The short answer
Trust the real-visitor (field) section of a report over any score out of 100. Start from the metric that is flagged, find the page or group of pages involved, and check what changed recently on the site. Lab scores help us debug but swing between runs. Send us the page address, device and a screenshot.
Which report is in front of you?
Speed reports come from several places and they do not all measure the same thing. Find yours in the table before you react to it.
For what the three measurements mean, the thresholds and the usual fixes, read Core Web Vitals. The glossary has short entries for Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift.
| What you have | What it is | How much to trust it | First step |
|---|---|---|---|
| PageSpeed Insights, top section about real users | Field data: real visits to the page | The one that matters for priorities | Read the pass or fail label, mobile tab first |
| PageSpeed Insights, score out of 100 | Lab data: one simulated load | Useful for debugging, varies between runs | Run it three times and take the typical result |
| Search Console, Core Web Vitals report | Field data, grouped by similar pages | Good for spotting which kind of page fails | Open the flagged group and note the metric |
| A score in an email, plugin or other speed tool | Almost always lab data | Treat it as a hint, not a verdict | Check it against the field section above |
Reading PageSpeed Insights, step by step
- 1Enter the page address and run the test. Open the mobile tab first.
- 2Look at the top section about what real users experience. A pass or fail label here is the one that matters.
- 3If it says there is not enough data, the page has too few real visits for a field result. You may see data for the whole site instead, or none at all. That is common on low-traffic pages.
- 4Only then look at the lab score and the lists of opportunities and diagnostics. These point to large images or heavy scripts, but they describe one simulated visit.
- 5Write down the flagged metric, the device and the time you ran it. That is the useful part to send us.
A lab score can swing between runs on the same page. A change of a few points is not a finding.
Reading the Search Console report
The Core Web Vitals report in Search Console sorts your pages into Good, Needs improvement and Poor, separately for mobile and desktop. It is built from real visits, so it lags behind any work you do. Use it to decide what to do next.
| What the report says | What it means for you | What to do |
|---|---|---|
| Mostly Good | Most visitors get a smooth experience | Keep an eye on it after each site change |
| Needs improvement | Close to fine, with room to improve | Note the metric and the page group, then plan it with other work |
| Poor on a group of pages | A shared template or script is probably the cause | Send us the group and the metric, since one fix often covers all of it |
| Poor on one page only | Something specific to that page, such as an oversized image or embed | Check the page for recent changes before asking us |
| No data available | Too few real visits to measure | Use PageSpeed Insights on the page and treat the lab result as a guide |
Match what visitors notice to what is flagged
You do not need to memorize the metrics. Open the page on your own phone and notice what happens.
| What you notice | Which measurement is likely flagged | What to check first |
|---|---|---|
| The main image or headline shows up late | Loading | Size of the top image, a new banner or video above it |
| A tap on a button seems ignored for a moment | Responsiveness | A chat widget, a slider or a newly added script |
| The page jumps as you go to tap something | Visual stability | A cookie banner, an embed or an image with no reserved space |
Before you ask us: five quick checks
These answers turn a vague worry into something we can act on quickly. They are also the questions we will ask first.
- Did the site change in the last month: a new plugin, a new banner, a chat widget, a new video or a batch of large photos?
- Is only mobile flagged, or desktop as well?
- Is it one page or a whole type of page, such as every product or every blog post?
- Does the same page pass on a different day? A single bad run proves little.
- Did the report change right after something we did together, such as a launch? If so, tell us the date.
Is it worth acting on?
We cannot promise any particular score or ranking effect, and good scores do not replace helpful content. If it helps you decide how much weight to give speed against the rest of your report, see how to read your monthly report.
| Your situation | Our usual advice |
|---|---|
| Enquiries or sales are healthy and the report is amber | Note it and fold it into planned work. Do not chase a perfect score. |
| A page that matters, such as a service or product page, is Poor | Worth fixing, because slow or jumpy pages cost visitors regardless of rankings. |
| A page you rarely use is Poor | Low priority. Check whether the page is still needed. |
| A page was passing and started failing after a change | Tell us the date of the change. That is the most useful clue there is. |
| Report says no data available | A quick manual check is enough. There is nothing to fix from the report itself. |
What to expect after a fix
- Lab results change right away. Real-visitor results take weeks, so judge a fix on that timescale.
- Low-traffic sites may never get field data, so the lab tools are all you have.
- A perfect 100 is not the goal.
- Anything we change goes to a preview link for your approval before it goes live, and a second person checks it first.
Example: Illustrative example only: the numbers are made up
A cafe site shows 4.6 seconds for loading on mobile in the real-visitor section of PageSpeed Insights, mainly from a 3 MB hero photo. After the image is resized and compressed, a lab test drops to 2.1 seconds the same day. The real-visitor figure still says Poor for a few weeks, because it still includes the slow visits from before. Weeks later it reads 2.4 seconds and Search Console moves the page group to Good. The owner nearly asked for more work after week one. Nothing was wrong; the report was simply catching up.
When to ask us
Ask us if Search Console shows pages as Poor or Needs improvement, if a page that was passing starts failing after a change, or if you want a view of which fixes would matter most. Use the support form and choose Report or analytics question. A person on our team will review the cause and what to change. For visitor-facing speed problems, see my site is slow.
- The page address, and whether the problem is on mobile or desktop.
- A screenshot of the PageSpeed Insights real-user section or the Search Console report.
- What changed recently: a new plugin, a new banner, large images added, with dates if you have them.
- Files as Google Drive, Dropbox or Loom links. Never send passwords.
