Making a page lighter is satisfying. Showing that you made it lighter, in a way that would survive a sceptical colleague or customer asking “how do you know?”, takes a little more care. The good news is that it doesn't take special tools. It takes a check before the change, a check after it, a fair comparison and a short written record. This article explains how to do that with GreenTracer, and what not to claim.
Take the “before” check first
The most common mistake is to start improving and only think about measuring afterwards. By then the starting point has gone. Before you or your web person change anything, check the page and keep the report. Note the date, the grade, the grams per page view, the data measured and which device the result is based on.
If you're planning several changes, such as new images one week and removing a plugin the next, a check between each one tells you which change did what. That's useful later when someone asks whether a particular fix was worth the effort.
Compare like with like
A before-and-after comparison is only fair if the two checks measure the same thing. In practice:
- Same page, same address. Your home page and your contact page are different pages with different weights. Comparing the old home page with the new contact page tells you nothing.
- Same basis. Each GreenTracer check runs Google PageSpeed Insights on mobile and on desktop, and the headline result uses the heavier of the two. The history table shows which device each result was based on. If the “before” was desktop and the “after” is mobile, your fix may simply have made desktop lighter than mobile; look at both figures in the report before drawing a conclusion. Our article on why desktop and mobile results differ explains the usual causes.
- Compare the data measured, not just the grade. Grades are bands. A page can lose a meaningful amount of weight and stay in the same band, or cross a boundary with a tiny change. The grade boundaries page shows where each band starts and ends.
Don't read too much into one check
Live web pages are not perfectly repeatable. The Lighthouse team's notes on variability name “page nondeterminism”, such as an A/B test that changes the assets loaded or a different ad, as a major source of difference between runs, and recommend running tests more than once before drawing conclusions. Google's PageSpeed Insights documentation also describes its lab data as a simulated load on a single device under fixed network conditions.
So if a result looks surprisingly good or bad, check again before you celebrate or worry. A genuine change to the page shows up as a lasting difference in the data measured across checks. If the history marks a check as a “reused measurement”, PageSpeed Insights returned a measurement it had already taken, so it isn't an independent new test.
Keep a simple record
When you check a page again, the saved report keeps the earlier checks as history, newest first, with the date, grade, grams per page view, data measured and device for each. That is your running record, and you don't have to do anything to keep it.
Two additions make it much more useful:
- A change log. A few lines in a document: the date, what changed (“replaced 2.4 MB hero photo with WebP”, “removed old chat widget”) and who did it. The history tells you the numbers moved; the log tells you why.
- A PDF at key moments. Every report can be downloaded as a PDF. Saving one before and one after a significant piece of work gives you a fixed snapshot to file, attach to a board paper or send to a client, alongside the report link.
Re-check yourself, and when
GreenTracer does not currently re-check pages automatically. Nothing will tell you if your page gets heavier after a plugin update or a new campaign banner, so the checking is up to you. Sensible moments to check again:
- straight after any change you made to reduce weight;
- after a redesign, a theme change or adding a new plugin, widget or embed;
- before you share a result publicly or display a badge.
Results also age. A carbon result is shown as out of date 180 days after the check, and a carbon badge on your site shows the same. That isn't a penalty; it reflects that websites change. Checking the page again gives you a current figure and adds another row to your history.
Describe the change accurately
When you report progress, say what was measured. “The home page now transfers 1.1 MB instead of 2.6 MB on a first visit, and its estimated emissions fell from about 0.385 g to about 0.163 g CO₂e per page view” is clear, checkable and honest. Those illustrative figures use the Sustainable Web Design Model v4 rate of about 0.148 g per MB on a standard host. Avoid turning that into “our website is now green” or “carbon neutral”: a lighter page has a smaller estimated footprint, not none, and the figures are estimates rather than meter readings. If you want to explain what the numbers mean, our guide to reading a website carbon report is a good thing to link to.
