Personalising for anonymous visitors without collecting personal data, realistic?
Our previous personalisation vendor leaned heavily on third-party data enrichment, and it failed our internal audit this spring. I direct marketing at a financial services firm, so I am starting again with a much stricter brief: first-party behavioural signals only, and nothing that identifies a person unless they have an account with us.
My honest question: is personalisation under that constraint actually useful, or do you end up with segments so coarse they are not worth the effort? Most of our traffic is anonymous and will stay that way. I would rather hear a realistic answer now than discover the limits after we have committed
1 answer
Yes, it is genuinely useful, and anonymous is the normal case, not the degraded one. Anonymous visitors get full behavioural profiles built from first-party signals: behaviour, campaign source, location, device, interests. Audiences evaluate on those profiles in real time without a single personal attribute existing anywhere.
Concretely, the segments are not coarse. These are all ordinary conditions:
user is anonymoususer is returningsession's stats' pageviews > 3Compose them with campaign, location, and interest signals and you get "returning anonymous visitor from the ISA campaign who has read three pages this session", which is plenty sharp to act on. This is exactly the pattern documented in the anonymous users playbook if you want it laid out as a recipe.
And for the visitors who do eventually open an account: identity resolution links the anonymous history to the identified profile without losing data. So there is no need to collect anything pre-emptively "just in case", which sounds like precisely the practice your audit flagged.
This is the answer I needed, particularly that identification can happen later without losing the history. That removes the last argument for pre-emptive collection. Taking the playbook to our compliance review.