Per-Action Pricing: Examples & Companies

What is it

Per-Action Pricing is a billing unit where each discrete action taken by an AI agent or automation is metered — common in browser automation and agentic workflow tools. Unlike tokens (which meter compute) or seats (which meter headcount), the action unit tries to meter work done: a browser step completed, a data row enriched, a meeting brief generated, or a 15-minute block of autonomous coding.

The unit emerged as automation platforms needed a billing granule that non-technical buyers could understand without knowing anything about the underlying LLM calls or compute cost. Clay is the canonical example, splitting its pricing into platform Actions (orchestration, always 1 per enrichment and under $0.01 each) and Data Credits (the marketplace cost of the data, roughly 0.5–10+ credits per record), so buyers can reason about what it costs to run a workflow independently from what it costs to buy the data. Bardeen applied a similar idea to browser automation: credits map to output rows (standard rows at 1 credit each, enrichment rows at 3 credits), so the bill tracks the lead list delivered rather than the compute consumed.

What makes this category challenging is the wild variance in what “one action” means across vendors — from a browser output row up to a 15-minute autonomous task. UiPath AI shows how deep the sub-unit can nest: its Intelligent Xtraction & Processing engine meters actions in Platform Units “at a rate of 0.2 Platform Units per chargeable action,” a granule buried inside a broader per-seat and per-robot licensing structure. That divergence makes the action the most definitionally contested billing primitive in this corpus.

How it works

The mechanical question is always: where does the action boundary sit? Vendors answer at several different levels of abstraction.

The pricing formula varies by vendor, but the common structure is:

Bill = (action pool or tier fee) + (actions consumed × per-action rate)

For Clay, the two-meter version is: Bill = Actions capacity tier + (Data Credits consumed, drawn from a rollover pool). Actions are a fixed capacity that resets each cycle; Data Credits are the variable pool. For Bardeen, the flat-fee version is: Bill = plan fee, with bundled credits consumed until the pool runs out and no rollover — unused credits simply expire. For Rox, the base fee buys a monthly Agent Action pool that resets each cycle, and Core users can “early-renew” to reset the allotment mid-cycle.

Worked numeric example — Clay, a mid-size GTM team. A Growth team at the $495/month base tier starts with 40,000 Actions and 6,000 Data Credits per month. A workflow that enriches 500 prospects — each requiring 1 Action (500 Actions total) plus roughly 12 Data Credits for phone and firmographic data (6,000 Data Credits total) — exhausts the entire included Data Credit pool in a single run, while barely denting the 40,000-Action allowance. The Data Credits are the bill driver; the Actions cost pennies. You can model the two meters against your own workflow volume with the Clay pricing calculator.

Worked numeric example — Bardeen, a lead-enrichment workflow. Bardeen’s own pricing-page example — scrape 10 profile links, extract contacts, verify emails with enrichment steps (3 credits each), and qualify leads with AI (1 credit each) — burns about 64 credits in one run. The $50/month Premium plan bundles 1,000 credits/month, which supports roughly 15 runs of that workflow before the pool empties; the $10/month Basic plan and the free tier both grant only 100 credits/month, so a single 64-credit run consumes more than half of it. Unused credits expire monthly with no rollover.

Companies using this

Seven in-corpus companies bill in actions — Clay and Bardeen as the canonical GTM/automation examples, plus Cognition (Devin), Rox, Automation Anywhere, UiPath AI, and Writesonic, each at a different granularity. The table below sorts by pricing model, billing units, and free-tier availability.

Patterns observed

Actions are a wrapper around cost opacity, not a replacement for it. Across all seven companies, the action abstracts away the actual compute and data costs the buyer never sees directly. Clay makes this unusually explicit by separating near-zero orchestration Actions from Data Credits (the real variable cost of buying data and AI from 150+ marketplace providers). Most vendors collapse both into one unit — buying simplicity at the cost of transparency. When Rox says an Agent Action’s cost “scales with task complexity and AI model used,” it is describing a floating per-action rate that can’t be quoted as a single headline number.

The action-boundary decision is the biggest pricing lever, and RPA vendors bolt it onto an older robot frame. Vendors who define “one action” at a low altitude (Bardeen’s output row) get a legible meter buyers can count; vendors who define it high (Cognition’s 15-minute ACU) get a meter that more faithfully represents business value but is harder to model up front — which is why Cognition keeps the ACU as its enterprise unit while wrapping self-serve buyers in a quota-plus-on-demand-credit structure. Meanwhile Automation Anywhere and UiPath AI both list “actions” as a billing unit, but only as a consumption sub-unit layered on top of their older attended/unattended bot and per-seat licensing — agentic work gets metered in actions, but the headline bill still runs through the bot scaffold.

Feature gating is giving way to action gating. Rox shows the shift cleanly: its November 2024 beta gated features by tier (a traditional SaaS approach) and priced Core at $20/month with per-tier account caps, and by 2026 Core was $50/month with all feature gates and account caps removed — making Agent Action volume the only upgrade lever. Buyers now upgrade purely because they want more agent work done, which is the action model operating at its purest.

Counterexamples & variants

Writesonic is the closest thing to a false positive in this cohort. Its billing_units field includes actions, but in practice actions are a minor footnote — Growth-plan ($399/mo) buyers get an Action Center trial (5 off-page + 5 on-page actions per month) and a small agentic-workflow trial allowance (10 → 50 → 100 runs by tier). The core value metric at Writesonic is “answers tracked daily” across AI search engines (50 → 300 → 600/day), not actions per se. Listing actions as a billing unit is technically accurate for the small slice of the product that uses them, but they don’t drive the bill the way they do at Clay, Bardeen, or Rox — buyers should weight the daily-answer tracking quota as the real consumption dimension.

Rollover mechanics reveal how confident each vendor is in its action definition, and the attended/unattended split is action-pricing in disguise. Clay’s Data Credits roll over up to 2× the monthly limit on Launch and Growth, acknowledging that enrichment work is bursty — whereas Rox’s Agent Actions and Bardeen’s credits both expire at cycle reset with no rollover, a “use it or lose it” dynamic that compresses burn and accelerates renewal. A distinct variant appears at Automation Anywhere, which explicitly prices unattended bots (fully autonomous) higher than attended (human-supervised) ones: the same action carries a higher price tag when it runs without oversight, because the vendor absorbs all the reliability risk. No other vendor in this cohort makes that distinction explicit on the price card.

What this means for buyers vs vendors

For buyers

Benchmark your specific workflows before comparing prices, not the headline action rate. A “1 credit per action” plan from Bardeen and a “1 Action per enrichment” plan from Clay can look identical on the surface yet diverge sharply once you account for how each vendor counts a multi-step enrichment — Bardeen charges 3 credits for an enrichment row, and Clay’s real cost lives in the separate Data Credit meter. The choosing the right usage metric guide recommends running your three highest-volume workflows through each vendor’s counting rules as a literal first step in evaluation. Then watch the rollover policy: it tells you whether you’ll be forced to burn actions on low-value tasks at month-end or can bank capacity for campaign bursts, and read the free tier carefully — at Rox, a full-featured Starter plan with 2,000 actions/month is a genuine product trial because all features are included, whereas at other vendors the free tier is a crippled demo.

For vendors

The action is the most defensible unit in agentic AI pricing when you can publish the counting rule clearly — but it demands a worked example on the pricing page that lets buyers model their own workflows, the way Bardeen’s scrape-extract-verify-qualify walkthrough and Clay’s two-meter split both do. Separating orchestration cost from data/model cost puts the variable spend in its own rollover pool and solves the floating-cost problem — techniques covered in the usage-based pricing fundamentals guide and the tracking and metering usage events guide. If you can’t write that example because your per-action cost floats with too many variables, your unit may actually be a capacity quota. And if your agents act continuously — monitoring, researching, alerting in the background — budget controls that cap daily action burn before the pool empties are table stakes; without them, always-on agents can consume a month’s allotment in a long weekend.

Explore this theme in the knowledge graph →