01

Choose representative pages before testing

Include the homepage, a populated inventory search-results page, several representative vehicle-detail pages and at least one service, finance or lead path. Record every URL, page state and test date.

Use stable in-stock vehicles where possible. An unusual unit, missing media or temporary vendor condition can distort a single VDP result.

The minimum page set for a dealership speed test
Page typeWhy it is included
HomepageThe only page most tests cover, and the least representative of entry traffic
New-inventory search pageLarge DOM, many images, filter interactions
Used-inventory search pageOften a different template and a different tag set from new
At least two vehicle detail pagesThe heaviest pages on the site, and where shoppers convert
A paid landing pageUsually carries extra measurement tags the rest of the site does not
The service scheduling entry pointA conversion path that is frequently third-party and rarely tested
02

Run repeatable laboratory tests

  1. 01

    Freeze conditions

    Record tool version, device profile, network throttling, location, cache state and consent state.

  2. 02

    Repeat across days

    Run multiple valid tests rather than selecting the best screenshot; the product-proof baseline uses at least nine valid runs per URL and device across three days.

  3. 03

    Save diagnostic evidence

    Keep the waterfall, main-thread trace, LCP element, layout shifts and interaction diagnostics.

  4. 04

    Compare like with like

    Use the same page set and conditions after a change and explain anything that could not remain constant.

03

Keep field and laboratory data separate

PageSpeed Insights may show both Lighthouse lab data and Chrome field data. Record whether field coverage is URL-level, origin-level or unavailable, plus the reporting window and device segment.

Laboratory traces diagnose a reproducible load. Field percentiles describe eligible real-user experiences. Neither should be relabeled as the other.

Google's Core Web Vitals thresholds at the 75th percentile: Largest Contentful Paint good at 2.5 seconds or less and poor beyond 4 seconds, Interaction to Next Paint good at 200 milliseconds or less and poor beyond 500 milliseconds, Cumulative Layout Shift good at 0.1 or less and poor beyond 0.25.
The thresholds a test is read against. A laboratory run produces a score; only field data is measured at the 75th percentile of real visits.
04

Test the functions shoppers use

  • Primary navigation, inventory filters, sorting, pagination and scrolling
  • VDP gallery, pricing, incentives and availability
  • Lead forms, service, finance, chat, trade and digital retailing
  • Consent, accessibility, keyboard focus and mobile viewport behavior
  • Required analytics events identified by the dealership
05

Report scope and limitations

State the pages, devices, conditions, dates, observations, likely causes and changes. One test run is not a universal shopper result, and a speed observation does not prove a ranking, lead, conversion or revenue outcome.

06

What makes a speed test repeatable?

A single test is an anecdote. What turns it into a measurement is recording enough context that the same test can be run again and the two results compared honestly.

That means the same pages, not the same page types: a specific homepage URL, a specific search-results URL with the same filters applied, and specific vehicle URLs recorded by VIN so you can tell whether the unit is still in stock when you retest. It means the same device profile and the same connection assumptions. It means the same time of day, because dealership traffic and third-party vendor load both vary across the day. And it means several runs per page rather than one, because the spread between runs on a script-heavy page is often wider than the difference any change will make.

Record all of that alongside the numbers. A result without its conditions cannot be compared to anything later, which means it cannot be used to justify or reject a change, and a measurement that cannot change a decision was not worth taking.

07

What should a dealership do with the results?

Read the field data first, if the page has any. It describes what real visitors actually experienced over the last 28 days, which is the question a dealer principal is asking. The laboratory score underneath it is a controlled simulation on a specific device profile; it is excellent for diagnosing why something is slow and it is not evidence of what anyone experienced.

Then look at what is loading rather than at the score. The third-party breakdown is where the explanation usually lives, and it is more actionable than the score because each entry has an owner inside the dealership or its agency. The free speed test on this site returns exactly that: the field Core Web Vitals graded against Google's thresholds, and the third-party providers ranked by how long each blocked the browser.

Finally, give any change four weeks before judging it. Field data is a rolling 28-day window, so a fix made today is diluted by three and a half weeks of the old experience for most of the following month.

Sources and further reading

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