A hybrid approach usually works best for large retailers. The deciding factor is rarely the clienteling features themselves, it is how well the tool connects to the data you already have.
Start with the integration map: list every POS, loyalty system and ecommerce platform, and check whether the platform you are considering can create one unified customer ID across them. If it cannot, the cost of workarounds grows quickly. Keep the customer data model under your control, whether that is in a CRM, a data warehouse or a customer data platform. Then the clienteling front end (store associate app, outreach tools) can be bought or replaced without losing history. Buy the commodity parts (profiles, appointment booking, messaging) and build only what is specific to your stores, such as custom recommendation rules or integration with a legacy POS. Pilot in a small group of stores first and measure associate adoption, since a tool that associates do not open will not return value, however good the features are.So the question is less build versus buy and more which layer you want to own.