Documentation
AI Visibility Features/Tracking Products

Tracking Products

Spyglasses can run an AI visibility report on a product, not just a parent brand. Use it when a brand has multiple distinct products with their own positioning, audience, or buyer journey — and you want to see how each one shows up in AI answers independently of the corporate-level brand.

What You'll Learn

  • When to track a product vs. just the parent brand
  • How to add a product and what the "Product page URL" controls
  • How the brand snapshot is built from a single seed page
  • How a product report differs from a brand report
  • How to use key messages and tags to slice product-level metrics

When to Use a Product Report

A product report is the right tool when a single domain hosts multiple things you want to track separately. For example:

  • A motorcycle manufacturer's Road Glide Limited vs. Pan America 1250 — same brand, completely different buyer prompts.
  • A SaaS company's Analytics and CDP product lines — different buyers, different competitors, different category framings.
  • A retailer's Apparel and Home Goods categories — same domain, different shopping intent.

If you only need a single rollup for the whole company, run a brand report instead — it crawls the root domain and gives you one view across everything.

Adding a Product

In Properties → Track something new, choose Product as the subject type, then fill in:

  • Product page URL — a representative page for the product, e.g. https://www.harley-davidson.com/us/en/motorcycles/road-glide-limited.html. This is the seed URL — the anchor for everything downstream. The crawl, the brand snapshot, the report's domain context, and the AI prompts all derive from this page rather than from a root domain.
  • Display name (optional) — a friendly name shown in the dashboard, e.g. Road Glide Limited. Falls back to the snapshot's extracted name if left blank.

The form validates the URL up front — it must be an http:// or https:// URL — so a misspelled or non-public link fails before the report starts.

The Brand Snapshot

A product property's brand snapshot is built from a seed-URL crawl, not a full-domain crawl. The crawler:

  1. Always crawls the seed URL itself so the product's own page is the anchor of the snapshot.
  2. Discovers sibling product pages by scanning the seed page's navigation and in-content links. Links sharing the seed's first path segment (e.g. other /products/ pages) get a scoring bonus so the page set stays in the product section instead of drifting into corporate nav.
  3. Falls back to domain-root discovery when the seed alone yields too few pages — useful for surfacing context like /about or /pricing that a deep product page may not link to directly.

The resulting BrandSnapshot contains the product's name, tagline, category, ICP, problems solved, features, pricing, differentiators, and other fields — same shape as a brand snapshot, just scoped to the product's positioning.

Manual entry

If the product page is JS-heavy or behind authentication and the crawl can't extract enough content, you can author the snapshot by hand. On the property's brand snapshot page, click Enter manually to create a blank snapshot, then fill in each field directly. Manual edits are preserved across rebuilds.

Rebuilding the snapshot

You can rebuild the snapshot at any time from the property's settings. For a product, the rebuild reuses the same seed-URL crawl path — if the page set has changed (new sibling products, updated copy), the rebuild picks up the new content. Manual field overrides survive the rebuild by default; pass Discard overrides to start from a clean crawl.

What's Different from a Brand Report

Product reports run the same discovery and recommendation pipeline as brand reports — AI Visibility section, competitor analysis, citations, historical metrics, funnel hero, tracked-attribute matching, the lot. With three intentional differences:

  • Crawl scope — pages come from the seed URL's neighborhood, not the root domain. This keeps the snapshot focused on the product's context instead of diluting it with corporate-level content.
  • No brand-consistency scoring — the "does AI describe this thing consistently?" score is a corporate-brand signal that doesn't translate well to product-level evaluation. Product reports get a snapshot for downstream use but skip the consistency loop. The brand consistency card and chart on the historical dashboard are blank for product properties.
  • Robots.txt and Brandfetch use the seed URL's host. If the parent brand's robots.txt blocks the crawler, the product report fails the same way a brand report on that domain would.

Everything else — discovery prompts, key-message tracking, citations, share of voice, share of influence, the AI Visibility Funnel, custom tags, and the nightly prompt runs — works identically.

Key Messages and Tags for Products

Two existing features become especially useful for products:

  • Product-specific Key Messages — capture the differentiators that matter for this product. A motorcycle product's positive key messages might be "Best touring motorcycle for long-distance comfort" or "Industry-leading cornering technology" — claims that don't necessarily apply to the parent brand as a whole. Positive key messages drive Share of Influence on the funnel hero.
  • Custom tags on discovery prompts — when a product has multiple sub-audiences or use cases, tag prompts with touring, sport, commuter (or whatever your segmentation is) and slice the funnel by tag to see how the product performs with each audience.

The Historical Metrics dashboard works the same way it does for brand reports, with one adjustment: the Brand Consistency card and chart are blank because consistency scoring is skipped for products. Everything else — the AI Visibility Funnel, Share of Voice trends, Mentions, Citations, the platform and tag filters — populates the same way.

Product reports also feed the nightly prompt runs — every overnight execution against the product's discovery queries updates the funnel and the trend charts, so you can see week-over-week movement without waiting for a fresh full report.

Practical Tips

  • Pick the seed URL carefully. It should be the page that best represents the product on its own terms — typically the product detail page, not a category landing page. A category landing page often dilutes the snapshot with cross-product content.
  • Author the discovery prompts before the first report. The auto-generated prompts use the snapshot, which is built from the seed page. They're a good starting point but rarely capture every nuance — review them and add product-specific ones (especially awareness-stage prompts that frame the problem the product solves, not just its category).
  • Use the parent brand as a benchmark. Run a brand report on the parent domain alongside the product reports. You'll often see the parent brand's SoV is higher than any individual product's — that gap is the opportunity for product-level optimization.
  • Tag prompts when you have multiple products on the same property. If you've tagged prompts as touring, sport, etc., you can switch between tag slices on the funnel hero without re-running anything.

Previous

Tracking Individuals

Next

Tracking Awareness

On this page