Read the fine print on the popular heat-map tools and you'll find something the marketing pages soft-pedal: most of them are session recorders. They capture each visitor's full visit (every mouse movement, every hesitation over a form field), store those recordings on the vendor's servers, and render the pretty heat map from the footage afterward.
Sit with that for a second. To learn that your pricing section gets ignored, you've hired a third party to film every person who visits your website, identifiably, and keep the tapes. The insight is aggregate; the collection is intimate. That mismatch is the entire problem, and it was never necessary.
The map never needed the movie
A heat map is a sum. If 1,000 people visit a page, you don't need 1,000 replayable movies. You need to know that 214 clicks landed on the photo and that most phones never saw the button. So the honest architecture is to collect exactly the sum and nothing else. Here's everything an AnvixaAI heat sample contains:
Anonymous counters, added up as visits happen. The last row is the point: there is no per-person anything to store.
No visitor ID, no session ID, no fingerprint, no cookie, nothing typed into any form. Not "collected then anonymized," simply never created. The batch a browser sends is a handful of integers that could belong to anyone, and the stored result is a running total that belongs to everyone. If someone demanded "all heat-map data about visitor X," the truthful answer is that no such data has ever existed.
What you give up: nothing that fixes pages
The fair question: do recordings tell you something totals can't? For improving a page (the actual job), no. Watching one visitor wander is an anecdote; page decisions come from the aggregate anyway ("do most phones reach the button?"), and the totals are the aggregate, with a bigger effective sample because nothing is throttled to keep recording quotas affordable. What recordings add is voyeurism and risk: footage that can leak, subpoena targets, consent-banner complexity, and a vendor relationship your privacy policy has to explain.
There's also a quieter benefit. Because samples are anonymous and tiny, the collection can ride the same lightweight snippet that serves your structured data: no second tag slowing pages, no new vendor in your stack, and per-device maps drawn over fresh screenshots of your real pages. One switch in the portal turns it on; the same switch turns it off.
- Most heat-map tools build the map from per-visitor session recordings
- The map is a sum: collect the sum, and the privacy problem never exists
- Samples: grid-bucketed clicks, a scroll decile, a device bucket. No IDs, ever
- Aggregate totals fix pages just as well, with nothing to leak or subpoena