How Intercom Powers Outcome-Based Pricing with Fin AI

Plus: Updates from Hugging Face, Wistia, and Jamf.

Welcome back to Good Better Best.

Later send than usual today due to a virus/stomach bug hybrid that annihilated our household this week. TGIF!

Everyone is buzzing about outcome-based pricing, but today we’re breaking down the keys to implementation — using Intercom’s Fin AI as an example.

After that, Ulrik breaks down what people are getting wrong about outcome-based pricing, and why it all comes back to the value chain.

Let’s get to it.

---

📰 Pricing News & Updates

• Hugging Face added value to its Pro and Enterprise plans.

• Wistia updated its plan positioning.

• Jamf added new plans.

---

---

Today’s Post is brought to you by Stripe

Is your billing system hindering your growth? 69% of leaders want flexible pricing, but most billing systems can't deliver.

Read Stripe’s latest report for strategies to implement creative pricing models, reduce involuntary churn, and grow revenue. Boost your revenue potential with Stripe Billing.

---

The Mechanics of Outcome-Based Pricing

The industry is buzzing about AI’s ability to power outcome-based pricing. While I’ve heard many folks talk about the impact on pricing models, I haven’t seen anyone write about how you actually implement it.

So I did a deep dive on Intercom’s Fin AI to better understand what they’ve done, and have tried to abstract the most important steps that other SaaS players can follow. Special thanks to Perplexity Deep Research and Claude, who assisted in this effort.

Here are the key steps as I see them.

1. Technical Architecture

Fin processes millions of monthly resolutions, and maintains an out-of-the-box resolution rate of 51% due to a sophisticated technical architecture:

1. Claude AI powers advanced reasoning and multilingual capabilities

2. Intercom’s Proprietary AI optimizes query interpretation and response validation

3. Real-time translation of 45+ languages helps preserve cultural nuances

4. Integrations with platforms like Zendesk, Stripe, and Shopify held drive resolution accuracy.

2. Operational Metric Definition

The foundation of Fin's pricing model hinges on a clear operational definition of what constitutes a "resolved" customer interaction. Intercom employs a dual-path system that captures both explicit and implicit success signals:

Hard resolutions occur when customers actively confirm satisfaction either by clicking the "That helped 👍" button or responding with affirmative text like "Yes" or "That solved my problem."

Soft resolutions are inferred when customers either close the conversation window without further requests or don't respond within 24 hours of Fin's last message.

This blended approach prevents both undercounting (when customers get what they need but don't explicitly confirm) and overcounting (by distinguishing genuine resolutions from abandoned conversations).

3. Billing System Definition

The billing mechanics further refine this model:

• Multiple resolutions within a single conversation count as just one billable event

• Customers have a 24-hour window to revisit and potentially change a resolution status

• Conversations started in the last 48 hours of a billing cycle roll into the next period

This structure prevents double-charging for follow-up questions while giving customers adequate time to confirm resolution quality. The 24-hour window particularly addresses edge cases where users might initially accept an answer but later discover inadequacies.

4. Outcome Optimization

Once technical, operational, and billing systems are able to power, define, and bill for outcomes, the next step is being able to deliver them.

The specifics will vary depending on the company and product, but I believe the following items are widely applicable.

→ Content Targeting

• Define audience segments based on key attributes (e.g., location, product)

• Map audiences to relevant content (e.g., Regional Knowledge Bases, Product-Specific Answer Banks)

→ Workflow Design

• Design clear escalation rules for complex situations where an outcome can’t be completed with AI alone

• Establish multi-step verification processes to ensure outcomes are actually completed

→ Continuous Model Training

• Create feedback loops from unresolved outcomes

• Deploy proactive satisfaction surveys to capture user feedback

5. Risk Mitigation

Now you’re successfully solving for outcomes. Great! But one of the obvious risks of an outcome-based model is that it can be far less predictable than a license model. To account for that, Intercom implements safeguards to ensure customers aren’t charged more than expected.

1. Usage alerts at 50%, 75%, and 90% of resolution limits.

2. Hard caps to automatically disable Fin upon reaching budget thresholds.

3. Custom Answer prioritization to maintain critical responses during limit breaches.

These controls enable predictable budgeting while preserving essential support functions during peak periods.

Conclusion

While I believe these 5 areas cover some of the most important mechanics of outcome-based pricing, they’re definitely an over-simplification. If there are other layers that I’m not thinking about, I’d love to hear them 👇🏼

---

🎯 Expert Insight

When I think about outcome-based pricing, I find myself slightly annoyed by how the term gets tossed around in SaaS circles. People frame it as binary — either you're doing outcome-based pricing or you're not. But after years of working with companies on their pricing strategies, I've come to see it differently.

The truth is that outcome-based pricing exists on a spectrum. It's not about whether your pricing is outcome-based, but to what degree it aligns with outcomes.

The Value Chain Perspective

What crystallized this thinking for me was visualizing the value chain—that sequence of events with causal relationships that transforms inputs into valuable outcomes.

I like to think of value chains as rainbows, with the pot of gold sitting at the end.

• At one end, you have inputs: users, budget decisions, resources allocated.

• At the other, you have outputs: increased revenue, improved efficiency, better customer retention, or whatever the ultimate business goal might be.

Let me share a real example that shaped my thinking.

Back in 2019, I worked with an AI company selling to banks. They priced based on a matrix combining users and models in production. During our discussions, the banks pushed back, saying: "Models make sense to us, but users? We don't care about users."

This was fascinating because both metrics correlated to value, but models in production sat much closer to the outcome the banks actually cared about. The users were just an input—necessary, but not directly tied to value creation.

What I began to understand was that each step in this value chain represents a potential pricing metric. The closer your metric sits to the end of the chain—to the ultimate business outcome—the more "outcome-based" it truly is.

For the banking AI company, we eventually shifted their pricing to focus primarily on models in production, with tiering based on the complexity and business impact of those models. This moved them further along the value chain, closer to true outcomes, without reaching the very end (which might have been impossible to measure reliably).

The Reality of Outcome-Based Pricing Today

What most companies call "outcome-based pricing" today isn't actually at the end of the value chain. It's just closer to the end than traditional input metrics like user seats or subscription tiers.

And that's okay. What matters isn't purity but improvement — moving along the spectrum toward metrics that more closely align with customer value.

I've watched vertical SaaS companies make this transition more successfully than horizontal ones. When you're building specifically for dentists or real estate agents, you can tailor your pricing to their specific outcomes. Horizontal solutions serving multiple industries face a greater challenge because the value chain looks different for each customer segment.

What I encourage my clients to do isn't to obsess over achieving "perfect" outcome-based pricing. Instead, I suggest mapping their customers' value chains and identifying metrics that move one or two steps closer to the outcomes that truly matter.

The companies that do this well don't just move along the value chain—they truly understand and maximize the outcomes their customers care about. That's the real promise of outcome-based pricing: not a binary choice, but a continuous pursuit of closer alignment between what customers pay and the value they receive.

---

Thanks for tuning in and see you next week!

Have thoughts on this post? I’d love to hear them. Hit reply or drop a comment.