Changelog v1.2: variant depth, live LLM probing, and Shopify apply
v1.2 is the biggest surface-area release since Surfacely launched. Seven phases shipped in one changelog window. This post walks through each one in order — what changed, what it does to your audit score, and what you can act on today.
Phase 10: variant-level schema depth
Before v1.2, your audit scored schema at the product level. A PDP with a valid Product block got credit regardless of whether the size, color, and material variants inside it were machine-readable or not.
AI engines do not stop at the product level. When a buyer asks ChatGPT "best black running shoes in size 10 under $120," the model needs variant data — price, availability, SKU, dimensions — to form a confident answer. If that data is missing or collapsed behind JavaScript, your product does not make the shortlist.
Phase 10 adds variant-level extraction to every audit. The report now shows you, per PDP, which variants carry complete schema and which are invisible to AI crawlers. The score lift attached to each fix reflects the actual coverage gap at the variant level, not a flat product-level pass/fail.
If your store sells products in multiple colors or sizes, this is the section of your new report to read first.
Phase 11: llms.txt and AI-readable exports
llms.txt is an emerging convention that tells AI crawlers where to find structured, trustworthy information about your store — analogous to robots.txt, but for language models. Without it, a crawler decides for itself which pages to trust.
Phase 11 adds two things to your audit:
- llms.txt detection — we check whether your store publishes an llms.txt file and whether its contents point AI crawlers to your canonical product data.
- AI-readable export — if your store does not have one, your audit report now includes a copy-paste-ready llms.txt block scoped to your domain and your top-priority PDPs.
Shipping the export took five minutes in our internal test. The score lift is modest on its own, but it compounds with the schema fixes in Phase 10 — AI crawlers that trust your structure read it more deeply.
Phase 12: live LLM probing
Schema analysis tells you what AI engines can read. Live probing tells you what they actually say about your store today.
Phase 12 adds live probing to every audit. After the crawl completes, we run category prompts against four engines — ChatGPT (GPT-4 family), Claude (Haiku 4.5), Perplexity, and Google Gemini — and record verbatim responses, citations, and sentiment scores.
Your report now includes a probing section that shows:
- Which engines cite your store and in which answer positions
- Verbatim excerpts from AI responses that mention your products
- Sentiment scoring per engine, per prompt
- A diff-ready baseline so Phase 17 monitoring can measure change over time
The prompts are category-level, scoped to what a real buyer would ask — not your brand name. If an engine does not surface you for "[your category] under $[your price point]," the probe records that absence. Most merchants find the absence more useful than the citations.
Phase 13: prescriptive playbook
Audit findings are only useful if they tell you what to do next.
Before v1.2, your report listed issues. v1.2 adds a ranked fix playbook — every issue now ships with:
- A specific, copy-paste fix (schema block, meta tag update, llms.txt line, or heading revision)
- The expected score lift, expressed as a percentage improvement against your current audit baseline
- The estimated implementation time in minutes
Fixes are ranked by expected lift, not by category. If fixing one image alt-text block on three PDPs outranks a heading hierarchy revision across your entire catalog, the image fix appears first.
The playbook is the artifact your developer or your Shopify agency opens when they sit down to act on your audit. It is also the input surface for Phase 14/15 — the Shopify embedded app reads the playbook and routes each fix to the right apply workflow.
Phases 14 and 15: Shopify embedded app with review and apply
This is the most concrete v1.2 deliverable.
Surfacely now ships as an embedded Shopify app. Connect your store once — the app requests only the scopes it needs to read and write product metafields and schema blocks — and your audit playbook surfaces inside your Shopify admin.
The apply workflow has three steps:
- Review — the app surfaces the AI-generated fix alongside the current live value. You see exactly what will change before anything is written.
- Approve or skip — approve a fix to queue it for apply; skip it to defer without losing it from the playbook.
- Apply — approved fixes write directly to your live PDPs. Schema blocks, metafield values, and structured data patches apply without a code deploy.
AI rewrites — product description copy optimized for AI extraction confidence — go through the same review/approve/apply gate. Nothing writes to your store without your explicit approval on that specific fix.
One-click apply is available on the Pro tier. The embedded app shell (review and approve without apply) is available to Free tier users who connect their store, so you can see exactly what would change before you upgrade.
Phase 16: user auth and audit history
Previously, audits were ephemeral — run a report, read it, lose it.
Phase 16 adds user auth and 30-day audit history for Pro tier users. Every audit your account runs is saved, addressable by URL, and comparable against earlier runs on the same domain.
Free tier users get their current audit saved for 7 days. Audit history is one of the clearest reasons to upgrade — the value of a second audit compounds when you can diff it against the first.
Phase 17: continuous monitoring and diff alerts
AI engines update their answers continuously. A store that appears in ChatGPT's recommendations for "best [category]" today may not appear next week — and without monitoring, you find out only when revenue drops.
Phase 17 adds continuous monitoring. Pro tier stores get weekly re-probes across all four engines, scoped to the same prompts your baseline audit used. When a re-probe detects a change — a new citation, a dropped mention, a sentiment shift — you get a diff alert that shows the before and after verbatim.
The monitoring dashboard (your audits → [domain] → monitoring) shows sparkline trends per engine, per prompt, over your full history window. Agency tier stores get weekly monitoring across all connected domains in a single view.
Monitoring is the product you run after you apply your playbook fixes. It tells you whether the fixes held, whether new issues emerged, and whether competitors gained ground in the answers your buyers are reading.
What ships on which tier
| Capability | Free | Pro | Agency | |---|---|---|---| | Variant-level schema depth | Included | Included | Included | | llms.txt detection + export | VERIFY | Included | Included | | Live LLM probing (4 engines) | VERIFY | Included | Included | | Prescriptive playbook | Included | Included | Included | | Shopify app shell (review + approve) | VERIFY | Included | Included | | Shopify one-click apply | — | Included | Included | | Audit history | VERIFY | 30 days | 30 days | | Monitoring + diff alerts | — | Monthly | Weekly | | White-label PDF export | — | — | Included | | Multi-domain | — | — | Included |
Run your updated audit
Every audit run from today forward uses the v1.2 engine — variant depth, live probing, and the full prescriptive playbook are included at no extra cost on the Free tier.
If you ran an audit before this release, your old report does not automatically update. Run a new audit against your domain to get the v1.2 results.
If your store is on Shopify and you want to move from reading fixes to applying them, upgrade to Pro and connect your store. The review/approve/apply workflow is the fastest path from audit score to live changes.