The library audit

When you finish connecting your reviews — Okendo, Judge.me, Yotpo, Loox, or a CSV — BetterReviews opens an audit page that summarizes what you brought in. The audit is the first step where you can see your actual library through BetterReviews instead of waiting for the dashboard to backfill.

What you’ll see on the audit

The audit is a single scrollable page. Top to bottom:

1. Your brand voice + theme colors. Two cards we filled in for you using your store’s settings. Confirm them with a single click each; you can edit either at any time later.
2. Pipeline. Four numbers across the top: total reviews imported, average rating, count with photos, and a count we call “detailed enough to feature.” The fourth number fills in over the first few minutes as we analyze the library — there’s a progress bar while that runs.
3. What’s been latent in your library. Three cards showing things you may already have but never used: reviews detailed enough to feature, ones ready for ads (review text plus a photo), and a count of support tickets we’d route away from the product page. Numbers are deliberate — we use a strict definition so the count is something you’d actually act on.
4. How rich your library reads. A rating distribution histogram, average word count, % of reviews over 50 words, and verified-buyer percentage. A side card surfaces “5★ reviews worth a second look” — reviews where the rating is five stars but the content reads negative or mixed. These are the ones a customer vented privately about then forgot to lower the star — worth a second look before quoting.
5. Photos. Total customer-uploaded photo count, plus a content-type breakdown once vision analysis runs (lifestyle / unboxing / product detail / accidental).
6. Your best review so far. The single highest-quality review in your library at the moment, with its photos. As analysis continues, the “best so far” tile can change — that’s intentional.
7. A footer to pick a product to test on. Step 3 of onboarding.

Why the audit looks the way it does

We use a diagnostic voice, not a graded one. “Detailed enough to feature: 1,968” rather than “Only 1.3% of your reviews are marketing-ready.” The numbers come from your own data — the framing leads with what’s there, not what’s missing.

We also use strict definitions for the latent-value counts. Validating against real merchant libraries, the LLM’s loose tags (marketing_potential = medium-or-high, support_ticket = true) over-counted real-world actionable reviews by 2-38×. The audit headline always uses the tightened version so the number you see is the number you’d actually quote.

How long the audit takes to fill in

Layer 1 numbers (rating, photo count, total) appear within a few seconds of landing — they’re aggregates over your imported library.

Layer 2 numbers (detailed-enough-to-feature, ad-ready, support-routed, 5★ mismatch) fill in over the first 2 to 5 minutes for a typical library. Larger libraries (50k+ reviews) take longer — we’ll keep filling in for up to 10 minutes and the page progress bar tracks the state in real time. The page polls every 5 seconds while analysis runs; once your library is fully analyzed it stops polling.

Buttons that aren’t active yet

Some buttons (Browse, Download) unlock as the next-step features ship — hover any disabled CTA for specifics. The numbers are real; the destinations are landing in subsequent releases.

The See what we’d ask this customer today → button under your best review is live — click it to preview the conversation we’d start with a customer who just bought that product. See Customizing the chat for how the preview surfaces work.

Refreshing the audit

If you change something upstream — new theme colors, a reworked brand voice, more reviews — use the Re-analyze my store button on the audit page to re-read your store and refresh both the brand-voice and theme-color cards, without leaving the page.

What we don’t include

We don’t include third-party benchmarks (“industry average is X reviews”) on the audit. Small samples per merchant produce noisy comparisons that read as graded rather than diagnostic; if you want comparisons, the dashboard’s discovery view will let you slice your own library more usefully.

We also don’t show personally identifiable information (PII) on the audit. The “your best review” card uses the public-attribution name you’d display in the widget, never the full name or email we have on file. Same redaction rules apply across the whole audit page.