Clay Tables Explained: How They Work, What They Cost, and When You Need Something Simpler

Clay tables are the interface behind the whole "GTM engineering" movement. If you have seen a screenshot of a spreadsheet where one column finds emails, another scrapes a website, and a third writes a personalized first line, you have seen a Clay table. This guide explains how they actually work, what they really cost per row, and when you need one versus something simpler.

What are Clay tables?

A Clay table is a spreadsheet with superpowers. Rows are companies or people. Columns are live actions instead of static cells: pull a LinkedIn profile, look up a verified email, check a company's tech stack, or ask an AI agent a research question.

That is the core model: spreadsheet plus enrichment. You import or build a list of rows, then stack columns that fetch, compute, or generate data for every row. Each column runs across the whole table, so 1,000 rows means 1,000 executions of whatever that column does.

The result feels familiar (it looks like Google Sheets) but behaves like a workflow engine. That combination is why Clay owns the "tools like Clay" category it created, and why it can feel like overkill if all you wanted was a clean, enriched list.

How Clay tables actually work

Four mechanics explain almost everything that happens inside a Clay table.

Columns are API calls

Every enrichment column in Clay is a wrapper around one or more API calls. A "Work Email" column queries an email provider. A "Company Headcount" column queries a firmographic source. An HTTP column calls any endpoint you want. When you add a column and hit run, Clay executes that call for each row and writes the response into the cell.

This is the honest mental model: a Clay table is a visual orchestrator for API calls, priced per call. Powerful, and also the root of every billing surprise (more on that below).

Since September 15, 2026, Clay also has a public REST API that runs your table functions and searches from your own code; our Clay API guide covers what it can and cannot do.

Claygent: the AI column

Claygent is Clay's AI research agent packaged as a column. Instead of hitting a structured data provider, it browses the web and answers open-ended questions per row: "Does this company hire SDRs?", "Summarize their pricing page", "Find their compliance certifications." Our Claygent guide covers what it costs per row, the models and how to prompt it.

AI columns are the part of Clay most teams actually fall in love with. Cleanlist AI runs the same kind of per-lead research as a skill, at 3 credits a lead to research the company or the contact and 5 for both, from the same credit balance as emails and phone numbers. The Cleanlist AI vs Clay comparison has the side-by-side.

Waterfalls across providers

Clay's marketplace connects 150+ data providers (Clay now also says 200+), and waterfall columns chain them: try provider A for an email, fall back to B, then C, until one returns a result. You pick the providers, the order, and you pay data credits per successful call.

Waterfalls are genuinely the right architecture for B2B data, because no single provider covers everyone. The difference between tools is who builds and maintains the waterfall. In Clay, you do. In a managed tool like Cleanlist AI, the 25+ provider waterfall is pre-built and server-side, which is how it returned 98% verified email on the 500-Lead Enrichment Benchmark without you touching provider logic.

Tables are becoming a platform (summer 2026)

Clay is currently mid-blitz on a "12 features in 12 weeks" release campaign, shipping weekly through the summer: Sequencer 2.0, a Workflows beta, sandbox and read-only tables, agent plugins, and more, per Clay's changelog and community announcements. The direction is clear. The table is turning into a full GTM platform with sequencing and multi-step workflows layered on top.

That is exciting if you want one deeply configurable system. It also means the learning curve is compounding, not shrinking.

What Clay tables cost

Pricing is where Clay tables stop being a spreadsheet and start being a procurement exercise.

The dual credit system

On March 11, 2026, Clay replaced its old plans with two self-serve tiers and split credits into two currencies, as covered in detail by independent analyses from Michael Saruggia and Salesforge:

Data Credits buy marketplace data (emails, phones, firmographics). Actions meter platform usage (AI runs, HTTP calls, formulas). Alongside the split, Clay cut marketplace data prices by 50 to 90% depending on the provider. Existing customers were grandfathered on legacy plans, and the window to switch closed on April 10, 2026, so every new team now prices against this model. Our full breakdown is in Clay pricing changes 2026.

The per-row math

Here is the math that matters. Per Michael Saruggia's 2026 teardown of the new plans, a full contact enrichment (verified email, phone, title, company data) burns roughly 14 to 34 data credits on the Growth plan, and up to 75 when you layer on deeper company data. That works out to roughly $0.70 to $3.75 per fully enriched contact, on top of the $495/month subscription.

Run 1,000 full contacts a month at those per-row rates and your real Clay bill commonly lands between $1,300 and $2,400 all-in. For comparison, the same full contact on Cleanlist AI runs about 11 credits (roughly $0.65 per contact on Pro at $89 per seat for 1,500 credits), with no separate action metering.

Where bills surprise teams

Three patterns produce most of the "why is our Clay bill 3x the subscription" moments:

• Waterfall fan-out. A waterfall that tries four providers can charge for multiple attempts per row. Multiply by 5,000 rows and a single column run gets expensive.
• Two currencies to forecast. Data credits and actions deplete at different rates. Teams budget for one and get capped by the other, mid-campaign.
• Re-runs. Fixing a prompt or reordering a waterfall means re-running the column. Every re-run bills again across every row.

None of this is hidden. It is just variable, and variable is hard to forecast. If your finance team wants one predictable line item, the model itself is the problem.

When Clay tables are the right call

Clay is not overpriced for everyone. It crossed $100M in ARR with enterprise net revenue retention above 200% as of June 2026, per Industry Lens reporting. Enterprise customers that adopt it double their spend on average, which only happens when a product genuinely delivers.

Clay tables earn their cost when:

• Someone owns the tool. You have a GTM engineer or a technically deep ops person whose job includes building and maintaining tables.
• Your workflows are genuinely complex. Multi-source waterfalls with conditional logic, custom scrapers, intent triggers, and webhooks that fire on specific signal combinations.
• You are consolidating tooling. If one Clay build replaces three point tools and manual research, the subscription plus credits can still be net savings.

If that describes your team, read our full Clay data enrichment review for where it shines and where it strains.

When Clay tables are overkill

Most teams evaluating Clay have a simpler job: list in, enriched list out, synced to the CRM. No conditional webhooks. No custom scrapers. Just accurate emails, phones, and firmographics on the accounts they already want, plus a few AI-researched fields.

For that job, a table-builder is the wrong shape. You end up paying for orchestration flexibility you never use, and paying a per-action tax on every AI question you ask.

That lane is exactly what Cleanlist AI is built for:

• $49 per seat a month on Starter (750 credits per seat), $89 per seat on Pro (1,500 credits per seat), one shared credit pool and no separate action currency. See pricing.
• 98% verified email and 85% direct dial on the 500-Lead Enrichment Benchmark, from a pre-built 25+ provider waterfall you never configure.
• AI research on every lead from Starter, at 3 to 5 credits a lead, with the source on every answer.
• Push leads to your CRM from Starter, with HubSpot and Salesforce sync on Pro.

The honest framing: Clay is a workbench, Cleanlist AI is a conveyor belt. If you want to build the machine, buy the workbench. If you want enriched lists coming out the other end every week, buy the conveyor belt. The side-by-side is in Cleanlist AI vs Clay, and if you are weighing Clay against a database vendor instead, see Clay vs ZoomInfo.

FAQ

What are Clay tables?

Clay tables are Clay's core interface: spreadsheets where each column executes an action per row, such as querying a data provider, running an AI research agent (Claygent), calling an API, or computing a formula. Rows hold companies or people, and columns enrich them. They function as a visual workflow engine for building and enriching lead lists.

How much does Clay cost per row?

On Clay's Growth plan ($495/month), a full contact enrichment (email, phone, title, company data) burns roughly 14 to 34 data credits, and up to 75 with deeper company data, per Michael Saruggia's 2026 teardown of the March pricing change. That is roughly $0.70 to $3.75 per fully enriched row on top of the subscription. Lighter lookups cost less; the exact number depends on which marketplace providers your waterfall hits. Full math in our Clay pricing breakdown.

Is there a simpler alternative to Clay?

Yes. If your workflow is list in, enriched list out, CRM sync, Cleanlist AI covers it from $49/month with a pre-built 25+ provider waterfall (98% verified email on its 500-lead benchmark) and AI research priced in the same credits. You skip provider selection, waterfall maintenance, and dual-currency credit math. Clay remains the stronger pick for complex custom workflows with a dedicated owner.

Do Clay credits roll over?

Yes, for data credits. Clay's docs say unused data credits on a monthly plan roll over up to 2x the plan's monthly amount, and annual plans that renew on the same or a higher tier get a 15% data credit rollover. Actions do not roll over. Enterprise contracts are negotiated case by case.