Broadly, there are three ways to price any subscription software product:

  • Consumption Based - where the price charged is based entirely on how much you use, and which features you use.
  • Tiered - This is similar to consumption based, but split into coarser tiers such as bronze, silver and gold as you add features and capacity.
  • Flat Fee - Where there is essentially a flat fee for using the product with all features and in an unmetered manner.

In the analytics industry, the vast majority of companies use heavily consumption based pricing. You'll pay per user, per gigabyte ingested, per alert, per agent, or sometimes some arbitary “credit” system.

Consumption based pricing is generally positioned as better for the customer. The price you pay is tied to the value you get, and you aren't paying for capacity or features that you don't need. It also creates a low cost of entry whilst you start, and the cost only grows as your usage increases and the value becomes proven.

Risks Of Consumption Based Pricing

The problem is that the consumption based pricing models can be a bit of a trap due to banana skins such as the following:

  • The unit costs look low, but you'll use a lot of them. For instance, there might be a “per gigabyte” cost for ingestion or a “per agent” cost, but to get any value at all from the system you will realistically ingest a lot of gigabytes or need at least a handful of agents. The cost will therefore be some multiple of that low initial price before you even get started.
  • Sometimes, you go in with your eyes open to the above, but the costs grow quickly over time as you have more people using the platform and process more data or connect more servers. Something which seemed small quickly becomes a large uncapped expenditure whilst all the time you are becoming more locked into the platform. At some point, the bill lands on the CFO desk and heads start to roll.
  • There are modules or features that you find you need which aren't included in the base price. Over time, more of these get switched on and costs grow further still.
  • You might pay to scale up your environment, but because people forget or you don't have the processes in place, you never scale down, so your prices only ever go up even though you are paying for an adaptive and elastic service.
  • The price quoted for base versions can be underpowered. To run a viable production version of the software you need at least a certain amount of memory, or a certain number of instances to avoid the risk of data loss. (This is one area where vendors are disingenuous, as a datastore configured with terrible performance or at risk of data loss shouldn't even be in the market.)
  • Sometimes, vendors recognise all of the various complexity and try to wrap it all under a “standard consumption unit”. Though well intended, these can be hard and opaque for customers to understand, predict and model, and are usually more expensive than they appear.

This isn't to say the any particular tools aren't great and don't deliver increasing value, but the financial risk and exposure to the customer is real. They will start with something which they think will be small and tactical, but can end up with quite a major expenditure which they are locked into, and need to battle against to keep costs down or potentially migrate away from.

Given a choice, Consumption Based Pricing is still appealing, but when entering into these agreements for analytics tools, go in with your eyes open after realistic planning.

Hands-On Training For The Modern Data Stack

Timeflow Academy is an online, hands-on platform for learning about Data Engineering and Modern Cloud-Native Database management using tools such as DBT, Snowflake, Kafka, Spark and Airflow...

Sign Up

Already A Member? Log In

© 2023 Timeflow Academy.