Honest Analytics

All posts

How to add privacy-friendly analytics to WordPress

“Privacy-friendly” is a marketing term before it is a technical one, and almost every analytics product now claims it. This is how to work out which claims hold, and how to get a working setup on a WordPress site.

What the label has to mean

Four questions separate the real thing from the branding. Ask them of any tool, including this one.

Where does the data go? If visitor data leaves your server for somebody else’s, you have a processor relationship, a transfer to think about and a third party whose policies can change. If it stays in your own database, you do not.

What identifier does it use, and how long does it live? This is the question that actually determines whether you are running a tracker. A cookie or a stable hash that persists for months can follow somebody. An identifier that is destroyed daily cannot.

Is the IP address stored? An IP address is personal data. “Anonymised” covers a wide range, from truncating the last octet to never writing it down at all. Ask which.

Can you check? Open source matters more here than for most software, because the whole question is what the code does with visitor data. A claim you can read beats a claim you have to trust.

Setting it up

Pick the tool. For WordPress the practical options are a plugin that stores data locally, or a hosted service with a script. Both can be privacy-friendly. Only one avoids sending anything anywhere. There is a comparison of the main options if you want the trade-offs.

Install it before you remove anything. Run the new tool alongside whatever you have for a few weeks. You want to see the shape of the difference before you lose the ability to compare.

Check it counts cached pages. This catches people out constantly. If your site uses a caching plugin or a CDN, the PHP that records a visit may never run. Caching breaks a lot of analytics setups, and the undercount is worst on your most popular pages, which is the worst possible place for it.

Verify from a real visit. Open the site in a private window as a logged-out visitor and confirm the visit appears. Logged-in traffic usually bypasses the cache, so testing while logged in tells you nothing.

Bring your history. If your replacement can import, do that rather than exporting CSVs you will never open again. Honest Analytics imports from Google Analytics 4, WP Statistics and Independent Analytics in the free edition.

If your banner exists mainly because of analytics, it is worth reassessing once analytics has changed. It is also worth being careful about, because a banner usually covers more than one thing.

Go through what else the page loads: embedded video, social embeds, ad tags, chat widgets, marketing pixels, fonts or maps from a third party. Any one of those can require a banner on its own, and removing the banner while one of them is still there does not improve your position.

Honest Analytics is cookieless by default and designed not to require an analytics consent banner in its default configuration. That is a statement about this plugin, not about your site. Whether your site needs a banner overall depends on everything else it runs and where your visitors are, and no plugin vendor is in a position to answer that for you.

What you should expect to lose

A privacy-friendly setup will report lower numbers than a cookie-based one, and it will report fewer kinds of number. Both are worth knowing in advance so neither reads as a fault.

Cross-day individual journeys go first. If the identifier is destroyed every 24 hours, there is no honest way to say “this is the same person as last Tuesday”, so unique visitors become daily estimates and returning-visitor rate disappears.

Demographics and interests go too. Those come from exactly the cross-site profiling the whole exercise is avoiding, so a tool that offers them is doing something you have decided against.

What you keep is the part most people actually use: how many, which pages, where from, on what, and whether it is going up.

The check nobody does

A month after switching, open the tool and ask which reports you have actually looked at. Most people find the answer is two or three. That is not a criticism of analytics; it is the argument for a simpler one, and it is the same test worth applying before adding any feature to a site.