Clay didn’t “change pricing.” Clay changed incentives.
On March 11, 2026, Clay published an internal pricing memo explaining why the old credit model was mispriced, why heavy users were costing Clay money, and why the new model splits what you buy into two meters: Data Credits (data) and Actions (work). (clay.com)
If you run a lead gen agency selling “GTM Ops” by the credit, this memo is your warning label.
TL;DR
- Clay pricing changes 2026 means two meters: Data Credits for marketplace data, Actions for orchestration (enrichments, AI tasks, HTTP calls, pushes to tools). (clay.com)
- Marketplace data got cheaper (Clay claims 50-90% cuts), but workflow-heavy teams now pay for the “plumbing.” (clay.com)
- The trap for agencies: you sell “per lead” or “per contact,” but your cost becomes “per step.” That gap eats margin.
- If your Clay tables look like a conspiracy board, Actions turns “cool workflow” into a recurring tax.
- The fix: tight ICP, fewer enrichment steps, fewer retries, fewer “nice-to-have” signals. Or stop buying a Frankenstack.
What Clay actually changed (March 11, 2026)
Clay’s memo lays it out in plain English:
- They mispriced credits in the Pro segment and operated at a loss for years. (clay.com)
- Fixed AI pricing broke once reasoning models made costs wildly variable. (clay.com)
- They split pricing into Data Credits + Actions to charge for platform work, not just data. (clay.com)
- They cut marketplace data prices 50-90% to make Clay’s marketplace closer to external costs. (clay.com)
Clay’s official FAQ makes the subtext explicit: if you used Clay mainly for orchestration with your own API keys, the old model could be “effectively free,” and now it is not. (clay.com)
This is the market maturing in public. Data got expensive. API workloads got real. Pricing models stopped pretending.
Define the meters like an adult
Let’s stop pretending “credits” are intuitive.
Data Credits
Data Credits = buying data from Clay’s marketplace partners (emails, phones, firmographics, intent, etc.). Clay says these got much cheaper under the new model. (clay.com)
Actions
Actions = paying Clay to run the machine.
Per Clay’s docs and memo language, Actions cover things like:
- enrichments (the step itself)
- AI tasks
- HTTP API calls
- pushes/syncs to third-party platforms (university.clay.com)
So yes, you can BYOK for data and still pay Clay for the orchestration layer. That’s the point.
Clay pricing changes 2026: the buyer math agencies actually need
Agencies do not buy “data.” Agencies buy margin.
You sell one of these:
- per lead
- per meeting
- per “qualified account”
- retainer with volume commitments
Clay charges you:
- per record (Data Credits)
- per step (Actions)
Those don’t map cleanly. That’s where surprises live.
The new hidden unit: cost per workflow, not cost per lead
A “lead” in an agency pipeline is usually a bundle:
- find account
- find contacts
- enrich company
- enrich contact
- verify email
- sometimes phone
- score fit
- score intent
- generate personalization
- push to sequencer/CRM
- route replies
In Clay, every extra step means:
- more Actions
- more points of failure
- more retries
- more “debug time,” which is the most expensive line item of all
Clay even says plans will “default with 4-5 actions per data credit.” (clay.com)
That’s a clue: power users burn Actions because power users build step-heavy workflows.
If you are a lead gen agency, you are a power user by definition.
Three traps that will hit agencies first
Trap #1: Unpredictable unit costs (aka your margin starts drifting)
When your pricing model is “credits,” your unit cost depends on:
- which providers you call
- how many calls you chain
- whether you re-run steps on missing data
- whether you run AI research on every record
- how much your “quality gates” filter out after you already paid
Third-party breakdowns of Clay’s new plans show how wide the cost range can get depending on enrichment depth. Example: Cleanlist’s analysis estimates full contact enrichment can vary significantly based on provider chain, even when you assume a base data-credit rate. (cleanlist.ai)
You cannot run an agency on “it depends.”
Agency takeaway: if you cannot quote a stable cost per delivered lead, credits pricing turns into margin erosion with better branding.
Trap #2: Workflow sprawl (Actions turns “nice ideas” into monthly fees)
The Clay superpower is also the Clay tax: you can keep adding steps forever.
- Add a tech install check.
- Add funding enrichment.
- Add “recent hires.”
- Add “AI first line” from website copy.
- Add another waterfall.
- Add another verification.
- Add another CRM push.
- Add another webhook to your internal system.
Congrats. You built a Rube Goldberg machine that looks productive and invoices you every month.
And when your client asks why performance dipped, you are debugging tables instead of shipping meetings.
Clay’s own FAQ explicitly frames Actions as tying platform cost to platform value. (clay.com)
That’s rational for Clay. It’s brutal for agencies selling outcomes.
Trap #3: Fragile enrichment chains (one broken provider breaks the “lead”)
Agency pipelines tend to rot in the same way:
- You add one more provider.
- That provider changes an API response.
- Your formula breaks.
- Your fallback triggers.
- Your fallback burns more Actions.
- Your “QA step” catches it late.
Now your cost per lead rises and your throughput drops.
You do not have a data workflow. You have a dependency graph.
Decision framework: when credits pricing makes sense vs when it becomes a tax
When Clay’s new pricing makes sense
Clay is still Clay. If you need orchestration, it’s a monster.
Credits pricing works when:
- Your workflow is stable
You run the same 3-6 steps. You do not improvise in production. - You monetize complexity
You charge for research-heavy outputs, not commodity leads. - You actually use marketplace data
Clay cut marketplace data prices 50-90%, so you benefit if you buy marketplace data instead of BYOK everything. (clay.com) - You have a hard cap on steps per record
If every record can only take N steps, your unit cost stops drifting.
Example use case that fits:
ABM campaigns where every account is high value, volume is low, and the workflow generates very specific personalization.
When it turns into a tax
Credits pricing becomes a tax when:
- You BYOK most data
The pricing memo and FAQ basically say you used to get orchestration “free,” and now you do not. (clay.com) - You sell at volume
Bulk lead lists. Weekly contact drops. “Just keep the top of funnel full.” - You run enrichment waterfalls by default
Waterfalls are great, until you do them on everyone. - You treat Clay like a backend
Heavy HTTP, heavy automations, heavy CRM sync. Actions meter punishes “Clay as infrastructure.”
If you are doing 5 figure monthly volumes, any pricing model that bills “per step” will find you.
The agency playbook to cap spend (without gutting performance)
1) Start with stricter ICP filters, not more enrichment
Most agencies do the opposite:
- they enrich first
- then they filter
That’s backwards. Filter before you pay.
Run a two-stage ICP:
- Stage A: cheap filters (domain, industry, headcount, geo, basic tech)
- Stage B: paid filters (titles, direct dials, hiring signals, intent)
If a company fails Stage A, it never gets Stage B.
This is the whole logic behind fit + intent gating. Gate spend before you burn it. Chronic covers this with dual scoring and suppression logic here: deliverability qualification gate for 2026.
If you want the implementation, build your scoring rules once, then keep the signal set small. Chronic’s native scoring approach lives here: AI lead scoring.
2) Stop enriching “for completeness”
Agencies love complete records. Clients love meetings.
Define the minimum viable record for each channel:
Email-only outbound
- first name
- company
- role/title (rough is fine)
- verified email
Cold call
- direct dial (or switchboard plus routing)
- role confirmation
- profile URL
- role
- company
Everything else is optional until it proves ROI.
3) Cut enrichment steps with a “one provider, one fallback” rule
Waterfalls are where budgets go to die.
Set a rule:
- Pick one primary provider for a data type
- Pick one fallback
- Stop
If you need three fallbacks, your list quality is the issue. Not your workflow.
Related: you will like this Chronic post if you keep blaming personalization for bad targeting: Personalization vs list quality.
4) Separate “R&D workflows” from “production workflows”
Do not ship experiments to the same tables that feed client campaigns.
- R&D table: messy, creative, expensive
- Production table: boring, deterministic, cheap
If Actions now meter the plumbing, your production workflows should feel painfully simple.
5) Price your service with a data cost line item
If you still sell “per lead,” you are volunteering to eat credit volatility.
Do this instead:
- Retainer covers strategy + ops + reporting + baseline volume
- Data and enrichment billed at cost + margin
- Clear caps and thresholds
Your client may complain. Then they will remember they hired an agency, not a charity.
What this means for “GTM Ops by the credit”
Buying GTM Ops “by the credit” used to sound modern.
Now it sounds like “we do not know our unit economics.”
Because you are not buying outputs. You are buying consumption.
The market is correcting:
- enrichment is not free
- orchestration is not free
- reasoning models are not predictable
- heavy API users will get metered
Clay is just saying the quiet part out loud. (clay.com)
The obvious counter-move: consolidate the Frankenstack
A lot of agencies built stacks like this:
Clay + email finder + verifier + sequencer + CRM + intent + spreadsheets + Zapier + whatever broke last week.
Then they wonder why their costs climb while outcomes stay flat.
If your goal is meetings, you do not need five tools arguing over who owns the record.
Chronic’s stance is simple:
- $99
- unlimited seats
- end-to-end outbound, till the meeting is booked
- no Frankenstack required
Chronic handles the core chain inside one system:
- Build and enforce ICP: ICP Builder
- Enrich leads: Lead enrichment
- Score leads on fit + intent: AI lead scoring
- Write outbound that ships: AI email writer
- Track the actual pipeline: Sales pipeline
If you want the philosophy behind consolidating tools in 2026, read: The 2026 sales stack cleanup.
One clean pipeline beats ten clever workflows.
Practical scenarios: choose your model in 60 seconds
Scenario A: Boutique agency, 20 accounts, heavy research
- Low volume
- High margin per account
- Custom personalization
- Multi-step enrichment
Clay still makes sense. Pay for orchestration because you bill for it.
Scenario B: Volume lead gen, weekly drops, clients want “more contacts”
- High volume
- Low tolerance for variance
- You sell “per lead” economics
Credits pricing becomes a tax fast. You need predictable unit costs, or you need to consolidate and cap spend with gates.
Scenario C: Internal outbound team with BYOK data stack
- You already pay for Apollo, ZoomInfo, Clearbit, or custom data
- You used Clay as the glue
Under Clay pricing changes 2026, you now pay for the glue. (clay.com)
Decide if the glue is worth a new recurring meter.
FAQ
What are Clay pricing changes 2026 in one sentence?
Clay split the old single credit model into Data Credits (marketplace data) and Actions (platform orchestration like enrichments, AI tasks, HTTP calls, and pushes), while cutting marketplace data prices and reworking plan tiers. (clay.com)
Why should agencies care more than in-house teams?
Agencies sell outcomes with fixed pricing. Clay charges inputs that vary with workflow complexity. That mismatch turns into margin drift unless you cap steps and gate enrichment.
Does the 50-90% data price cut mean Clay got cheaper?
Only if you buy a lot of data through Clay’s marketplace. If you mostly BYOK and use Clay for orchestration, the new Actions meter can raise your effective cost per record. (clay.com)
What is the biggest cost trap with Actions?
Workflow sprawl. Every extra step feels small. At scale it becomes a permanent bill for “nice ideas” that do not book meetings.
How do agencies cap Clay spend without killing quality?
Use stricter ICP filters first, limit waterfalls to one fallback, and separate R&D workflows from production. Spend Data Credits only after a record passes cheap filters.
When should an agency ditch credits pricing entirely?
When you sell high volume deliverables with thin margins, or when your pipeline depends on long enrichment chains that break often. Predictable unit economics beats flexible workflows.
Make the call, then make it boring
If you love building workflows, Clay will keep rewarding your curiosity and billing you for it.
If you love booked meetings, make the system boring:
- Gate early.
- Enrich less.
- Ship more.
- Consolidate the stack.
Pipeline on autopilot beats GTM Ops by the credit.