI vs Anthropic for safety and coding | Strong; depends on current model benchmarks and policy pages. | 8.5 |
| Recommendation | Best AI for enterprise teams | Strong, but stack incumbency favours Microsoft and Google. | 8.3 |
| Transactional | How do I get started? | Clear ChatGPT, API and business paths. | 9.0 |
| Trust & proof | Is OpenAI reputable and safe? | Strong official and third-party proof. | 8.4 |
| Support intent | What if I have a problem? | Enterprise support is clearer than consumer support. | 7.8 |
| Contextual intent | Best AI platform for developers | Very strong, on documentation and API ecosystem. | 9.1 |
| Differentiation | What makes OpenAI different? | Strong entity story; homepage understates it. | 8.2 |
Average 8.50 · answer risk low-to-medium. The risk is comparison ambiguity, not absence.
Findings
- OpenAI wins Signal because its owned ecosystem exposes strong developer, business, safety, support and trust surfaces designed for retrieval.
- Anthropic wins Brand because the safety-first research story, Claude product surfaces and enterprise materials reinforce a single coherent narrative.
- Lovable is the clearest challenger: builder language, fast action pathways, pricing, docs and security are unusually legible — but entity depth and third-party verification trail the giants.
- Gemini and Copilot carry parent-ecosystem drag: high authority and broad channels, fragmented product boundaries.
- The shared category gap is comparison readiness. No site in the set makes it easy for a human or a machine to explain when to choose one AI platform over another.
What the sector should fix first
- Publish one canonical products-and-plans source page mapping every product, plan and API to a use case and a next step.
- Give the homepage a source-level identity layer: who we serve, what you can do, where to start — before handing off to product surfaces.
- Publish comparison-ready pages and FAQs for high-intent questions rather than leaving the comparison to third parties.
- Consolidate support and enterprise pathways so one journey does not cross four properties.
- Keep safety, trust and security material as first-class corporate proof, not footnotes.
Limitations
- Publicly observable evidence only. Private analytics, internal process and unpublished work are outside scope.
- Point-in-time measurement. Scores reflect the evidence visible on the evaluation date.
- Scores are comparable within this study and this methodology version only.
- Phase 1 snapshot: no crawler logs, page-by-page structured-data inventory or Core Web Vitals lab data.
- Brand Integrity and Signal Strength are reported separately and are never blended into one composite.
Sources, inclusion and corrections
- Kumu Brand & Signal Benchmark, AI Firms Phase 1 snapshot, September 2026.
- Official site fetches for openai.com, anthropic.com, gemini.google.com / Google Cloud AI, copilot.microsoft.com / Microsoft 365 Copilot and lovable.dev; reviewed 13 September 2026.
- Named baselines: Google Search Central, Core Web Vitals / Lighthouse, WCAG, Nielsen Norman Group, Baymard Institute, structured data and freshness signals.
Inclusion in a Brand & Signal benchmark is not an endorsement, and the organizations evaluated are not clients, sponsors or partners of the benchmark. Evaluation is unsolicited and based on publicly observable evidence.
Any organization may request a correction. We review the evidence, and where a factual error is confirmed we republish with a dated change note. Corrections address factual accuracy only — scores cannot be purchased, negotiated or removed, and no commercial relationship changes a result.