Why are email open rates unreliable now?
Since Apple Mail's privacy changes, many opens are reported automatically by the mail client even when nobody reads the message. Click rates are more reliable, though still an aggregate.
Blog
The website metric that is quietly flattering your numbers
Often it is not. A finance firm I worked with believed its website converted at nearly 2%. One sister domain, recording 31.9% because its visitors arrived partway through signing up, was pulling the site average up to 1.93%, and the true rate across everything else was 0.4%. Blended averages also hide device gaps, untagged email gets filed as direct traffic, and two tools can disagree by a factor of ten. Break conversion down by source and device before you spend against it.
How can one referral source distort a conversion rate?
By converting at a rate that has nothing to do with the website. On that finance firm's site, a sister domain was sending traffic that recorded a 31.9% conversion rate because of how it was tracked: those visitors were, in effect, already partway through signing up when they arrived. That one source dragged the whole site average up to 1.93%. Strip it out and the true rate across everything else was 0.4%, against an internet median nearer 4%.
Nobody had done anything dishonest. A set-up left alone for a few years, with everyone trusting the top-line figure, produces exactly this, and acting on a flattering number is how money gets spent solving the wrong problem.
Where else do website numbers hide the truth?
In averages, in untracked channels, and in tools that disagree. The same firm's mobile conversion was 0.11% against 0.31% on desktop, and more than a third of its visitors were on a phone, so a blended number made a failing mobile experience invisible. Email traffic that is not tagged does not disappear; it gets filed as direct, so email looks weak and direct looks strong. On one campaign, one report showed 822 conversions and another showed 73. When systems disagree by a factor of ten, at least one of them is wrong, and the figure quoted is usually the kinder one.
The word "engagement" hides as much as any average. In Google Analytics 4 a session counts as engaged if it lasts more than 10 seconds, includes a conversion event, or has at least two pageviews: a rough filter against bots, and no measure of whether content is working. Email open rates have been unreliable since Apple Mail's privacy changes began reporting many opens automatically, and a 5% click rate cannot tell 5% of recipients clicking once from 1% clicking five times. Social engagement is a number the platform has every reason to show you high.
Which numbers are worth tracking instead?
The ones tied to an outcome. Conversion rate by traffic source, because the overall rate is an average that can hide a paid source converting poorly beside an organic source converting well, and moving budget between them is often the most useful decision a marketing team can make. Time from first visit to conversion, because a B2B buyer may visit for months before filling in a form. Form completion by step: if step one to step two completes at 80% and step two to submit at 40%, step two is the problem.
Some familiar metrics can be ignored. Bounce rate cannot tell a visitor who found the answer from one who hated the page. Average session duration cannot tell a confused user from an efficient one. Pageviews from people who never convert add little. The test for any metric is whether you would make worse decisions if you stopped tracking it tomorrow.
How do you check your own numbers?
Be suspicious of a comfortable number. Break conversion down by source and ask whether any single source is doing suspiciously well. Split it by device and see whether mobile is quietly failing. Check that goals fire on the final thank-you page rather than halfway through a form. If two systems report the same thing differently, find out why before quoting either.
The finance firm was about to invest in a rebuild against a conversion rate five times better than reality, aimed at a desktop experience while its customers had moved to mobile. The audit caught it before the money went out, which is why checking the numbers is the cheapest work on any project.
Related evidence: How a mid size regional UK law firm grew online enquiries by 190% without rebuilding their website. Next in this journey: What counts as a lead? Most law firm marketing reports can’t tell you.
Questions people ask
What counts as an engaged session in GA4?
One that lasts more than 10 seconds, includes a conversion event, or has at least two pageviews. That filters out some bot traffic but says little about whether the content is working.
Not sure where to start?
Most clients begin with a conversation. No pitch, no hard sell.
Just a straightforward discussion about where you are and whether I can help.
Book a free 30-minute call