AI already knows Bibado is an award‑winning category leader. It just never brings that knowledge to the questions parents actually ask.

Name Bibado and AI knows them cold. Ask about a product and it’s a coin flip. Ask the question a parent actually types — the problem, not the product — and they are effectively absent. Blended across every unbranded prompt (category and product combined), Bibado is named 29% of the time.

Source: AI Answer Snapshot — official APIs, 150 answers, GB, 8 Sep 2026

The raw numbers looked dramatic: answers that searched the live web named Bibado 61% of the time; answers from training memory alone, 15%. That would read as “corroboration works, brand memory doesn’t” — except it doesn’t survive the controls, so we’re not using it.

What survives: engines choose to search on commercial queries and answer from memory on informational ones. Bibado’s presence tracks the question bucket, not whether the engine happened to search.

Across all 150 answers, Reddit is cited nearly twice as often as Bibado’s own site. Reddit plus Mumsnet together (90 citations) outweigh every publisher combined. In this category, AI leans on parents talking to each other — not on brands, and not even on journalists.

Source: AI Answer Snapshot — official APIs, 150 answers, GB, 8 Sep 2026. Gemini’s vertexaisearch redirect wrapper is excluded — it masks the real domain. uk.trustpilot.com + trustpilot.com total 20 combined.

• An “authority & citation building” workstream aimed only at journalists targets the wrong surface — community presence is the bigger lever, and Bibado has thousands of genuine advocates to work with.
• bibado.co.uk sits at #2 in the citation table above — they are not invisible. Their own site is one of the most‑cited sources in the category. They simply don’t win the unbranded answer.
• John Lewis (14) and the Evening Standard (13) are live citation surfaces; Trustpilot (20 combined) confirms review platforms feed AI answers directly.

On 11 of 50 prompts, one engine names Bibado on the identical question that another misses. That is a far more actionable position than invisibility — small movements in the citation layer should flip several of these. Eight of the eleven, for illustration:

Source: AI Answer Snapshot — official APIs, GB, 8 Sep 2026. Table shows 8 of the 11 split prompts on record.

Across 24 answers where a named competitor appears and Bibado doesn’t, the pattern is consistent: Bibado holds the bib category reasonably well, then loses the entire cutlery / tableware / cup range — the exact attach‑and‑cross‑sell products with higher order value, and the ones whose developmental story is closest to Bibado’s own positioning. ezpz (23 mentions) and Olababy (11) own the self‑feeding tools conversation.

Share of voice across all 150 answers — Bibado’s total is the highest of any brand, but it’s inflated by brand‑name prompts. Strip those and the picture reverses:

Source: AI Answer Snapshot — official APIs, 150 answers, GB, 8 Sep 2026

A UK‑indexed source audit — reading the actual pages behind the roundups, not just the generated answer — found Bibado present in every readable result, often near the top:

Source: UK retrieval‑surface audit — direct source reads (WebFetch), partial sample, 8 Sep 2026. Four further roundups (momjunction, littlebabygear, babycarestudio, theartofparenthood) returned unreadable pages (406/503/empty) — Bibado’s position there is TBC, not assumed.

Direct inspection of both domains — robots.txt, hreflang, canonicals and structured data — separates what’s already working from what isn’t.

Source: live site inspection (curl) of bibado.co.uk and bibado.com, 8 Sep 2026

• Bibado owns its brand space and is marginal‑to‑absent in unbranded discovery.
• The informational layer — where parents actually start — sits at 7%.
• Reddit and Mumsnet are the dominant citation surface in this category, ahead of every publisher.
• The cutlery / tableware / cup range is being lost outright, not just underperforming.
• Presence is threshold‑level on 11 prompts, not zero — so it should move with the right inputs.

Not claimed: anything about why in model‑internals terms, and nothing about what a specific consumer app shows a specific parent on their phone. Also not yet run: Ahrefs Brand Radar at UK locale, blocked on connector authentication at snapshot time — the three‑engine API baseline above is independently complete; Brand Radar would add scaled, ongoing monitoring on top of it.