Appen's Competitive Moat, page-6

I think you guys disregard the value of Appen retaining the IP on the data they collect / provide and the ability to re-sell this. Whilst it's unlikely two clients will look for exactly the same thing - there will be similarities and so they won't need to start from scratch for each project and this is where they have a moat.

They'll build up this data set over multiple projects which may be why they are seeing less margin on these new client / projects - they are having to build the data sets from scratch.

Therefore, I would say they have a moat in the advertising/technology space (which seems to be tapering off) and are currently building one in the new verticals / products.

I work at a data company (Marketing Data) and if someone wanted to throw resources at building a large dataset from scratch I say good luck to them - it is not easy.