Comparing your website to a competitor's is one of the easiest things to do badly. Both sites are public, both can be opened in a browser, and any two numbers can be put side by side — which is exactly why most comparisons end up confirming whatever the person running them already believed.
A comparison is only worth acting on if the two sides were measured the same way, at the same moment, against the same rules. Almost none of that happens by default.
The trap
Most comparisons are rigged before they start.
The usual version goes like this: open your own site on the page you know is weakest, open the competitor on their polished homepage, and conclude that they are ahead. Or the reverse — your best landing page against their thin contact page — and conclude that you are. Neither result means anything, because two different kinds of page were compared.
Timing does the same damage more quietly. A site measured during a deployment, a cache purge, or a brief outage will look worse than it is. Measure the two sides an hour apart and you have introduced a variable bigger than most of the differences you are looking for.
None of this requires bad faith. It is simply what happens when a comparison has no rules, and it is why the first question about any competitor benchmark should be how it was run rather than what it found.
- Comparable page types — homepage against homepage, service page against service page
- Both measured at the same moment, not across a session
- The same checks applied to both, in the same order
- The same weighting, decided before the results were seen
Method
The ruleset has to be fixed before the result exists.
A fair comparison decides its rules in advance and applies them identically. Same moment, so neither side is caught mid-deploy. Same limits on what is fetched, so one site is not judged on more evidence than the other. Same category weights, so the thing you happen to be good at does not quietly become the thing that counts.
That last point is where most benchmarks fail. It is trivially easy to pick weightings after the fact that make either side win, and almost nobody notices they have done it — you simply look at the categories, decide which ones 'really matter', and produce a conclusion that was determined by the choice rather than by the sites.
This is also why the comparison on this site publishes its fairness contract on the result itself: same scan timestamp, same response limits, same category weights, a draw declared inside three points, and a critical blocker able to override a small lead. You do not have to agree with those rules, but you can see them, and they were the same for both sides.
Scope
Some things are comparable. Most of the interesting ones are not.
What compares cleanly is what both sites publish. Page titles and descriptions, heading structure, whether HTTPS is properly configured, whether tracking is present in the source, image handling, structured data, crawlability signals, the shape of a contact path. These are observable on both sides with the same method and they mean the same thing on both.
What does not compare is everything on the other side of the server. You cannot see their conversion rate, their traffic, their ad spend, their close rate, or whether the enquiries their form generates are any good. A competitor with a worse-scoring website may be winning comfortably because their sales follow-up is faster than yours, and no amount of page analysis will show you that.
So the honest framing for any technical comparison is narrow: this is how two sites compare on public, observable signals. It is a real and useful thing to know. It is not a verdict on two businesses.
Reading it
A small gap is not a finding.
Scores carry noise. A two-point difference between two sites usually reflects which specific rules happened to fire rather than any real gap, and treating it as meaningful leads to work with no return. A comparison that declares a draw inside a small margin is being more useful than one that always names a winner.
The differences worth acting on are categorical rather than numerical. One side has structured data and the other has none. One has a working contact path and the other has a form with no visible destination. One is served over HTTPS without warnings and the other is not. Those show up as whole categories, not as a couple of points.
It is also worth checking whether a single critical issue is dragging a score down. A site that is strong everywhere except one blocking problem is in a completely different position from a site that is mediocre throughout, even when the two end up on the same number.
- Ignore gaps of a few points — they are noise, not findings
- Look for categories where one side has something and the other has nothing
- Check whether one critical blocker explains the whole difference
- Re-run after changes, against the same competitor, with the same rules
Doing it
The useful output is a list of your own gaps.
A comparison earns its keep when it stops being about who won. The competitor's site is a source of specific, checkable observations: they have a schema type you do not, their pages state a price and yours do not, their contact path takes two clicks and yours takes four. Those are the outputs worth writing down.
Run it against two or three competitors rather than one. A gap that appears against every one of them is a real gap in your category. A gap against a single site is just that site doing something differently, which may or may not be working for them.
Then re-run the same comparison after you have made changes, against the same competitor, with the same rules. Movement measured the same way twice is worth considerably more than a single snapshot that named a winner.
Keep reading