Welcome back to Good Better Best.
Each week, we break down real pricing, packaging, and product moves from SaaS and AI leaders and share the ideas worth stealing.
This week we're breaking down Notion's new AI credit system. Every Business and Enterprise seat now gets two separate wallets, split by who picks the AI model. I'll walk through how it works, why Notion drew the line where it did, and why I think it’s way too complicated.
Before we get there — next week, we’re hosting an AI Credits Roundtable with 3 pricing leaders who are in the trenches right now with AI credits: Rita Sherman (Monetizaton Lead at Gong), Jeff Curran (Pricing Lead at Miro), and Manu Mehra (Strategic Pricing Lead at Databricks).
It’s going to be an awesome session. Grab your seat here.
On to today’s post.
🔌 PricingSaaS Partners power the next era of SaaS pricing

This Week in Pricing, Packaging, and Product
This week we observed 260+ changes. The highlights:
Cursor rebuilt its lineup around a Grok/X bundle: Plus $8, Premium $30, Super $100, with Teams cut to $30+ [Link]
Microsoft 365 now flags usage-based AI charges beyond license fees [Link]
Intercom dropped the Fin Million Dollar Guarantee and its 90-day money-back pledge [Link]
ClickUp raised its Core plan from $8 to $9/user/month (billed yearly) [Link]
Figma added a Specs beta to paid seats [Link]
Vercel added Daemons with usage-based Active CPU and Memory pricing [Link]
Stripe added a Postgres database sync tier to Data Pipeline [Link]
GitHub renamed Copilot secret scanning and added AI push protection detection [Link]
Mixpanel cut its Growth yearly price to $100/month at 18M events a year [Link]
Dialpad Sell slashed Premium to $79/user (-47%) and Advanced to $59 (-38%) [Link]
Basecamp added AI agent integrations (Claude, ChatGPT, Grok Bot) to all plans [Link]
Lindy hiked its Team plan to $49.99 per user and bumped credits to 5,000 [Link]
Mixmax swapped tiered plans for one all-features plan priced by recipients [Link]
Orb ditched opaque tiers for $6/credit self-serve pricing [Link]
PagerDuty revealed its AI Actions quota: 1,000 to 100,000 a year by tier [Link]
Sumo Logic added a DojoAI Credits add-on for pooled monthly AI prompt capacity [Link]
Wistia added AI credits to every plan [Link]
DeepSeek launched Search and Fetch APIs at $0.75 per 1K requests [Link]
Fal doubled its Flux 3 image edit price to $0.048 per megapixel [Link]
Vendr folded its Negotiation Agent and Procure-to-Pay into core plans [Link
Impact.com revealed its Enterprise price ($10,000/mo) and restructured tiers [Link]
Unity set a $0.75/GB Asset Manager storage fee across all plans [Link]
Check out more updates on PricingSaaS →

I love Notion. I run my business on it and pay for it every month.
So when Notion emailed me Monday about a new AI credit system, I was intrigued as both a customer and a pricing strategist.

My first read was that the new model is unnecessarily complicated, and after diving deeper, I’m even more convinced that’s the case.
As AI makes products more complex, there will inevitably be new considerations under the hood that drive pricing and packaging. I still believe it’s the job of SaaS and AI companies to abstract away that complexity as much as possible to make it easier for users to grasp. I also believe the companies that do a good job of this will be rewarded. The better you are at explaining pricing dynamics, the better story you can tell around your product, it’s value, and how it drives results for your customers.
Before we get into the specifics of Notion’s update, I want to do a quick recap on how Notion’s AI strategy has evolved. As you’ll see, it has gotten increasingly complex over time, but until this update, it was still easy to understand. I’d argue that’s no longer the case.
Notion’s AI Pricing Evolution
First, Notion launched Notion AI as an add-on, charged at $8 per month (or $10 on a monthly plan). With the add-on, there was no usage metering — just a straightforward up-charge.

Next, they bundled the add-on into Business and Pro plans, and introduced their agent. They still didn’t introduce a usage component, but the Notion AI trial on Free and Plus plans suggested they would start metering in some way.

Next, they introduced Custom Agents, and with it launched a simple Credit Model that cost $10 per 1,000 credits. Notably, there was still no credit balance for Notion Agent, which remained included in the Business and Enterprise plans.

Now, with the latest update, they’ve introduced two different credit balances: Standard and Select, and both Notion Agent and Custom Agents consume credits.

What Changed?
Going forward, each Business seat gets:
Standard credits: 1,500 per member per month. These pay for work where Notion picks the model: Notion Agent, writing help, and Meeting Notes.
Select credits: 500 per member per month. These pay for work where you pick the model, plus Custom Agents.
Both balances reset monthly and don't roll over. You can't combine, trade, or share them. If you run out of one, that work pauses until the reset, even if the other wallet is full.
As if that wasn’t enough, behind those two sits a third credit type: Notion credits.
This is a shared workspace balance that can be used to replenish either Standard or Select credits. Here’s how Notion breaks it down in their FAQ:

If this seems confusing, that’s because it is!
If Notion credits can cover Standard or Select credits, why have both allowances? Why not just have a single pool that all work can draw from?
And if you absolutely HAVE to split credits into multiple balances, there has to be a better way to do it than by who chooses the model.
For example, keep Custom Agents on their own credit balance. Then users who don’t have Custom Agents don’t have to think about those credits at all, which is how it was before.
It’s worth mentioning that Notion introduced a lot of new functionality as part of this announcement, and it would appear that may be influencing the new pricing.
Along with the new credit model, Notion announced:
An improved router that automatically selects a model based on cost and speed
A better harness that can perform long-running tasks cost-efficiently
The ability to bring your own AI (ChatGPT for now, others on private Alpha)
To recap, Notion now offers the ability to use their built-in agent, create your own agent, or both. Along with that, you can use Notion’s router, select the model yourself, or bring your own key from another AI tool.
It’s not lost on me that offering this level of flexibility adds complexity. When I posted about this earlier in the week, I heard from two pricing leaders I really respect that offered different explanations.
Sam Lee, Head of Monetization at Stripe, suggested it could be due to technical challenges that will resolve over time.

Jeff MacMillan, who led pricing at Autodesk, suggested it might have to do with cost forecasting.

These explanations make sense, but it’s still on Notion to figure out how to deal with this complexity rather than handing the job off to customers. While Notion is far more than a note-taking app now, most of its users aren’t billing engineers.
Why it Matters
I recently watched a great video from Sequoia Partner Pat Grady breaking down the current state of AI. In the video, he shared a slide about The Diffusion Gap, or the gap between AI capabilities and AI adoption.

It’s easy to forget that most people aren’t using AI agents every day. I’m reminded every time I talk to the rest of my family, and they look at me like an alien when I describe how I’m using AI (and I’m not even doing anything special!).
It’s clear that AI cost management has become a fundamental discipline for software companies. It’s also clear that flexibility is becoming really important for end users. But the combination of those two can get really complicated. The further we get into all of this, the more deeply I believe that the people that can make this stuff easy to understand will be able to bridge the diffusion gap and have a massive advantage.
Notion is one of the companies I think should be leading the way, which is why this updated model felt like such a miss to me.
Thanks for reading! If you’re working on AI monetization and want to learn more about how we help, book time here.
Until next time,
Rob