Every year, we've seen the same SaaS analytics startup get funded by the same VCs over and over again.
"X startup raises Y capital from Z firms promising to solve analytics", providing SDKs for sending data to their centrally hosted cluster.
If you are looking to this class of tools to accomplish data insight and impact at scale, you should be wary.
SaaS analytics tools will not scale with your success. Here are three main reasons:
1) Opinionated schema means format inflexibility
Developers frequently start their journey in SaaS analytics with a tool like Google Analytics, Mixpanel, etc. It's usually trivial to add SDKs to get the data from your app into their centrally hosted cluster.
However, usually these SDKs come with an opinionated schema you must adhere to. Why?
Because the purpose for an inflexible, opinionated schema is for these companies to precalculate and cache answers to questions they deem important (not you) to achieve the illusion of speed and scale.
But as your product evolves, and there comes the need to iterate your data formats to keep up with product changes and growth. The SaaS analytics tools will hinder you from fulfilling this need.
2) Scale is unachievable
Once your product reaches a moderate level of scale, you'll have many sources of data you'll want to unify to achieve a holistic understanding of your product.
What if one of these sources of data is your SAAS analytics tool?
It will not be trivial. Having to write scripts to batch export the data from these SAAS analytics solutions is manual, slow, and not a pleasant experience.
Additionally, having to send this data over the internet to your datacenter will be sluggish due to network latency.
It is important that you prioritize these data sources be housed inside the same datacenter from the start to circumvent these avoidable challenges.
3) Privacy becomes a second-class citizen
Privacy becomes a second-class citizen when sending data outside of your servers to a SAAS analytics company.
Health startups with sensitive patient records are required to achieve a level of privacy called HIPAA compliance, something they could never accomplish by sending data to a third party SAAS tool's centrally hosted cluster.
If protecting the data of your users is at the top of your priority list, it's best not to send data to a cluster you don't own and operate.
In summary, working with SAAS analytics tools can often hinder your productivity rather than improve it. Inflexible formats can slow down product iteration, and true scale and privacy are often not achievable with this class of tools.
Instead, choose a single tenant or on-premise solution that can scale with your success.