01

Create one measurement dictionary

Define every metric, percentile, date range, device category, page template, and sample threshold once. Apply the same definitions across the group so a dashboard label does not mean different things by rooftop.

What a group speed report must define before any rooftop is compared to another
DefinitionWhy it cannot be assumed
Which pages are tested at every rooftopA homepage-only report hides the pages shoppers actually enter through
Field data or lab dataThey measure different things and must never be averaged together
The percentile reportedA median and a 75th percentile tell different stories about the same site
The date rangeField data is a rolling window; two rooftops read on different days are not comparable
The device classA mobile figure and a desktop figure are not the same measurement
What counts as insufficient dataA low-traffic rooftop may have no field data at all, which is not a failure
02

Preserve rooftop context

  • Website and platform
  • Included page templates
  • Installed third-party environment
  • Major campaigns or releases
  • Traffic and device composition
  • Available sample and field-data coverage
03

Aggregate only comparable evidence

A group average can hide a weak rooftop or overweight a high-traffic store. Present rooftop results first, then aggregate with the weighting method and eligibility rules explained.

A controlled dealership speed study over time. The original and optimized experiences run concurrently on split traffic, across the same days and the same conditions, instead of one after the other.
Two rooftops read on different days are not comparable. A group report has to state its window before it states its numbers.
04

Make reports actionable

Connect each issue to a responsible system, recommended action, risk, verification plan, and current status. DealerSpeed Engine reporting is intended to document the control and optimized conditions without turning one store’s observation into a universal claim.

05

Can group reporting be automated?

  1. 01

    Use the Chrome UX Report API

    Free, public, and it returns the same field data Google assesses, for any origin, programmatically.

  2. 02

    Query monthly

    Field data moves on a 28-day window. Querying weekly reports mostly noise.

  3. 03

    Store the history

    The value compounds. A single month is a snapshot; twelve months shows which rooftops are drifting.

  4. 04

    Flag movement, not absolutes

    A rooftop that just fell below a threshold it used to pass is the actionable event.

06

Why do group speed reports mislead?

Because an average across rooftops answers a question nobody asked. A group of thirty stores with twenty-eight healthy sites and two badly broken ones produces a reassuring mean and a real problem, and the mean is what ends up on the slide.

Weighting makes it worse instead of better if it is not disclosed. Weight by traffic and the flagship store decides the group number; weight by rooftop and a low-volume satellite counts as much as the highest-revenue location. Neither is wrong, but a report that does not say which one it used is not interpretable, and two reports using different weighting cannot be compared even within the same group.

The other structural problem is that rooftops are rarely comparable. Different brands, different template versions, different agency-installed tools and different inventory volumes mean two stores on the same platform legitimately produce different numbers for reasons no group-level action would fix.

07

What should a group-level report contain?

  • Pass or fail per rooftop on all three Core Web Vitals, mobile and desktop separately, with the change since the previous month
  • One definition of each metric, percentile, date range, device category and page template, applied across every rooftop
  • A minimum sample threshold, with rooftops below it shown as insufficient data instead of as a number
  • Page templates reported separately: a group homepage average tells you nothing about vehicle detail pages
  • The third-party inventory per rooftop, ranked by contribution, because that is usually where the differences between stores live
  • Platform noted per rooftop, without ranking platforms against each other
  • A named owner for each issue, since most fixes belong to an agency or a vendor instead of to the store
  • Every change that could move the numbers: platform releases, campaign starts, incentive periods, tool additions

Sources and further reading

External sources support the general technical guidance on this page. They do not represent a DealerSpeed Engine performance result.