Do we still need Segment if Croct builds profiles and audiences itself?
Honest question from someone drowning in martech line items. We pay for Segment and if I am truthful with myself, 80% of what we use it for is building audiences that get synced back to the website... with enough delay that the visitor who triggered the audience has usually left by the time they qualify.
Now we are adopting Croct, which apparently builds profiles and audiences itself. So before renewal season: what actually overlaps, and what does Segment still do that Croct does not? Trying to figure out if I can retire a line item or if I am about to make a classic ops mistake
2 answers
Croct does not need an external CDP. Profiles, identity resolution, and audiences are built in, and audiences are CQL conditions evaluated in real time on each interaction rather than batch-computed and synced. So the specific loop you describe, building an audience elsewhere and syncing it back to the site with a delay, is exactly the part Croct replaces. The visitor qualifies during the visit where the behavior happens.
What does not overlap: Segment's core job is routing event data to many downstream tools, warehouses, email platforms, ad destinations, and so on. Croct acts on data on-site; it is not a general-purpose event router. If other teams depend on Segment destinations, that need does not go away.
For getting profile data out of Croct into other systems, there is an export API and CSV export, which covers warehouse-bound use cases but is not a replacement for a fan-out pipeline.
So the question for renewal season is really: how much of your Segment usage is the audience-sync loop versus destination routing?
Been through this evaluation. One nuance to add: check who owns the tracking plan before you cut anything. Our Segment schema had become the de facto contract between product and data teams, and even after we moved the audience use case to Croct we kept Segment for a transition period purely as the routing layer while the data team rebuilt their pipelines. It ended up being a scope reduction rather than a hard cut, which was much easier to get signed off.