The Knowledge Graph Teardown: Diagnosing a Weak Entity Record
Objective: Given two specimen snapshots of Stitch Fix's entity signals, find the real defects blocking Knowledge Panel and AI Overview visibility, without flagging cosmetic non-issues.
You're a growth marketer at Stitch Fix investigating why a competitor's styling-subscription brand shows up in AI Overview answers about 'personal styling services' and Stitch Fix does not, despite ranking higher on-page.
Review the schema snapshot and the off-site mentions snapshot, then list every entity defect against a fixed answer key.
What entity defects are blocking Stitch Fix from Knowledge Panel and AI Overview visibility that a higher-ranking competitor already has fixed?
Before you start
What you'll need
- —Familiarity with Organization JSON-LD and the concept of sameAs corroboration
- Co-citation
- an unstructured mention of a brand alongside category peers (in a roundup article or podcast), which Google weighs as corroboration alongside structured data.
- Wikidata
- a structured, machine-readable knowledge base Google triangulates against when building an entity record, distinct from Wikipedia's prose pages.
Free path (everything below is enough to finish)
Free, no account friction for a single comparison table
Paid upgrades (optional, faster/deeper)
The teardown itself needs no paid tool, manual search-operator checks ("Brand Name" -site:brand.com) approximate the same off-site mention count Ahrefs automates.
Paid tier needed for ongoing mention-volume tracking beyond a one-time manual check
The process
Specimens to review
Review this Organization JSON-LD snapshot pulled from Stitch Fix's homepage and /about page.
=== HOMEPAGE JSON-LD ===
{
"@type": "Organization",
"name": "Stitch Fix",
"url": "https://www.stitchfix.com"
}
=== /ABOUT PAGE JSON-LD ===
{
"@type": "Organization",
"name": "Stitch Fix, Inc.",
"url": "https://www.stitchfix.com/about",
"founder": "Katrina Lake"
}Specimen: synthetic, realistic
Review this snapshot of Stitch Fix's off-site presence compared to a competitor's, both in the 'personal styling subscription' space.
=== STITCH FIX OFF-SITE SNAPSHOT ===
Wikidata: no item found
Wikipedia: page exists, last edited 2019
Podcast mentions (last 12 months): 2
Roundup articles ('best styling subscription services'): appears in 3 of 10 checked
=== COMPETITOR OFF-SITE SNAPSHOT ===
Wikidata: item exists, 14 properties filled
Wikipedia: page exists, last edited this year
Podcast mentions (last 12 months): 19
Roundup articles: appears in 9 of 10 checkedSpecimen: synthetic, realistic
Analyze your findings
What to look for
- Entity consistency
- Do all Organization blocks across pages share a single @id and identical name string?
- Structured vs. unstructured corroboration
- Is the gap in schema (Wikidata, sameAs) or in off-site mentions (podcasts, roundups), or both?
- Magnitude of the gap
- How large is the difference from the competitor's numbers, not just whether a gap exists?
- Distractor discipline
- Is a cosmetic detail (URL path, HTML ordering, edit date alone) being mistaken for a structural defect?
Make the call
Stitch Fix has no Wikidata item, 2 podcast mentions and 3-of-10 roundup appearances in 12 months, versus the competitor's fully populated Wikidata item, 19 podcast mentions, and 9-of-10 roundups. Which factor most directly explains why the competitor appears in AI Overview answers and Stitch Fix does not?
Recommendation · Priority: High
“Unify the Organization JSON-LD across homepage and /about under one @id with a single consistent name string, and nest the founder as a linked Person entity. In parallel, create a Wikidata item, since Stitch Fix currently has none, and pursue placements in styling-subscription roundup articles and podcast mentions to close the co-citation gap versus the competitor's 19 mentions and 9-of-10 roundup presence.”
Common mistakes
What trips people up
Treating two separately-defined Organization blocks as a minor duplication — without a shared @id, this creates competing entity signals rather than one entity Google can confidently merge, a critical defect, not cosmetic.
Overlooking the founder as a plain string — a string value gives Google a name with no way to connect it to a corroborated Person entity, losing the founder-authority boost.
Assuming ranking higher on-page should guarantee AI Overview visibility — AI Overviews and Knowledge Panels draw on entity corroboration signals (Wikidata, co-citations) that are independent of on-page ranking strength.
Flagging the Wikipedia page's 2019 edit date as equivalent to having no Wikidata item — an existing-but-stale Wikipedia page is a smaller gap than a completely missing Wikidata item, which is an entire missing corroboration channel.
Final deliverable
A defect log across both specimens, each entry tagged with severity and which entity-building concept it violates.
See a reference example
Rent the Runway entity defect log (excerpt) [CRITICAL] No Wikidata item found despite 8 years of press coverage -> How Google Builds an Entity Record [MODERATE] founder listed as plain string, not a linked Person entity -> How Google Builds an Entity Record
Success criteria
You're done when you can:
- Correctly flags both critical defects (missing Wikidata item, low co-citation volume) across the two specimens
- Does not flag either distractor as a defect
- Ties each real defect to the correct lesson concept
Key takeaway
Entity strength comes from two separate channels: structured data (a unified Organization block with a stable @id and a populated Wikidata item) and unstructured co-citations (roundups, podcasts, press). A brand can out-rank a competitor on-page and still lose AI Overview visibility if either channel is significantly weaker.