Unlocking Customer-Level Margins and Set Value-Based AI Pricing

Before Attribute, the pricing and finance teams at the company operated with a fundamental blind spot: they knew what customers were paying, but they had no way to see what it actually cost to serve those customers.

Their internal data warehouse was rich on the revenue side, seats, tiers, add-ons, BI, but completely dark on the cost side at the customer level. That made three critical conversations impossible to have with any rigor:

1. Which customers are unprofitable, and why?
2. What should a new AI product cost, and how do we protect margin as usage scales?
3. Is our enterprise pricing tier correctly priced for the workloads it actually generates?

Their pricing leader summarized it this way: “We have very clear pricing models, but we’ve never actually understood the customer context behind them.” Without that context, every pricing decision was effectively a simulation built on assumptions, not the real consumption data coming out of production.

Attribute’s runtime cloud cost intelligence plugged directly into the company’s environment, no tagging project, no instrumentation requests to engineering. Within weeks, Attribute was attributing cloud spend down to the customer level and streaming it into the customer’s existing BI environment alongside their revenue data.

For the first time, the team could answer questions like:

• What does it actually cost to serve our largest enterprise accounts?
• Which customers on which tier are driving the most cloud consumption?
• Are our highest-revenue accounts also our highest-margin accounts?

Attribute connected cost-per-customer to the existing revenue data model, and the results were immediate and surprising.

Looking at roughly 7,500 of the most expensive accounts, the biggest cost drivers running inside their infrastructure, Attribute surfaced a set of numbers the business had never seen before:

• ~360 accounts were unprofitable, COGS exceeded revenue
• ~$1.3M in aggregate losses across those accounts
• A clear gap inside the Enterprise Standard tier, customer usage patterns didn’t match what that tier had been priced to deliver

As one team member put it: “We never had customer context. We saw total deal value and application-level data, but never how the customer actually uses the system.”

That last finding, the enterprise tier gap, was a total miss. The team could see specific accounts where the blended consumption of dashboards, AI assistants, per-user add-ons, and complex tier math was producing economics the pricing model had never accounted for.

With customer-level cost context in place, the pricing team turned to the conversation every SaaS company is having in 2026: how do we price AI features so they don’t destroy margin?

Most pricing teams today are guessing. They run simulations, build hypothetical consumption models, and ship pricing into the market without a clean read on what AI features actually cost to deliver per customer. As one leader at the company put it, “It’s changing now, it’s even more implicit. This is nonsense.”

Attribute closed that gap. By tracking token consumption by topic and attributing it to specific workloads, features, and customers, Attribute gave the pricing team a real equation between:

• The value action the customer is taking (a note-taking AI, a sidekick, an AI block)
• The consumption footprint that action drives (tokens, inference calls, supporting infrastructure)
• The resulting unit economics the pricing model needs to protect

Instead of pricing AI products from a PowerPoint simulation, the team can now price them from actual production data, the same data that shaped the new pricing models for their core AI assistant and the incoming AI sidekick add-on.

For a work management platform with a growing AI portfolio and increasingly complex pricing, customer context isn’t a nice-to-have, it’s the foundation every margin decision now runs on. Attribute delivers what no other cost tool can:

• Customer-level profitability for every account, every tier
• Token consumption attribution for AI products, enabling true value-based pricing
• Workload-level cost context without a tagging project

Pricing, finance, and product teams can stop guessing and start making margin decisions from live production data.

• ~360 unprofitable accounts identified, previously invisible
• ~$1.3M in negative-margin revenue surfaced and routed to pricing for action
• Enterprise Standard tier gap identified, feeding a pricing model redesign
• AI pricing grounded in production data, not simulations, a first for the pricing org
• Zero tagging work required from engineering to produce any of it