Why scaling SaaS platforms cost effectively is so hard...

Estimated read time: 2 minutes

Scaling of eCommerce sites is relatively easy - scaling a SaaS service is much harder for the following reasons:

• Demand is outside of your control
• Some users have disproportionate impact on capacity
• Reluctance to change existing sizing models
• Setting and delivering the customer promise on performance

Demand is outside of your control

An ecommerce business may not know exactly how users will behave, but it's largely in the control of the marketing department. SaaS providers don't have control over how they behave - for instance if the service is an ecommerce platform, you have no control over the customers' marketing teams.

Some users have disproportionate impact on capacity

The capacity impact on an ecommerce site is very collective as it's based on a large number of users, each with a small impact. However for a SaaS service, there will be customers who have a much larger individual impact on capacity, making it much harder to predict and manage hosting cost.

Reluctance to change existing sizing models

The uncertainty around demand results in a reluctance to change existing sizing models. In our experience, SaaS providers tend to oversize because of a lack of confidence in performance. Capacity (and therefore cost) will be wasted in the wrong places, while customers still experience performance incidents, because the business of getting the capacity right is so complex and full of uncertainties.

Setting and delivering the customer promise on performance

Most SaaS providers make very little commitment to customers on performance. It's common for SaaS services to have no performance SLAs at all, or a very minimum commitment. In our experience this is driven by a lack of confidence in performance within the SaaS provider. It results in poor trust with customers and blame being shifted onto the SaaS provider.

To deliver a scalable, cost effective SaaS platform, which meets customers' performance expectations, is difficult. You need to start with deeper insights into customer types, their performance and behaviour.