Most speed reports measure the wrong visit, which is why so many sites pass a test and still feel slow to the people who matter.
Key Highlights
- Core Web Vitals are scored on field data at the 75th percentile of all page loads, not on an average and not on your own device.
- Search Console uses a rolling 28-day window of Chrome User Experience Report data from real visitors.
- A URL group takes the slowest status assigned to it, so one poor metric marks the whole group poor.
- Interaction to Next Paint replaced First Input Delay as a Core Web Vital on 12 March 2024.
- Google states there is no single page experience signal, and that good scores do not guarantee top rankings.
- Sites with too little traffic get no report at all, which is common for smaller UAE businesses.
- Search Console blends data from every country, so traffic from slower-connection markets drags the whole figure down.
What a Core Web Vitals Audit Actually Measures
Three things, each describing a different kind of frustration. Largest Contentful Paint measures how long the main content takes to appear, and is good at 2.5 seconds or under, needs improvement up to 4 seconds, and poor beyond that. Interaction to Next Paint measures responsiveness when someone taps or clicks, good under 200 milliseconds, needs improvement up to 500, poor above. Cumulative Layout Shift measures how much the page jumps about while loading, good below 0.1, needs improvement to 0.25, poor above.
A proper audit shouldn’t stop at those three numbers. It needs to show where those numbers came from, which pages they actually relate to, whether there’s enough data to take them seriously, and which of the three is really costing you. A good digital marketing agency in Dubai should be able to lay out all four before suggesting even a single development ticket. Otherwise, there’s no real way to know what needs fixing. The recommendation should come from the metric that’s failing and, just as importantly, the template where it’s happening.
The audit also has to separate what you can fix from what you cannot. Server response time, image handling and third-party scripts are yours. The visitor’s phone, their network and their location are not, and a report that treats those as failures is setting up work that cannot succeed.
Why the 75th Percentile Changes Everything
Because it means your own experience of the site is almost irrelevant. Core Web Vitals are scored, in Google’s words, “based on how well they perform in the field at the 75th percentile of all page loads.” Search Console puts the same idea plainly: the group value is the point at which “75% of page requests took this amount of time or less.”
That one rule explains the most common argument between a business and its developer. The developer loads the page on office fibre and a recent laptop and sees two seconds. The 75th percentile visitor is on a three-year-old Android handset on a busy mobile network, and sees five. Both are real. Only one is being scored.
It also changes what counts as progress. Improving the fastest quarter of your visits does nothing to the figure, because the figure is set by the slower end. Any digital marketing agency proposing performance work should therefore be optimising for the slow quarter deliberately, which usually means weight and server response rather than polish.
Is Field Data the Same as a Lighthouse Score?
No, and conflating the two is the single most common error in performance reporting. Search Console’s Core Web Vitals report uses field data from the Chrome User Experience Report, which gathers anonymised metrics from actual users visiting your URLs. A Lighthouse or PageSpeed Insights lab score is a simulated load on a standardised device and connection.
Both are useful and they answer different questions. Field data tells you what real visitors experienced over the last 28 days, which is what Google assesses. Lab data tells you what is wrong with the page and lets you test a fix immediately, which field data cannot do because it lags. Use lab tools to diagnose and field data to judge.
The practical rule is that nobody should accept a performance claim based only on a lab number. A green Lighthouse score on a developer machine and a poor Core Web Vitals assessment in Search Console are entirely compatible, and the second one is the one Google uses.
Does a Slow Site Really Hurt Your Rankings?
Yes, and less dramatically than most pitches suggest. Google states that Core Web Vitals are used by its ranking systems and recommends achieving good scores. It also says good scores in the report do not guarantee top rankings, and that chasing perfect scores purely for search may not be worth the effort.
The bigger point is really about how the whole thing works. Google has made it clear that there isn’t one single page experience signal, and its core ranking systems look at several signals together. Page experience tends to matter more when a lot of genuinely useful pages are competing for the same search. So, in simple terms, a slower page with the best answer can still win, while a very fast page with a weak or thin answer can still lose.
So the honest version of the ranking argument is narrow. Speed is a tie-breaker in competitive results rather than a lever that lifts weak content. Where it pays unambiguously is in what happens after the click, which is the part of the case most agency decks skip.
Thresholds, Groups and the Slowest-Status Rule
The mechanics of the report decide how your results look, and two of them surprise people.
URL grouping: Search Console groups URLs with similar experience and applies one status to the whole group. Fixing one page in a group of forty will not move the group.
The slowest status wins: A group’s status defaults to the slowest status assigned to it for that device type. A group with poor Cumulative Layout Shift and only a needs-improvement Largest Contentful Paint is reported as poor.
Sampling and indexing: The report shows a sample of pages and includes only indexed URLs, so it is not a complete list of your site.
Actual URLs, not canonicals: Data is attributed to the URL that was loaded rather than to the canonical, which matters on sites with parameters or duplicates.
A 200-row ceiling: The table lists up to 200 rows, so large sites need the API or a crawler to see everything.
Read together, those rules mean a single bad template can mark a large part of your site poor. That is bad news for reporting and good news for prioritisation, because it also means one template fix can move a great deal at once.
When Search Console Shows You Nothing
This happens constantly to smaller businesses and it is not a fault. A URL group without enough data is left out of the report entirely, and Google states that a minimum amount of data is required in order to respect user privacy. No numeric threshold is published.
Search Console has a fallback. Where a URL group is too small, it can report at origin level instead, covering every URL under that protocol, host and port. If the origin also lacks sufficient data, nothing is shown, and a property displaying “No data available” is either new or does not have enough Chrome User Experience Report data for the device type selected. New properties can take a few days for existing data to appear.
For a modest UAE site with a few hundred monthly visitors, the realistic position is that field data will never be available, so the audit has to run on lab tools and server metrics instead. Saying so at the start is more useful than producing an empty report, and any best digital marketing agency in Dubai shortlist should include whoever tells you that first.
Counting the Commercial Cost Honestly
There is a real study behind the business case, and it needs its limits attached. Google commissioned research conducted by 55 and Deloitte, published as “Milliseconds make millions”, covering 37 leading European and American brand sites and more than 30 million user sessions across retail, luxury, travel and lead generation. Mobile load times were monitored hour by hour for 30 days at the end of 2019.
The reported effects of a 0.1 second improvement were substantial. Retail conversion rates rose 8.4% with spending up 9.2%. Travel conversion rates rose 10.1%. In luxury, progression to the Contact Us page rose 20.6%. In lead generation, progression to the form submission page rose 21.6%, with pageviews up 7%.
Four caveats matter before anyone puts those numbers in a proposal. The study was commissioned by Google. It is observational rather than a controlled experiment, and the published write-up offers no caveat about causation. The metrics improved were First Meaningful Paint, Estimated Input Latency, average page load time and Time to First Byte, which are not today’s Core Web Vitals, and First Meaningful Paint has since been deprecated in favour of Largest Contentful Paint. And the sample was European and American brands in 2019, with no UAE or wider Gulf data in it. The direction is credible and well evidenced. The specific percentages are not a forecast for a Dubai business, and a digital marketing company in UAE quoting them as one is overreaching.
Who Owns Each Fix?
Performance work stalls when the audit lands on the wrong desk, so the findings should be sorted by who can actually act.
| Finding | Usually fixed by | Typical effort |
| Slow server response time | Hosting provider or backend developer | Low to moderate, sometimes a plan change |
| Oversized or unoptimised images | Content team with a build step | Low, and the most common quick win |
| Layout shift from ads or embeds | Front-end developer reserving space | Low per instance, tedious at scale |
| Render-blocking scripts | Front-end developer | Moderate, needs testing |
| Third-party tags and chat widgets | Marketing, by removing what is unused | Low, and politically awkward |
| Theme or framework overhead | Developer, sometimes a rebuild decision | High, and worth questioning |
| Fonts loading late | Front-end developer | Low |
The pattern across most audits is that marketing owns more of the problem than it expects, because tag managers accumulate. Before commissioning development work, it is worth asking a best digital marketing company in Dubai candidate how many third-party scripts they would remove, and whether anyone has audited what each one is for.
A Six-Week Audit You Can Actually Run
Keep the sequence cheap at the start, because half the findings cost nothing to establish.
Weeks one and two, establish whether you have data: Open the Core Web Vitals report and check whether URL groups appear, whether the view has fallen back to origin level, and whether either device type shows nothing. If there is no field data, say so in writing and plan the audit on lab tools.
Weeks three and four, diagnose by template rather than by page: Identify which templates the failing groups represent, run lab tests on a representative page from each, and record server response time separately from front-end findings.
Week five, sort by owner and effort: Split the findings across hosting, development, content and marketing, and start with the items that need no code.
Week six, fix one template and measure the right thing: Change a single template, confirm the improvement in lab testing immediately, then wait for the 28-day field window to turn over before judging the report. Treat those weeks as a working order rather than a schedule, since hosting constraints and development queues will stretch them.
Five Reports That Flatter, and One That Helps
A single PageSpeed Insights screenshot: One lab test on one page, presented as a site-wide verdict.
A score out of 100 with no metric detail: The score is a weighted lab composite, not the three metrics Google assesses.
Desktop results only: Mobile is usually where the 75th percentile visitor sits, and the two are reported separately.
A green result taken from the developer’s own browser: Fast device, fast connection, no bearing on the field assessment.
Averages instead of percentiles: An average hides the slow quarter that sets your actual score.
A digital marketing agency in Dubai that produces any of those five is not measuring your site, it is reassuring you. The report that helps shows the field assessment per URL group and device, names the template behind each failing group, separates what hosting owns from what development and marketing own, and states plainly where there is not enough data to judge. Asking a digital marketing company in Abu Dhabi or anywhere else for that specific format is the quickest way to find out who understands the measurement.
From Audit to Something Worth Billing For
A Core Web Vitals audit is cheap to run and easy to misread. Measure the field assessment rather than your own browser, work by template rather than by page, separate what hosting owns from what marketing owns, and accept that a small site may have no field data to judge at all. The commercial case for speed is real and the published percentages belong to other people’s websites. Any digital marketing agency worth retaining will say both halves of that sentence.
At Savit Interactive UAE, performance sits inside our web development practice rather than beside it, which is why a technical implementation brief and a Core Web Vitals finding usually reach the same team. Our process puts discovery and audit before strategy for exactly this reason: the audit frequently changes what the strategy should be, and occasionally it tells a client to spend on hosting instead of on us. We offer a free digital marketing audit as the starting point, with reporting that covers performance alongside visibility rather than burying it, and our in-house specialists across SEO, development and content mean a layout shift problem does not need three suppliers to resolve.
Footnotes and References
[1] Google Search Console Help. “Core Web Vitals report.” Google states that the report uses Chrome User Experience Report field data from actual users over a rolling 28-day window, that a URL group’s value is the 75th percentile, that a group takes the slowest status assigned to it, that groups without enough data are excluded in order to respect user privacy, and that data from all locations is combined.
[2] web.dev. “Interaction to Next Paint becomes a Core Web Vital on March 12.” Google confirms that Interaction to Next Paint replaced First Input Delay on 12 March 2024, that Search Console dropped First Input Delay on that date with a six-month deprecation period in other tools, and that all Core Web Vitals are scored on field data at the 75th percentile of all page loads.
[3] Google Search Central. “Understanding Core Web Vitals and Google search results.” Google gives the good thresholds of 2.5 seconds for Largest Contentful Paint, 200 milliseconds for Interaction to Next Paint and 0.1 for Cumulative Layout Shift, and states that Core Web Vitals are used by its ranking systems while good scores do not guarantee top rankings.
[4] Google Search Central. “Understanding page experience in Google Search results.” Google states that there is no single page experience signal, that several signals are considered together, and that page experience matters most where many helpful pages compete for the same query.
[5] web.dev case studies. “Milliseconds make millions.” Research commissioned by Google and conducted by 55 and Deloitte across 37 European and American brand sites and more than 30 million sessions, with mobile load times monitored for 30 days at the end of 2019, reporting conversion rate gains of 8.4% in retail and 10.1% in travel associated with a 0.1 second improvement.


