Beyond the view: the decision room's measurable zones 10 min live
view_item says a shopper arrived at the product. It says nothing about what she did there. The PDP has zones, and each zone can fire an event: the image gallery (GA4's select_content, or a custom image_swipe), the reviews block (a custom review_open), the size guide (a custom open), and the wishlist button (add_to_wishlist - a standard GA4 ecommerce event, no customization needed). With those events flowing, Mango Lane can define an engaged PDP session: viewed 2+ gallery images, OR opened reviews, OR opened the size guide.
LiveThe engagement events, one by one5 min▶
The instrumentation shopping list, cheapest first:
- add_to_wishlist - already in GA4's standard ecommerce event set, same family as add_to_cart. If your ecommerce events flow, this one is nearly free.
- Gallery interaction - GA4's generic
select_contentworks, or a customimage_swipewith an index parameter so you can distinguish "saw image 2" from "saw all 8". - review_open / size_guide_open - custom events on the tap or expand. Remember b1's trap: if reviews open in a modal sheet, auto screen tracking may miss it entirely - log the event manually.
Self-studyWhy wishlist adds sit outside the engaged rule2 min read▶
Mango Lane tracks add_to_wishlist but leaves it out of the engaged definition on purpose: a wishlist add is deferred intent ("not today"), while gallery, reviews and size guide are active evaluation ("deciding right now"). Mixing the two muddies the split - a high-wishlist session often converts NEXT month, which this month's CR cannot see. Keep wishlist as its own metric; it becomes interesting again in re-engagement work (the push and deeplink doors from b1).
The correlation trap, stated honestly 6 min live
Engaged sessions convert at 9%, plain at 4.5%. It is very tempting to read that as "if we get plain sessions to engage, they will convert at 9% too" - and that reading is wrong. Engaged shoppers were more interested to begin with; the swipe did not create the intent, the intent created the swipe. Forcing engagement (auto-playing the gallery, popping the size guide open) does not transplant the intent.
LiveDiagnostic, not causal promise4 min▶
What the engagement split IS good for - and it is genuinely good for this:
- Which zones correlate with buying. Split the engaged 72,000 further by zone: sessions that opened reviews vs sessions that only swiped images. The zone with the strongest order correlation is your first candidate for investment - and for a test.
- Where plain sessions bail. 72,000 sessions arrive and touch nothing. Do they bounce in 3 seconds (wrong product, bad entry - a b2/b3 problem) or linger and leave (the page failed to invite them in - a PDP problem)? Elapsed-time views in funnel exploration separate the two.
- A before/after tripwire. If a redesign moves the engaged share from 50% to 60% but the engaged CR collapses, you diluted the pool with forced engagement - the metric caught you inflating it.
Causal proof - "does making the gallery better CAUSE more orders?" - needs an A/B test, where you change the page for a random half and compare. That craft is its own course: learn-experimentation-with-phoebe.
Self-studyThe selection effect in one sentence2 min read▶
People who read hospital reviews are sicker than people who do not; hospitals did not make them sick. The engaged 72,000 selected themselves into engagement because they were already closer to buying. Any observational split - engaged vs plain, searcher vs browser (b3's 2.67x), push-receiver vs not - carries this selection effect. The honest phrasing habit ("X sessions convert at 2x", never "X doubles conversion") is the cheapest insurance an analyst owns.
The evidence for the zones, precisely phrased 7 min live
The trap says the 2x is not proof. It does not say the zones are worthless - there is real external evidence that images, reviews and size guidance matter. The discipline is quoting that evidence exactly as strong as it is, and no stronger.
LiveImages first, reviews measured, sizing honest5 min▶
- Images are the first act. Baymard's PDP research finds 56% of shoppers' first action on a product page is exploring the images. The gallery is not decoration; it is where evaluation starts for most sessions.
- Reviews have academic numbers. The Spiegel Research Center's 2017 review (with PowerReviews data) found: a product showing 5 reviews had a 270% greater purchase likelihood than the same product with 0 - note the phrasing, it is 5-versus-0, not "reviews lift sales 270%". For higher-priced products the gap widened to +380%. A Verified Buyer badge added about +15%. The ratings sweet spot was 4.0-4.7 - near-perfect 5.0 averages read as fake and convert worse. And the benefit plateaus after roughly 5 reviews.
- Fit and size drive returns. Poor fit is the #1 apparel return reason - sources put it anywhere from ~53% to ~70% of fashion returns (McKinsey at the high end). Size guidance is return prevention as much as conversion. But no credible "size guides cut returns by X%" number exists - if you see one quoted, ask where it came from.
The plateau is the actionable part of Spiegel. If benefit flattens after ~5 reviews, the highest-leverage review program is not stacking review #51 on the bestseller - it is getting the first 5 reviews onto every zero-review product in the catalog. Merchandising teams routinely get this backwards because bestsellers are the products they look at.
Self-studyHow to file each evidence grade2 min read▶
| Claim | Grade | How to quote it |
|---|---|---|
| 5 reviews vs 0 = 270% greater purchase likelihood | Academic (Spiegel 2017) | Quote the exact comparison, with the plateau caveat |
| 56% start with the image gallery | Industry research (Baymard) | Quote as a behaviour share, not a conversion claim |
| Fit is the #1 fashion return reason, ~53-70% | Range across sources | Give the range and name the high-end source |
| "Size guides cut returns by X%" | No credible number | Say so out loud - do not invent one |
The engagement audit ★ 15 min · everyone works
Three drills, calculator out. Drill 1 proves the split reconciles; drill 2 crosses engagement with b3's paths; drill 3 finds the slice worth testing.
LiveStep 1 · verify the split reproduces total orders4 min▶
Engaged orders: 72,000 x 0.09 = 6,480.
Plain orders: 72,000 x 0.045 = 3,240.
Sum: 6,480 + 3,240 = 9,720 - exactly the canon order count. The split passes reconciliation.
Always run this check on any behavioural split before presenting it. A split whose segments do not cross-foot to the known total means the segment definitions overlap, leak, or were computed on different date ranges - all three happen constantly in the wild.
LiveStep 2 · engagement rate by path5 min▶
Cross the engagement split with b3's paths. Mango Lane's numbers:
Search-path PDP sessions: 43,200. Of these, 60% engage: 43,200 x 0.60 = 25,920 engaged.
Browse-path engaged sessions: all engaged minus search engaged = 72,000 - 25,920 = 46,080.
Browse-path engagement rate: 46,080 / 100,800 = 45.7%.
The punchline: searchers arrive warmer (b3: path CR 12% vs 4.5%) AND engage more once they arrive (60% vs 45.7%). Intent shows up at every layer of measurement - one more reason the naive "make browsers engage more" reading of the 2x chart fails.
LiveStep 3 · find the actionable slice5 min▶
The interesting sessions are the ones that break the pattern:
Plain but added to cart. Sessions that touched no zone yet carted anyway. What did they NOT need? Often: repeat buyers who already know the product. If this slice is large, your engaged rule may be measuring novelty, not quality.
Engaged but never carted. Swiped images, opened the size guide - then left. The size-guide subset is the loudest: they wanted to buy and could not resolve fit. This is where Part 3's returns evidence points your first test.
Write it down: for each slice, name the ONE zone you would A/B test first and what the success metric is. That sentence is the handoff to the experimentation course.
Try it yourself - this week ◐ 20-25 min total
- Write your own engaged-PDP-session rule (some version of: 2+ gallery images OR opened reviews OR opened size guide). Check which of the events already exist in your tracking.
- Compute the two conversion rates - engaged sessions to order, plain sessions to order - and check the segments cross-foot to your total orders.
- Present the split to one colleague using the honest phrasing: "engaged sessions convert Nx" - and watch yourself resist saying "engagement Nx-es conversion".
- Count your zero-review products. Spiegel's plateau says that list, not your bestsellers, is where review collection pays.
Sources this session draws on
The engagement evidence base is a mix of academic work, industry research, and honest gaps - flagged accordingly:
Three questions before you go 🎯 ◐ 90 seconds
1 · Engaged PDP sessions convert at 9%, plain at 4.5%. Which caption is honest?
Engaged shoppers were more interested to begin with - the intent created the swipe, not the reverse. The split diagnoses which zones correlate with buying; only an experiment can prove a zone change causes orders.
2 · What did Spiegel 2017 actually find about the 270% number?
The finding is a 5-versus-0 comparison, not a per-review multiplier or a blanket lift. Precision matters: the effect was larger for pricier products (+380%) and plateaus after roughly 5 reviews.
3 · Spiegel's benefit plateaus after ~5 reviews. What does that imply for a review program?
If the curve flattens after ~5, review #6 onward buys little - but reviews 1-5 on a zero-review product buy the whole 270% gap. The long tail of unreviewed products is where collection effort pays.