Credits pricing sucks and users hate it

Before we pivoted into billing, we experienced why credits are broken. Data was an important ingredient in our first product, so we prepaid $60k in Fivetran credits. When we pivoted, we no longer needed Fivetran’s product.

‍

This kicked off our credit nightmare. We spent countless hours talking to account managers, only to realize their pricing was super complicated:

‍

1. The usage unit (aka billable metric) was monthly active rows
2. The price ber monthly active row decreased the more you used

‍

Imagine you buy a credit for $1, and at first, 100,000 MARs cost 1 credit. But once you use more than 1,000,000 MARs, suddenly 100,000 MARs cost only 0.5 credits. How do you track your credit spend?

‍

When we asked for our money back, we spent countless hours talking to account managers.

‍

‍

After many emails, much sighing and a lot of eye-rolling, our credits expired and the money had disappeared into a Fivetran-shaped hole. 

‍

Don't get me wrong. Fivetran's product is fine, the account managers just did their jobs and I’m sure all of this is in accordance with their TOS. But it was an expensive lesson that illustrates why users hate credit-based billing:

‍

Credits combine the worst of subscriptions (pay whether you use it or not) and usage-based pricing (can’t know exactly how much you’ll spend.

‍

And we’re not the only ones who dislike credits. Here's uLogMichael on OpenAI replacing free payments:

‍

‍

Yet credits are everywhere in AI, DevTools and infrastructure pricing. They enable companies to charge for usage without directly charging for usage. 

‍

But users hate buying credits. 

‍

They’re basically a purchase guarantee: The user commits to buying a certain amount, whether or not they use it all. That’s great for companies (they pad their margins with unused credits), but harms users. The upside of usage-based pricing should be that you only pay for what you use. Credits negate that benefit.

‍

But credits make it far too complex. Especially because credit pricing never seems to be “1 credit = 1 dollar” but some weird formula. The next variable is how much usage a credit actually buys? 

‍

Basically, my complaint is that buying B2B infrastructure shouldn’t feel like this:

‍

‍

It obscures the real costs and makes it hard to know what you’re actually paying for. Our users tell us they dislike credit-based billing for a few core reasons:

‍

1. Prepaid usage is a barrier to adoption: No one wants to prepay to try a product. Credits force commitment before the user actually realizes value. This hurts product adoption and conversion rates. 

2. It diminishes trust: If you lose your credits with no possibility of a refund (and especially if they expire quickly), you’re less likely to want to do business with them (again). 

3. Opaque pricing blurs the real cost. You’re using an API, but do you really know what you’re paying for? If your invoice shows “10,000 credits used,” does that mean $10 or $10,000?

‍

But credits also make life harder for companies. We speak with a lot of finance teams that struggle with revenue recognition and predictability. 

‍

If a customer buys 100 credits, you can’t recognize revenue until they’re used up. If credits expire or roll over to the next period, it adds even more complexity. Breakage adds another variable. And so on.

‍

All of this gets worse when you have multiple products that cost different amounts of credits. Credits have one major upside: A user can’t rack up a bunch of usage and abscond without paying. But is that enough to burden your team and users with credits? No. We need a way that gives users flexibility and helps companies grow. 

‍

From our customers, we often see progressive billing as an alternative that has all of the upsides, but none of the downsides.

‍

Why progressive billing beats credits 

‍

Progressive billing basically means charging users incrementally. As an example, our customer Together AI lets customers start using their product without so much as a credit card (rare for companies that charge on usage). But once they rack up $1 of usage, they get a bill. Once they pay that, they can rack up to $10 until they get a bill. After that, they get a limit of $100. If they pay all of their bills, Together AI trusts them to use the product as much as they want. 

‍

Instead of forcing users to prepay for a vague bucket of credits, progressive billing makes things easier for everyone:

‍

• No pre-payments: Users only pay for what they actually use.
• Transparent pricing: $1 = $1, not 37 credits.
• Insulation against fraud: People can’t rack up a meaningful bill and run away
• Freemium for usage-based pricing: The threshold for user acquisition gets lower because they don’t have to buy anything. 
• Easy revenue recognition: There’s just bills and clear, itemized invoices.

‍

If your model is more blended (e.g. a user gets some amount of usage included via credits), you can replace this with entitlements and include the usage directly. 

‍

Why this is extra important for AI companies

‍

This matters even more when you’re an AI company. Conventional SaaS margins are high enough to counteract the messiness of credit-based systems. But AI thins your margins by increasing your costs, which means you need to ensure you get paid.

‍

If you want your users and your finance team happy, get rid of credits and start progressive billing. 

‍

The best way to do that is with Lago. Find out how to ship progressive billing with Lago here.

‍