"It’s always more work to work with a company to get the last mile of the custom use case… than it would have been to just do it yourself."
This is true if all you need is a out of the box basemap with "data layers"
Many companies need to process data in real time or at least nightly to get it into a format that has all the information they need usually cross referenced and enhanced with multiple related data sources. These can be related via a primary key and simple or related via a simple or complex spatial relationship or spatial plus primary key relationships. This will require custom ETL pipelines that a utility company, for instance, doesn't have the in house expertise to create and maintain.
This also assumes you have the supporting elements already in place. Utilities often need to host data themselves for security or legacy reasons so they can't just use a cloud service. There are usually tools built on top of the data that need to be updated. A new geospatial system usually involves a lot more than just a new data layer. If that's all you need, then maybe these companies like felt are a good solution.
Felt seems to be primarily a visual map annotation tool ala building visualizations. I doubt the data part would need to be that complex and custom. It could be a generic 'view' layer providing just the top layer and leave the data sources wide open and flexible but for the current offering it makes sense it's building off a 'library' of sources.
Although every B2B product will have customers who demand startups be their personal software consultants. That doesn't always translate into a repeatable product.
In my own experience, either the product needed very little map features, and it didn't warrant paying in full just for that (I remember even live Google Maps integration being dismissed as "too costly" and moving to pre-rendered optimized maps).
Either we needed a lot more maps, and having full access to the underlying data made more sense to limit computational overhead and better adapt to our use case.
Basically map/location related features are hard and often costly, I expect companies to either avoid them completely or bite the bullet and go full in.
I can’t find it now, but a while ago there was a post about map/data startups that basically boiled down to: Don’t.
The argument went like this:
People who want geospatial visualisations almost always want something custom.
Companies selling geospatial products believe that they can churn the data into “data layers” that serve “use cases” that mostly address this.
It’s always more work to work with a company to get the last mile of the custom use case… than it would have been to just do it yourself.
There aren’t that many paying customers.
…
I don’t know how true that is in reality, and the article was more nuanced than my dumb summary, but… really? A subscription product to annotate maps, which has built in “data layers” so “you don’t have to”?
Sounds very familiar.