Clay’s New Pricing Model: Data vs Actions vs Outcomes (And What Small Teams Should Actually Buy)

Clay didn’t “raise prices.” Clay changed the unit of pain.

Before, most teams budgeted Clay like a tool. Now you budget it like cloud spend: two meters running at once, with different failure modes. Data costs money. Actions cost money. Outcomes still cost you if your workflow burns credits and books nothing.

TL;DR

• The Clay pricing actions model splits spend into Data Credits (buying data and some AI) and Actions (platform usage like enrichments, exports, syncs, AI runs).
• This model is spreading across RevOps because vendors finally found a way to charge for “work,” not seats.
• Usage-based can be fair. It’s also the easiest way to accidentally fund someone else’s growth plan.
• Small teams should buy the model that matches their bottleneck: data volume, workflow volume, or meetings booked.

What changed in Clay’s pricing: data vs actions vs outcomes

Clay announced a new pricing model on March 11, 2026. The headline: cheaper marketplace data, advanced features moved down-tier, optional migration for existing customers. (community.clay.com)

The real headline: Clay separated the bill into two separate meters.

1) Data Credits: you pay for the stuff you pull

Clay calls these “data marketplace purchases.” Data credits are what you spend when you buy enrichment data from vendors inside Clay’s marketplace. (university.clay.com)

Clay says marketplace data costs dropped 50% to 90% under the new model. (community.clay.com)

That’s the “data” side.

2) Actions: you pay for the work the platform does

Actions are “platform usage capacity.” Clay charges actions when you run enrichments, run AI, push to CRMs, export to sequencers, and more. (university.clay.com)

Key detail from Clay’s docs: “Cost per enrichment: Always 1 action.” (university.clay.com)

So even if the enrichment data is free, cheap, or BYOK, the platform work still ticks the meter.

3) Outcomes: what you actually wanted

Clay doesn’t bill on outcomes. You do.

Outcome pricing is still rare across SaaS. One survey from Price Intelligently (SBI) found 2 out of 321 companies using outcomes-based pricing (under 1%). (sbigrowth.com)

So the market compromise looks like this:

• Not seats.
• Not pure outcomes.
• Actions.

Because actions are measurable. Outcomes are messy.

The plain-English definition of the Clay pricing actions model

If you need a one-liner for buyers:

Clay pricing actions model = you pay once for data, and you also pay every time Clay touches a row.

Clay’s own breakdown:

• Actions = “platform usage: enrichment & GTM execution”
• Data credits = “data & AI from vendors in Clay’s marketplace” (clay.com)

And in the docs, Clay gets even more specific: actions are consumed for enrichments, AI usage, signals, exports to sequencers, and CRM exports and syncs. (university.clay.com)

Why this pattern is spreading across RevOps tooling

RevOps tools used to sell seats because it was simple. Then CFOs started bullying everyone with “prove ROI.” Then AI showed up and the tools started doing actual work.

So vendors needed a new pricing unit:

• Seats don’t map to value when one operator can run an AI-driven workflow that touches 50,000 records.
• Outcomes are hard to attribute cleanly (and risky for the vendor).
• Actions sit in the middle. Billable. Trackable. Defensible.

Even pricing analysts are calling out that pricing complexity is exploding across SaaS. (arxiv.org)

And buyer sentiment is moving toward usage-based and shorter-term models, especially as AI becomes “always included.” (4099946.fs1.hubspotusercontent-na1.net)

Translation: more “meters.” More math. More chances to get surprised.

The buyer checklist: when usage-based pricing is fair vs when it quietly explodes

Usage-based pricing is fine when the meter matches a cost driver you control.

It’s a trap when the meter is tied to:

• retries
• fallbacks
• “oops, run the table again”
• background syncs
• AI steps you can’t predict

Here’s the checklist small teams should run before committing.

1) Define an “action” like your budget depends on it (because it does)

Clay defines actions broadly: enrichments, AI runs, exports, CRM syncs, signals, sequencing adds. (university.clay.com)

Questions to ask any vendor using an actions model:

• What counts as an action?
• Is it per record or per step?
• Does a “waterfall” cost 1 action or multiple actions?
• If the action fails, do I still pay?
• Do dry runs count?

If the salesperson can’t answer in one minute, you’re buying mystery meat.

2) Rate limits: where pipelines go to die quietly

Ask:

• What are the per-minute and per-day rate limits?
• What happens when I hit them?
• Does the system queue, fail, or retry?
• Do retries count as new actions?

Rate limits matter more than unit cost. They determine whether “usage-based” becomes “usage-stalled.”

3) Retries: the silent budget killer

Every workflow eventually hits:

• provider timeouts
• temporary blocks
• flaky endpoints
• CAPTCHA pages
• “no result, try another provider”

Ask:

• Are retries automatic?
• Can I cap retries per record?
• Can I set “stop conditions” like “stop after email found”?
• Do retries consume actions?

This is where “cheap data” turns into expensive automation.

4) Enrichment fallbacks: who pays when the first source fails?

Clay supports multi-provider waterfalls and BYOK. (clay.com)

That’s powerful. It’s also where costs multiply if you do it wrong.

Ask:

• If Provider A returns nothing, and Provider B runs, is that one action or two?
• Can I set fallback rules by persona, region, or domain type?
• Can I prevent “phone lookup” unless the lead hits an intent threshold?

5) Rollover rules: do your unused credits vanish?

Clay’s docs show a split:

• Actions rollover: No
• Data credits rollover: Yes (up to 2x monthly limit) (university.clay.com)

That changes how you should buy:

• If actions do not roll, overbuying punishes you.
• If data rolls, you can stockpile for campaigns.

Always ask rollover rules. Always.

6) BYOK: “free data” isn’t free if the platform still charges actions

Clay explicitly says you can connect your own API keys for enrichment tools. (clay.com)

But Clay’s docs also say using your own API keys still consumes Actions. (university.clay.com)

So BYOK can reduce data credit spend, while action spend stays.

That’s not bad. It’s just not the “get out of jail free” card people think it is.

The real-world math: how small teams get surprised

Clay’s pitch is straightforward: cheaper data, clearer pricing, advanced features on the Growth plan. (community.clay.com)

The surprise comes from workflow shape.

If you run “one enrichment then export,” you’re fine

Simple flow:

1. import list
2. find email
3. export to sequencer

Predictable actions. Predictable data.

If you run “research + personalize + sync + sequence,” actions stack fast

Common “modern outbound” flow:

1. enrich company
2. enrich person
3. run AI research per account
4. generate personalization snippet
5. push to HubSpot
6. export to sequencer

Clay’s docs say enrichments, AI runs, CRM exports, and sequencer exports consume actions. (university.clay.com)

So your action count can exceed your “lead count” quickly.

That’s the point of the model. You are paying for the work.

Three team archetypes: what pricing model fits, and what to buy

You asked for small teams. Small teams have one rule: cash is oxygen. Anything that makes costs unpredictable gets scrutinized.

Here are the three archetypes and what to buy.

Agency teams: “We run 6 clients. Everyone wants custom.”

Reality

• Many workflows.
• Lots of iterations.
• Constant table re-runs.
• Client churn means spend swings.

Best fit

• Actions-based can work if you control process and templates.
• It can also explode because agencies love “just one more step.”

What to buy

• Buy the minimum tier that supports the integrations you need.
• Build 1 to 2 “golden workflows,” then clone.
• Gate expensive steps behind rules:
  • only run AI research for Tier 1 accounts
  • only run phone lookups after a reply signal

Agency checklist

• Does the vendor support per-client workspaces or cost reporting by project?
• Can you cap actions per table run?
• Can you prevent junior ops from re-running the whole workflow?

B2B SaaS outbound team: “We need 20 meetings a month, not a data science project.”

Reality

• You care about speed to pipeline.
• You run sequences weekly.
• You need enrichment, scoring, writing, and booking to behave like one system.

Best fit

• A hybrid model is fine.
• But pure actions pricing makes forecasting annoying unless your process is locked.

What to buy

• If you use Clay, keep it for what it’s best at: orchestration and experimentation.
• Do not turn it into your entire outbound stack unless you accept variable spend.

What small SaaS teams should prioritize instead

• Lead scoring that reflects reality: fit + intent, not vibes.
• Enrichment that does not force you to play provider roulette daily.
• Email writing that ships fast, with guardrails.
• A pipeline view that shows what is blocked today.

(Yes, this is where an autonomous system wins, because it removes the “tooling math.”)

Founder-led sales: “I will not learn another credits system.”

Reality

• Low volume.
• High stakes per lead.
• Founder time is the most expensive input.

Best fit

• Flat pricing. Outcome focus. Minimal config.
• Usage-based is fine only if it’s truly tiny and predictable.

What to buy

• If you are founder-led, avoid paying for “actions” that represent fiddling.
• Buy something that runs end-to-end, till the meeting is booked.

Because the founder shouldn’t debug a waterfall enrichment at 11:30 PM.

What to ask before you commit (copy-paste for procurement)

Send this list to any vendor with usage-based pricing. If they dodge, you learned everything you need.

1. Define an “action” with examples. List what is free.
2. Do failed actions bill?
3. Do retries bill? Can we cap retries?
4. What are the rate limits? What happens at the limit?
5. Do background syncs bill as actions?
6. What rolls over? What expires?
7. Can we set hard spend limits that stop execution?
8. Can we see cost reporting by workflow, user, client, and destination?

Clay publishes a lot of this publicly, including what consumes actions and how actions vs data credits work. Start there, then verify how it behaves in your exact workflow. (university.clay.com)

The uncomfortable truth: “cheaper data” can still mean higher bills

Clay says it cut marketplace data prices and moved advanced features into a lower-priced plan. (community.clay.com)

Both can be true, and you can still pay more.

Because the new model charges for:

• the data, and
• the orchestration

So the winner profile looks like:

• enrichment-heavy teams
• fewer workflow steps per record
• fewer reruns
• tight governance

The loser profile looks like:

• experimentation-heavy teams
• lots of AI steps per record
• messy CRM sync habits
• “just run it again” culture

Chronic’s pricing contrast: stop doing tooling math

Clay is a GTM engineering platform. It’s good at that. It also makes you think like a billing analyst.

Chronic takes a different stance: outcome focus, flat price, unlimited seats. One number. No credits spreadsheet.

• $99. Unlimited seats.
• Pipeline on autopilot.
• End-to-end, till the meeting is booked.

And if you want the pieces:

• Lead scoring that prioritizes who matters now: AI lead scoring
• Enrichment without juggling five vendors in a waterfall: lead enrichment
• Personalized outbound that ships fast: AI email writer
• Visibility into what’s stuck and what’s moving: sales pipeline
• Define your ICP once, stop arguing in Slack: ICP builder

If you’re evaluating AI layers seriously, read this before you buy anything you can’t control: AI CRM reliability checklist. Downtime plus usage billing is a special kind of pain.

And if you want the bigger macro shift behind this pricing change, this lays it out clean: HubSpot’s outcome-based agents and why seat pricing dies next.

FAQ

What is the “Clay pricing actions model” in one sentence?

It’s a two-meter system: Actions bill for platform work (enrich, AI, exports, syncs) and Data Credits bill for marketplace data purchases. (university.clay.com)

Does bringing your own API key make Clay free?

No. BYOK can reduce data credit spend, but Clay’s docs say using your own API keys still consumes Actions. (university.clay.com)

What counts as an “Action” in Clay?

Clay lists enrichments from any provider, AI usage, signals, GTM execution like sending emails or posting to Slack, exports to sequencers, and CRM exports and syncs. (university.clay.com)

When is usage-based pricing fair for a small RevOps team?

When the unit is predictable and under your control: stable workflow steps, limited retries, clear stop conditions, and cost reporting by workflow.

When does usage-based pricing “quietly explode”?

When your process includes lots of retries, fallbacks, AI research steps per record, frequent table reruns, and background syncs that trigger billable events.

What should a founder-led team buy instead of an actions-based platform?

A flat, outcome-focused system where cost does not rise because you ran one more workflow step. Founder time is expensive. Tooling math is a tax.

Run this play today: pick the model that matches your bottleneck

1. If your bottleneck is data coverage: buy data-first pricing, measure cost per usable contact.
2. If your bottleneck is workflow automation: actions-based can work, but cap retries and lock templates.
3. If your bottleneck is meetings booked: stop buying meters. Buy outcomes. Chronic runs outbound end-to-end, till the meeting is booked.