Why your numbers look different to ours

If you have opened your own analytics, or run a speed test yourself, and got a different answer to the one in your monthly email, this explains why. In almost every case both numbers are correct and they are answering different questions.

The short version

Your own reports tell you how the month went. Ours tells you what the Speed Layer changed. Those are not the same question, and a slow month with a well performing Speed Layer is a perfectly normal result.

I ran PageSpeed on my own site and got a different score

That is expected, and it is the most common reason people get in touch.

Google's PageSpeed score is not a fixed property of your website. It is the result of one test, run once, from one place, on a simulated phone and connection. Run it three times in a row and you will usually get three different scores. We see swings of twenty or thirty points on the same page within minutes, with nothing on the site having changed. It moves with network conditions, with which page you tested, with what your third party tags happened to be doing at that moment, and with how busy Google's own testing infrastructure is.

That is exactly why we do not report a single score.

The two numbers in your email are measured against each other, on your site, in the same conditions, over the same period. One is with the Speed Layer running and one is with it switched off. Whatever noise affects one affects the other, so the difference between them is meaningful even when the individual numbers move around.

So if you test your homepage once and see a number that does not match ours, that is not a contradiction. Compare like for like, or ask us and we will run it with you.

My analytics say traffic is down, your report says things improved

Both can be true at the same time.

Your Google Analytics shows the total for the period: every visitor, everything that happened. If this month was quieter than last, that shows up, and nothing we do changes it.

Our report is not a total. It is a comparison between two halves of your own traffic over the same days, visitors who got the Speed Layer and visitors who did not.

Your report answers "how was this month?" Ours answers "what did the Speed Layer change this month?"

Why do you hold some of my traffic back?

Because otherwise we could not prove anything.

If we simply switched it on everywhere and compared against last month, every seasonal swing, every campaign, every inventory change and every model launch would land inside the number we showed you. You would have no way of knowing which part was us.

So a slice of your visitors always runs with the Speed Layer switched off, and we compare the two groups over the same days on the same site.

It does cost a little performance on that slice. It is the only honest way to tell you the gain was ours.

Why does the comparison only cover Mondays?

This applies to sites running the Speed Layer for one hundred percent of visitors.

If everyone gets it, there is no comparison group for most of the week. So on Mondays we split the traffic and measure both sides, then go back to serving everyone for the rest of the week.

That means a monthly conversion figure can rest on four Mondays, which is a small sample. It is why we prefer to look at conversion over sixty or ninety days, where a handful of leads landing on one side or the other stops swinging the result.

Page speed does not have this limitation. It is measured continuously.

Can I see the split myself in Google Analytics?

Yes, as an audience.

When a visitor is served the Speed Layer we fire an event that places them in a Speed Layer audience in GA4, so you can segment your own reports by it.

Two things to know. It exists in GA4 only, not in Search Console. And if your site is at one hundred percent, you need to filter to Mondays for the comparison to mean anything, which is awkward to do by hand. Ask us and we will build the report for you.

My conversions are down. Is that the Speed Layer?

Look at the on and off numbers rather than the total.

If conversions are down across both groups, something else moved: campaigns, inventory, the season. If they are down on the Speed Layer side only, tell us and we will investigate the same day.

Plenty of things on a dealership website move these numbers and are nothing to do with the Speed Layer: your digital retailing tool, your chat widget, where your calls to action sit, how your lead forms are built. If you are on a manufacturer program, your field representative can go through those with you, and we are glad to join that call.

Why is this month's percentage different to last month's?

Because it is a real measurement, not a fixed specification.

The figure reflects what actually happened on your site over those thirty days: which pages people landed on, what your other tags were doing, what Google's testing conditions were like. A number that moved from month to month is a number that is being measured rather than repeated.

If it moves a long way, or turns negative, we would rather you asked us than assumed. Sometimes it is genuine noise. Sometimes it tells us the tag has been moved, or that something new on the site is loading ahead of it, and that is worth knowing quickly.

How often will I hear from you?

Once a month, with your numbers in the email itself rather than an invitation to go and find them. Reply to any of them and we will set up a call.