This article was contributed by Griffin Parry, CEO and cofounder of m3ter.

Usage-based pricing is quickly becoming popular with software companies. It’s easy to see why: software-as-a-service (SaaS) businesses with usage-based pricing have experienced 29.9% year-over-year revenue growth, compared to 21.7% for those that don’t. What’s more, OpenView Partners estimates that seven out of the nine recent software IPOs with the best net revenue retention operated usage-based pricing. 

But usage-based pricing is surprisingly difficult to implement, and one of the biggest stumbling blocks is the quality of a company’s usage data. After all, intelligent pricing can only work if it’s built on good data. 

Here’s how companies can get the proper foundations in place for a successful, intelligent pricing strategy. 

Choose the right usage data to measure

Avoid limiting your focus. The temptation is to measure only usage vectors that drive your current pricing, but you should spread the net and measure all usage vectors that you could potentially price against, and all that drive variable costs. Ultimately, pricing improvements depend on observed behavior, so it’s better to measure first and then decide what is important, rather than the other way around.

To illustrate, imagine that you were running a backend-as-a-service for video games (which, in a previous life, I did). You are currently charging based on API calls, but think you may be leaving money on the table. You decide to start differentiating between different API calls so you can break out usage by type. 

You also decide to measure other usage vectors that could be a good basis for pricing, such as active players or the number of items in your virtual goods catalog, and that you know drive costs, such as total storage and data egress. After observing the results for three months, you start seeing patterns in the data that suggest pricing changes could drive significant bottom-line improvements.

Make sure your usage data is clean

Do you trust that your usage data is clean, complete and accurate? Or would that be pushing it? For many businesses, there’s room for improvement here. In fact, a recent IBM report found that 83% of companies suffer from data inaccuracy. 

As the old adage goes, “garbage in, garbage out.” There’s no way you can hope to run a sophisticated consumption-based pricing model if your usage data is riddled with duplications, errors and missing information. Check that your quality assurance procedures are adequate, and if you have any doubt, consider adding a dedicated data QA engineer to your team who can help to design and implement a robust testing regime. 

Be able to apply pricing flexibly

Usage-based pricing models can become complex, quickly. You’ll almost always want to offer a volume discount mechanism to reward higher consumption, but there are various options out there. Volume-based, tiered and stairstep pricing are just some of the most common. You’ll also likely consider a mechanic where you trade discounted rates for a commitment to a minimum level of spend, because this reduces perceived risk on the part of the customer and helps bring cash forward for you. 

On top of that, you might also consider hybrid models that introduce usage-based elements to feature-based tiers or per seat models. Usage allowances are a good example: when the customer reaches their allowance, they are either throttled or pay usage-based overages. 

You need to be able to store these complex pricing configurations, deal with custom terms for specific accounts, and then bring together pricing, usage data, and account details to calculate spend amounts so these can be passed to your invoicing system.  

You also need the flexibility to evolve your pricing. As mentioned above, the ideal choice for your business might not be immediately obvious — you might be surprised by customer behaviors you’re yet to discover, and it would be a missed opportunity if you couldn’t adapt to them as they emerge.  

Remember there are multiple consumers of usage and spend data

Usage and spend data are needed to drive billing on a monthly cycle. But other use cases require usage and spend data to be available, more frequently, in other parts of the stack such as the Sales CRM, the Customer Success platform, and the SaaS platform itself. 

You need to decide whether to have a single source of truth that centralizes usage and spend data in one place, or design a data governance strategy that enables data to sit in various places and assuages concerns about repetitions, inaccuracies, and manual data management leaving scope for human error.  

Linked to this is a decision about canonical mapping vs. point-to-point mapping when designing your strategy for sharing data between different systems. Canonical mapping involves a significant commitment of time and resources, and there is execution risk. But if deployed successfully, standardization allows different systems to easily communicate with each other — every application can translate its data into a single, common model that all other applications also understand.

Don’t forget your margins

Optimizing usage-based pricing is rooted in understanding the relationships between usage, spend (how usage converts to revenue), and margin (the gross profitability of that usage).  

To understand margin by customer, you need to record costs and, if these are shared between customers, allocate them. An example is cloud costs — you might have a large cloud services bill that’s driven by resources that support multiple customers (such as compute instances, databases, and storage). If you can allocate this to individual customers based on their usage, you can identify the lower margin customers and take action (usually with pricing changes) to improve overall profitability. 

Be customer-obsessed

The good news is that with usage-based pricing, your customers won’t be concerned about shelfware, waste, or failing to get the full value from their investment. They can just scale their usage and spend up or down depending on their needs.  

Also, if you’ve chosen the right pricing metric that aligns with the customer’s target outcomes, their cost will only increase if they are successful themselves. That’s great for the customer relationship and breeds a cycle of lasting loyalty. 

But remember that although customers might be fine with usage-based pricing in principle, if they underestimate their usage or have an unexpected spike, they may get a nasty surprise in their monthly bill — a risky moment that could mark the end of your relationship. You need to be able to proactively manage these situations by responding to usage signals.

Remember too that in a usage-based world, pricing is part of the product experience. Customers want to know how much they’re spending and how their usage drives that spend. Make sure you always have that data available in as close to real-time as possible, so they can easily check their running totals at any time and see a forecast of their next bill.

Making a success of usage-based pricing 

Ultimately, software companies are turning to usage-based pricing to capture their true value, driven in particular by trends in automation and Product-Led Growth. 

The model works because it enables value-based pricing that scales easily with a customer’s success — but it isn’t easy to implement, needs to be designed well, and can’t get off the ground without an underpinning of trustworthy, reliable data.

Choosing to meaningfully invest in your data infrastructure is a good place to begin, giving you the foundation both you and your customers can be confident in when navigating the complexities of intelligent pricing and billing.

Griffin Parry is the CEO and cofounder of m3ter.