How Plausible and Fathom work
Both are hosted web analytics services built on the same idea: a very small script, no cookies by default, and a single readable page of reports instead of a reporting suite. They count visitors, pages, referrers, countries, devices and campaign tags, and they let you record a handful of goals such as a form submission or an outbound click. Plausible is open source and can also be self-hosted; both are sold as subscriptions, priced against the volume of pageviews you send.
Because they avoid cookies and cross-site identifiers, they do not follow a person from visit to visit the way a cookie-based tool does. Returning visitors are estimated rather than individually recognised, and there is no user-level profile to interrogate. That is a deliberate trade: less depth, in exchange for a lighter page and a simpler privacy position.
Why Plausible and Fathom matter
The everyday argument is weight and clarity. The script is tiny, so it barely touches page speed — which counts for more on mobile data in Nepal than it does on office broadband — and the dashboard can be read by a business owner without training or a course. Many small sites never used more than the top few rows of a full analytics report anyway.
The privacy argument matters more to some organisations than others. With no cookies and no personal identifiers, the consent conversation becomes shorter, and there is no advertising platform on the other side of the data. If a client contract or an internal policy makes that a requirement, these services answer it without the server work that self-hosting brings.
Common mistakes with Plausible and Fathom
Choosing one and then asking it to do a job it was never built for is the usual error. There is no funnel exploration, no cohort analysis, no user-level path report and no native route for feeding conversions into Google’s bidding. If you run paid campaigns that depend on conversion signals, you will still need the ad platform’s own tracking alongside.
The second is expecting the totals to agree with your old tool. A cookieless service counts a returning visitor differently from a cookie-based one, and any script can be blocked. Compare the direction of travel rather than the absolute numbers, and never present two tools’ figures side by side as though one of them is simply wrong.
What to do about it
Start from the decisions you actually make. If the questions are whether traffic is growing, where it comes from, which pages get read and whether enquiries are rising, these tools answer them and remove a large amount of configuration you would otherwise have to maintain and explain. If the questions involve segments, funnels, customer lifetime value or feeding an ad platform, they will not, and something fuller is required.
Running both for a period is a legitimate answer. Put a lightweight service on for the numbers the business reads every week, keep GA4 for depth and for advertising, and be explicit in reports about which tool a figure came from. Whichever you settle on, write down what is measured and how, because an undocumented setup gets misread long before it gets outgrown — that documentation is most of what good analytics and tracking work consists of.