How quickly
does it load?
Separate Google PageSpeed mobile reports with SpeedLayer off and on. These are scores, not shopper outcomes.
Open the two Google reports
JENKINS / PERFORMANCE REVIEW
Find your store
Across 23 measured stores, the SpeedLayer-active group recorded 3.05 lead actions per 100 measured users, versus 0.84 in the not-active group. These are calls, texts, chats and form events, not unique leads.
Observed audience comparison on qualifying Mondays, not a randomized test or proof SpeedLayer caused the difference.
We show mobile speed tests and shopper-action counts separately. A faster lab result and a higher observed action rate are not the same measurement.
Separate Google PageSpeed mobile reports with SpeedLayer off and on. These are scores, not shopper outcomes.
Open the two Google reportsWe count calls, texts, chats and form events per 100 measured users in two observed groups. One person can take multiple actions. The difference is not proof of extra sales.
How we measuredStep 1: open the “without” Google report. Step 2: open the “with” report. Compare the mobile scores for the same website; each link opens Google in a new tab.
Find your store for a plain-language comparison and a clear next check. The five stores missing usable measurements are listed separately below.
Start with a Jenkins store and your own monthly numbers. The engagement and lead-action comparisons are observed, not a controlled test. The dollar and vehicle examples below are scenarios, not measured SpeedLayer return.
Select a measured store to explore its observed engagement and lead-action rates.
Compare all-session engagement and bounce rates. Entered ad spend is applied to each rate only to show scale; these are not measured paid-ad dollars saved.
No paid-traffic segmentation or ad attribution is available.
GA4 engagement is measured on qualifying Mondays; the spend you enter is monthly. This scenario assumes the same mix for paid traffic, which has not been verified. A bounced session is not automatically wasted ad spend.
Use your actual same-month totals and verified close rate. The off estimate applies the store’s observed lead-action event-rate ratio to your real qualified leads. It is a sensitivity test, not a forecast of lost cars.
Digital ad spend per total car sold
Enter your monthly numbers.Digital ad spend per modeled total car sold
No measured off-month sales.Enter your numbers to see the modeled difference.
Same ad spend in both cases. Off qualified leads = entered leads × (observed off/on website lead-action event rates). Modeled fewer cars = missing leads × entered close rate; off total sales subtract only this hypothetical difference. Website events are not unique leads, the cohorts were not randomized, and total sales include non-ad sources. Spend ÷ all cars is not ad-attributed acquisition cost. No inputs leave this browser.
These five properties do not have a usable shopper-action comparison. We do not rank them as poor performers. Here is the measurement step for each.
A decision to stay deserves evidence you can inspect, not a magic percentage. The source workbook is retained privately; individual Google PageSpeed mobile reports are linked from the store views.
Lead actions are forms, texts, chats, and mobile click-to-calls. The rate divides total action events by deduplicated GA4 users, not visits; the group sums those store-level user counts, so a person visiting two stores can appear twice. This is a lead-action signal, not deduplicated leads, sales, revenue, or actual ROI. The audiences were observed, not randomly assigned.
Before retiring the experience across the group, review each property’s evidence, repair the missing signals, and agree on a clean next measurement window.
Start with your store