Reading your analytics

The four sections of your dashboard#

Overview#

The first thing you see when you log in. Designed to tell you “is everything working?” at a glance.

Total events (last 7 days)#

The volume of shopper actions your store fired into CartAmplify. Trending up = healthy growth. Trending down or flat = check that your event tracker is still firing (sometimes a theme update breaks the snippet).

Revenue attributed to CartAmplify#

Of the revenue your store generated in the period, how much came through a CartAmplify surface — a search result clicked, a browse category clicked, a recommendation clicked. This is the “is this worth what I’m paying” number.

Per-language model status#

A row for each language you sell in, with one of three states:

• Cold start — Under ~10,000 events. Personalization isn’t active yet for this language.
• Personalized — A model has trained and is serving personalized results.
• Stale — The model exists but hasn’t retrained recently. Usually fine; happens to low-traffic languages.

If a language has been in cold start for more than a month, consider bulk-importing historical events to fast-forward it.

Catalog size#

How many products CartAmplify is tracking, per language. Should match your store. If it doesn’t, check the sync status banner on the Products page for errors.

Search#

This is the section to check weekly. It’s where the highest-impact small fixes live.

Total queries#

Volume per day. Look for sudden drops (tracker broke) or sudden spikes (a viral product or marketing push).

Zero-result queries#

Probably the single most actionable list in the entire dashboard.

Every query in this list is a shopper who searched, found nothing, and likely bounced. Examples of what shows up:

• Misspellings the system can’t recover — rare, but they happen. Add a synonym.
• Different naming — your catalog says “earbuds”, they searched “ear buds” or “wireless earphones”. Add synonyms.
• Things you don’t sell — useful catalog research. Maybe worth adding the product line; maybe worth a polite “out of stock” landing page.

Click-through rate (CTR)#

The percentage of searches that result in at least one product click. A healthy fashion store sits around 35–55%; electronics around 25–40%; grocery much lower (3–10%) because of habitual repeat shoppers.

There’s no universal “good” number — but a falling CTR on a query that used to work is a flag worth investigating. Usually a relevance regression after a category rename or attribute change.

Revenue from search#

Same idea as the overview number, scoped to search only. Compare this to revenue from your other surfaces to understand where shoppers are converting.

Browse#

Available on Growth and above. Shows category-page performance.

Per-category traffic#

Sorted by visitors. Useful for spotting categories that get traffic but no conversions — those are candidates for editorial reordering or a better landing layout.

Per-category CTR and conversion rate#

CTR = of the visitors who landed on this category, how many clicked into a product. Conversion = how many ultimately bought.

A category with high CTR and low conversion is doing something right (the products are interesting) but failing on the next step — usually a product detail page issue, not a category page issue.

Per-category revenue (Amplify+)#

Top revenue-driving categories. Useful for inventory and marketing prioritization, not for tuning CartAmplify itself.

Recommendations#

Per-widget breakdown. The way to figure out which widgets are earning their slot on the page.

Impressions#

How many times the widget rendered for a shopper. Low impressions usually mean the widget is below the fold — consider moving it up.

Click-through rate (CTR)#

The percentage of impressions that resulted in a click. A useful rough benchmark:

• Above 2% — Strong. Widget is well-placed and matched to the audience.
• 0.5–2% — Normal range for most widgets.
• Below 0.5% — Weak. Either wrong widget type for that surface, or wrong placement.

Conversion rate#

Of the clicks, how many led to a purchase. Combines widget quality with the rest of your funnel — if every widget converts at 1% and one converts at 0.1%, that one widget is the issue. If every widget converts at 0.1%, your checkout might be the issue.

Revenue per impression#

The ultimate number for ranking your widgets. A widget can have lower CTR than another and still earn more if the products it surfaces have higher AOV. This is what you optimize for if your goal is “make more money”, not “get more clicks.”

What to act on, by frequency#

Common questions#

Where to go next#