However, it's essential to recognize that while usage-based pricing has proven highly effective for certain businesses, it may not be the optimal choice for all. Some companies may find greater success with a fixed-price subscription model, while others might benefit from a combination of both usage and recurring pricing options. In some cases, the pricing model may not significantly impact the overall performance of the business. An example when we compare Apple TV, which charges per movie (usage pricing), and Netflix, which employs a flat monthly fee (subscription pricing). Both companies have achieved considerable success despite adopting different pricing strategies.
Examining trends in various industries sheds further light on the matter. For instance, the telecommunications sector has transitioned from usage-based pricing to flat fee models. Remember the days when phone bills were determined by the number of minutes called? Of course, they still use usage pricing in many areas, such as international calls but such calls are now more and more conducted via much cheaper online video calling. For telcos, usage pricing is not the future, it’s the past. And if anyone has a black belt in pricing, it’s the telcos!
Still, many SaaS companies continue to explore the potential of usage pricing. The allure of this approach lies in its ability to directly align value with costs, potentially appealing to customers and vendors alike. So, let’s look at various factors to consider before deciding to go all AWS with your pricing:
1. Does usage align with the value delivered?
AWS employs usage pricing out of necessity. Their services cater to application developers, each with a unique use case offering different value while representing different workload on AWS. Given the diverse range of applications and their varying reliance on different AWS services, the company has no choice but counting tasks, jobs, containers, gigabytes, and API calls to accurately measure usage and bill accordingly. That’s common to many infrastructure software companies, including Snowflake and Stripe1.
For software designed as a business application, it’s usually easier to tie pricing to the delivered business value and the most common metric for a business is the number of employees (users). The most notorious example of this approach is Salesforce, the godfather of SaaS, which successfully employs user count as its primary metric. For other companies, metrics such as the count of vehicles or number of sites may be suitable, depending on the software's functionality. If this metric aligns with the value the software delivers, businesses can implement a flat monthly price, charging for each unit regardless of usage frequency. There is a lot to like about that!
Some business applications have the option to adopt usage-based metrics like the number of contracts, invoices, miles driven, or revenue under management for their pricing models. In some scenarios, that kind of pricing may better align with the value they deliver for every customer. Before jumping into usage-based pricing, figure out the metric that best aligns with your value and structure the pricing accordingly.
2. Do you like the predictability of recurring revenue?
One of the main reasons companies prefer the flat subscription model is its predictability. CFOs appreciate the reliability of recurring revenue because it keeps coming in quarter after quarter, as long as you keep your churn rates under control. The recurring models were the big winners of the pandemic, as the subscribers – consumers and businesses alike – remained loyal to their subscriptions even through the times of economic uncertainty.
The problem with usage-based pricing models is that they are not recurring. When customers consume, they pay you, and all is great. When usage slows, your revenue dwindles. Take Uber as an example, a company that purely charges based on usage. As the picture shows, their quarterly revenue is quite bumpy: up and down with consumption.
Compare that to Netflix with its flat recurring subscription model during the same time period, and you get the idea.
Now, some usage is more predictable than others. Storage, as measured by gigabytes or terabytes consumed, will be less volatile than transactions. Storage is cumulative; it never goes down. Even in tough times, customers generate data, and they are unlikely to delete any. Storage is a metric that effectively only goes up; sometimes faster and sometimes slower, but it always keeps growing. That’s why usage works so well for companies like Snowflake. On the other hand, other types of usage such as transactions, miles, or contracts can fluctuate dramatically.
3. How do your customers budget?
Nobody likes a surprise bill. Your energy bill probably stays within a certain range every month, but when you receive a higher than usual bill, you take notice – and you are unhappy. Customers appreciate predictable spending, and a consumption model works well only if it is predictable. Unpredictable consumption fees are only acceptable if they are negligible, and the customer doesn't bother to care about the amount.
This aspect is particularly relevant when selling enterprise technology. Enterprises have budgets set for the year and any significant overspending or underspending can be challenging, especially for your champion within the account. That's where the predictability of the flat fee model shines. When dealing with a usage model, enterprise customers prefer to lock in a committed amount of usage at a predictable cost to budget for it. That helps the budgeting, but this type of drawdown model adds a significant amount of billing complexity and frequently leads to customer satisfaction issues when they run into overages.
If you have an enterprise product, consider how your pricing will fit the customers' budgeting process.
4. Do your customers see the value in usage pricing?
Usage pricing sounds like a very fair model, where the customer pays for what has been used. It seems to work quite well for many consumer applications, especially where the metric reflects two key factors:
1. The perceived value the offering delivers
2. The perceived cost
A good example is the pricing for Uber, which is primarily based on the distance traveled. The value perception is obvious – it's more valuable to travel 50 miles than to travel 5 miles. Moreover, most consumers understand that Uber’s cost is based on the cost of fuel, the vehicle depreciation, and the driver’s time. Charging for mileage is a reasonable proxy for all that.
However, it is a little more difficult to justify usage pricing in the payments industry, where credit card companies have been pricing based on a percentage of the payment amount. The value seems aligned, but only up to a certain point. After that, the fee becomes too large. A 3% fee on a $10,000 payment is $300, and at that point, you may prefer to be paid by a check or debit. That’s why we rarely see credit cards used for large amounts.
On top of that, the cost perception does not add up. The payment amount is just a number in a computer, and the credit card company's cost is the same to process a $10 payment as it is to process a $10,000 payment. That is why the payments industry remains unsettled, with new payment methods (and even currencies) popping up all the time.
5. Does your usage pricing inhibit usage?
When you charge for usage, your customers become more aware of their usage behavior. Everything is fine as long as the cost is negligible, but the moment it becomes a budget item, your customers will adjust their behavior to optimize their spending. Enterprises will literarily incentivize their employees to reduce their usage. That is probably not what you want. You want them to use your product a lot, as long as you get compensated for the increased cost that results from increased usage.
Imagine if Spotify charged users for every song they play. Some customers would just play fewer songs or dust off the radio while others might find ways to circumvent the limitations. That is not the behavior you would want, and this is precisely why the Spotify flat fee model became so successful - and completely disrupted the music industry.
If you want your product to succeed, you need customers to use it. A lot. Your pricing should not stand in the way of adoption. That means that it needs to be either:
1. Negligible compared to the customer's buying power. A $3.99 price to rent a movie is negligible for most Apple TV customers while $19.99 is perhaps too expensive. Or,
2. Valuable compared to the alternative. Uber's usage charges may not be negligible, but they are quite valuable when compared to the cost of a traditional taxi or limousine services.
Usage pricing can lower the barrier to adoption for new customers, as it allows them to try the product without committing to a fixed fee. However, as their usage grows, so do the charges. If your pricing isn't comparatively valuable and/or negligible, it may discourage usage. When implementing usage pricing, it is crucial to ensure that the pricing structure encourages customers to adopt and use the product.
6. Can you measure the usage?
If you charge customers based on a usage metric, they will want to be able to validate the accuracy of your metering. When you get a higher than usual phone bill, you want to see what caused the spike. Similarly, if your pricing is based on a percentage of transaction amount (e.g., payments, billing, invoicing), customers will want to compare your charge with their books. If those numbers do not match, they will start asking difficult questions.
To accurately meter the usage and handle customers' questions, you will need to build the right instrumentation into your product. Adding such instrumentation is not trivial – we are talking about productized code that provides accurate counters and can be exposed to your customers via a self-service portal. That logic needs to be secure, auditable, and maintained just like any other product functionality. And unlike other product features, you probably won't be able to charge for this one.
More importantly, you will need to deal with customer service scenarios such as credits, refunds, errors, disputes, promotions, proration, service freezes, vacation holds, etc. Those capabilities need to be built, and your customer service team will need to be large enough to handle any such inquiries. Deploying usage pricing comes at a cost. Counting users is much easier.
7. Are you ready for variable considerations?
Recurring revenue models come with a fair amount of accounting complexity – from quoting to billing to revenue recognition. Quoting becomes challenging for the Sales teams as they must help customers estimate their usage and spending commitments. Revenue recognition is pretty complex for your accounting team even with the flat subscription model. In the simplest form, you may be billing your subscribers annually upfront, but you can only recognize 1/12th of the revenue after each month upon service delivery.
Usage pricing can require much more complex accounting. It may force you to adopt variable considerations, which will add significant accounting complexity, especially when there is a fair amount of fluctuation in usage between each revenue period. Your auditor may require you to estimate your revenue in advance of each period and then reconcile it with actual revenue. This accounting complexity will add additional pressure and cost on your accounting team.
8. Have you considered a mixed model?
Do you really want to adopt usage pricing? Obviously, it is the right model for many products, and for some, it may be the only feasible model. All the points mentioned above are not intended to discourage you from doing the right thing. However, there are many factors to consider before embarking on that journey.
That said, your pricing does not need to be an either-or decision. We see the emergence of mixed models that combine a monthly flat fee with some additional charges to account for usage. For instance, Apple and Disney+ offer access to some of their content for a flat monthly fee, yet they charge a usage premium for premium content, such as new releases. The advantage of such mixed models is more predictability, better value alignment, and hopefully higher revenues. However, it also adds more complexity to the pricing structure and its operationalization.
There are many possibilities out there. Choose wisely!
