Sample report: The Sill — 51/100

SAMPLE REPORT · PILOT AUDIT, OCTOBER 2026

AI Visibility Audit — The Sill

51.3

Half-Seen

Mention rate		8.0 / 30
Prominence		4.3 / 20
Accuracy		19 / 20
Sentiment & framing		10 / 15
Citations & sources		10 / 15

Measured on the ChatGPT consumer web interface, logged-out, 16 core questions × 3 repetitions = 48 sessions (47 valid). Mention rate: 8 of 30 unbranded observations (26.7%, 95% range 14–44%). Every figure in this report can be recomputed from the saved answers.

Key findings

1. Missing from most recommendations

In category recommendation questions, The Sill appeared in only 1 of 3 rounds — and in "best for beginners" questions, 0 of 3. Yet when asked for alternatives to its competitor Bloomscape, it appeared 3 of 3 times. The model knows the brand; it just doesn't think of it first.

2. Invisible where intent is highest

Problem-based questions — "my plants keep dying", "greenery without the upkeep", "pet-safe plants" — produced zero mentions in 9 observations. The Sill's strengths (beginner education, pet-safe range) are not connected to the topics in AI's answers.

3. Facts are nearly perfect

Only one deduction across 17 direct answers: a founding detail that mixed the 2012 online launch with the 2014 Chinatown flagship (minor, −1). No invented prices or policies. Two claims we couldn't verify either way were marked uncertain, not scored.

4. Loved when asked, hedged on value

Direct brand questions drew concrete praise (design, beginner-friendliness). Value-for-money questions were weaker — 1 of 3 answers hedged.

Who AI recommends instead

In unbranded answers, the names that kept appearing alongside (or instead of) The Sill: Dahing Plants, Brooklyn Horticulture, Horti, The Old Yew, Plant Kitchen, and Bloomscape — each at a similar count to The Sill overall, but winning the beginner questions outright.

Honest-measurement notes. This pilot was measured from a Korea-based node; on a location question about New York, answers drifted to Korean local shops — location questions are region-sensitive, and reports state their measurement region. With 3 repetitions, ranges are wide (shown above); we never present a single answer as a measurement. On Perplexity and Gemini spot-checks, the same recommendation question did surface The Sill — visibility differs per assistant, and this report's scores are for the stated conditions only.

What we'd fix first

Tier 0 — Facts: clarify the founding timeline on the brand's own about page, so AI stops blending two dates.
Tier 1 — Citations: the beginner and pet-safe buying guides AI cites don't include The Sill. Getting into those sources is the single biggest lever.
Tier 2 — Compounding: re-measure recommendation questions monthly to track whether mention rate moves.

Get this report for my brand — $79