Sierra Pricing Explained: How Outcome-Based Contracts Work

Sierra charges per outcome, an agreed fee each time its agent completes a task you define, such as a resolved support conversation. Unresolved and escalated conversations are usually free, and the rate is set in your contract. Dograh, the open-source AI voice agent platform we build, bills by the minute on Dograh Cloud, or you can self-host it.

Key Takeaways

• Sierra bills per completed task, with each outcome's criteria agreed upfront.
• Unresolved and escalated conversations carry no charge in most cases.
• Routing and greeter calls can be priced per conversation instead.

Paying only when the agent resolves something sounds simple, until you ask who decides what resolved means. We cover how Sierra's outcome pricing works, what is usually free, buying through Google Cloud Marketplace, what to settle before signing and how it compares with Dograh's per-minute pricing. It is part of our guide to Sierra alternatives.

How Sierra's outcome pricing works

Sierra is paid only when its agent completes a task you have agreed on. Sierra's post on outcome-based pricing names the kinds of outcome it means: a resolved support conversation, a saved cancellation, an upsell or a cross-sell. Clear criteria for each outcome are agreed upfront, so both sides know what triggers a charge before the agent takes its first call.

The fee per outcome is a contract term. Your rate depends on the outcomes you choose, how they are defined and the volume you expect, and it comes out of the sales conversation with Sierra.

Sierra keeps the model costs on its side of the line. When it launched Horizon, its platform for agents that follow a customer goal over days or weeks, it wrote that you pay for business outcomes delivered rather than tokens, and that it bears the burden of managing token spend. For a buyer, that means the price does not move when Sierra swaps or upgrades the models behind your agent.

Sierra pairs this with a large customer base. In its latest growth update, it reports more than $400 million in annual recurring revenue and says it works with almost half of the Fortune 50.

What is usually free, and what isn't

A conversation the agent cannot finish usually costs nothing. Sierra says that if a conversation is unresolved, in most cases there is no charge, and the same goes for a case that has to be escalated to a person. That is the heart of the model: you pay when the work gets done.

The words "in most cases" matter, so read your contract for the exceptions. Some calls never aim for an outcome. Sierra notes that routing or greeter-style interactions, such as a call that only finds the right department, may fit consumption-based pricing better, paid by conversation count whatever the outcome. If a share of your line is routing, part of your bill will follow call volume.

Buying Sierra through Google Cloud Marketplace

Sierra is also sold on Google Cloud Marketplace, which changes how the money moves more than how much. Sierra's Marketplace announcement says you contract through your existing Google Cloud relationship, and that qualifying purchases count toward your Google Cloud commitment.

For a company with committed spend on Google Cloud, that can turn an agent purchase into budget it has already promised to use. Sierra's charges then appear alongside the rest of your spend in Google Cloud Console, and you still pay for the outcomes your agents deliver. Ask your Google Cloud account team which Sierra purchases qualify before you count on it.

Open Source Alternative to Vapi / Retell

Self-hosted voice agent platform — no per-minute fees

dograh-hq/dograh

Star on GitHub

Questions to settle before you sign

Outcome pricing is only as clear as its definitions, so pin these down in writing.

Start with what counts as resolved. A customer who stops replying may not be a happy customer, so ask how Sierra tells a real resolution from a dropped conversation. Then ask what counts when the agent starts the job and a person finishes it, and whether that call is free, billed or split.

Next, ask which calls fall under per-conversation pricing, and how a call gets classed as routing. Ask whether build and tuning work by Sierra's agent development team, which has supported hundreds of deployments, is priced separately or inside the outcome rate. Finally, ask about minimum volumes and how rates change as volume grows.

Our voice AI platform buyer's checklist turns questions like these into RFP lines. PolyAI is the other sales-priced vendor buyers often weigh, and our breakdown of PolyAI pricing covers how its quotes work.

Outcome pricing vs per-minute pricing

Outcome pricing follows resolutions, and minute pricing follows call volume. Neither unit is better on its own, so work out both on your own numbers.

Fin shows that a per-outcome price can be published. Fin's pricing lists $0.99 per outcome, while Fin Voice is offered on a custom plan.

Here is illustrative arithmetic with your own figures in place of ours. Say your line takes 10,000 calls a month, the agent resolves 6,000 and the average call runs four minutes. Under outcome pricing, the bill is 6,000 times your outcome rate, plus any per-conversation routing calls. Under minute pricing, it is 40,000 minutes times the per-minute total, resolved or not. The lower your resolution rate, the more outcome pricing protects you from paying for calls that fail.

Per-minute platforms are simpler to forecast, because call volume is easier to predict than resolutions. Run Dograh on your own servers and the forecast holds three lines with no per-minute platform fee among them: servers, model keys and phone minutes. On Dograh Cloud, a 1¢-a-minute platform fee joins the model and phone costs. Our post on AI voice agent cost per minute breaks those costs down, and our row-by-row comparison of the two products covers everything beyond price.