An illustrative quarter
A composite drawn from patterns seen across catalogs in the first half of 2026 — figures illustrative, not a published audit. A $40M-GMV outerwear brand, six months into brand-level AI tracking:
| Metric | Reading |
|---|
| Brand mentions, quarter over quarter | +41% |
| Share of voice vs two nearest competitors | 28% → 34% |
| Franchise SKUs (~60% of D2C revenue) named in category answers | <1 in 10 |
| Engines most often cited review roundups that excluded those SKUs | — |
Every brand-level number in that table is green. The business is losing the answers that sell its three most important products.
Why brand tracking cannot see this
It is a unit-of-analysis problem, not a tool problem. Brand-level platforms count occurrences of the brand string. They do not record whether the mention resolved to a product, whether the cited URL was a PDP or a homepage, or whether the answer helped a buyer decide. Those distinctions are not in the schema, so no report can surface them.
Seeing which SKUs are silent requires a system that stores each product as its own entity, runs the prompts that would naturally surface it, and records whether the product — not just the brand — appeared. That is the core primitive of SKU-level tracking and the reason eCommerce Insights was built around it from the first commit.
The math of branded-only mentions
Even before conversion, the ratios matter. On most D2C catalogs, revenue concentrates at the head: the top tenth of SKUs carries the majority of revenue. If your brand is mentioned 1,000 times across a prompt set and your top-revenue SKU appears in 40 of them, that is a four percent resolution rate on the product that matters most. Brand-level tracking reads the same data as "mentioned 1,000 times" — healthy. SKU-level tracking reads it as "best seller named 4 percent of the time" — urgent.
The ratio shifts by category. Fashion resolves to SKU more often because buyers describe the literal product; beauty resolves less often because buyers describe outcomes; electronics sits between. Whatever your category, the ratio is knowable only if you measure both levels.
They track your brand. We track your SKUs.
The honest positioning. Profound, Peec, Brandlight, and Otterly build good products for brand-level visibility. Ahrefs Brand Radar and Semrush's AI Visibility Toolkit bolt the feature onto suites their customers already run. None resolve to SKU at meaningful scale as of mid-2026 — per their public materials, not a knock.
eCommerce Insights tracks every SKU per engine per prompt intent, parses each answer for product names, variants, and cited PDP URLs, and rolls everything into a catalog-wide AI visibility score. When a product drops out of an answer, the ledger says which SKU, on which query, and what likely changed — usually on the PDP, sometimes on a review site you can reach, sometimes a structured-data field gone stale. The SKU-level AEO guide covers the full method.
Running both: the common pairing
Most teams do not rip out brand-level tracking. They keep it for comms and add SKU-level tracking for merchandising and SEO. The split is clean: comms owns the brand dashboard, ecom owns the SKU ledger, leadership reads both in the monthly readout. Consolidation usually happens only when a team bought a brand tool reflexively, watched a quarter of flat product-level movement, and realized it was reporting the wrong dimension.
What to do this week
Pull your brand-level report. Find the three prompts where your brand was mentioned most. Run each yourself in ChatGPT or Perplexity with a fresh session. For each answer, check whether a specific SKU is named, which SKU it is, and whether it is one that drives revenue. If the answer is "no SKU" or "the wrong SKU," that goes on the ecommerce roadmap — it is the line item the brand dashboard is not showing you. Shopify's own writing on AI in commerce is useful context for where the platform itself sees this heading.
Key takeaways
- Brand-level AI tracking is a comms metric. Ecommerce lives at the SKU.
- Rising brand mentions can mask falling product citations — the failure mode that costs revenue.
- Profound, Peec, Brandlight, and Otterly serve the brand-level buyer well; none resolve to SKU at scale as of mid-2026.
- Common pattern: keep brand-level for comms, add SKU-level for merchandising and SEO.
- First diagnostic: run your top brand-mention prompts manually and note whether a revenue-driving SKU is named.
