By category
The category decides more than the brand does. Work out which group you are shopping in and the choice inside it gets much easier.
Most bad analytics purchases are category errors rather than brand errors. A product analytics platform bought to measure a marketing site, a heatmap tool bought expecting traffic reports, an enterprise suite bought by an organisation with nobody to run it. Each of these groups answers a different question, and the first job is working out which question you have.
Tools built so that the consent banner never becomes your problem. They set no cookies, store no IP address, and answer the ordinary questions about traffic in a kilobyte or two of.
Tools for software products rather than websites. They want a designed event taxonomy and they repay it with funnels, cohorts, retention curves and experiments. Installed on a marketing.
Tools that show you what people did on the page rather than how many of them arrived. Click maps, scroll maps and recordings. The category scores low here on script weight and privacy.
The large, configurable, consent-hungry platforms that most large organisations run. Deep, capable and slow to implement, and in two cases here sold without a published price.
Tools you can run on your own hardware, where the data never leaves your infrastructure and the software costs nothing. You pay in operations instead of subscriptions, and whether that is a.
Analytics that come attached to something else you already pay for, plus the long-running independents. Cheap or free, shallow by design, and often the right baseline layer underneath a.