B2B list building with Clay: signals, waterfalls, verification and CRM sync

What Clay is and how it charges

Clay is a data platform built around tables. Each row is a company or a person, and each column is a lookup, an enrichment, an AI step or an export. That layout suits list building, because every decision about a record (where the data came from, whether it was verified, how it scored) stays visible next to the record.

Pricing as published on Clay's pricing page, checked 18 October 2026 (Clay pricing):

The page is not consistent with itself. Its FAQ shows $185 for Launch and $495 for Growth, and one place lists 3K Launch data credits instead of 2.5K. Confirm the current figure with Clay before you budget. Annual plans are discounted.

The cost model has two meters. Data Credits pay for data bought from Clay's marketplace of 150+ providers and start at $0.05 each. Actions pay for platform work such as running tables, calling AI models, exporting and sending data to other systems, and the page says they cost less than a cent each. Clay's FAQ says "You'll only be charged Data Credits when Clay successfully returns data", and that enrichments that return nothing or hit an API error use neither credits nor Actions (Clay FAQ, checked 18 October 2026). Enrichments run through your own API key use an Action but no Data Credits.

Signals as the starting point

A list built from a one-time export ages from the day you pull it. A list built from signals adds rows when something relevant happens at a company, which gives your first line a real reason to exist.

Clay's signals documentation lists topic intent, web intent, new hires, news and fundraising, job changes, promotions, and custom signals (Clay University, checked 18 October 2026). The job change signal monitors contacts already in a table, needs each contact's LinkedIn URL as the identifier, and charges 1 Action plus 0.2 Data Credits per check (Clay University, checked 18 October 2026). On the pricing page, job change signals appear under Launch and web intent signals under Growth.

We pick one or two signals per segment and tie each to a line in the email. A hiring signal supports a first line about open roles, and a funding signal supports one about growth. Record the date of each signal in its own column so you can drop stale ones before sending. The signal-based outbound playbook covers which triggers are worth acting on.

The base list usually comes from elsewhere. In our UK B2B SaaS list-building work, we built 50K+ verified leads with Sales Navigator, Crunchbase and enrichment tools, and campaign performance rose 3x. Clay then does the enrichment, scoring and routing on top.

Enrichment waterfalls

A waterfall tries several data providers in a fixed order for the same field, such as a work email, and stops at the first confident result. Clay's documentation describes the sequence as running providers "in a predetermined sequence, so you don't duplicate tasks or spend extra credits", and lets you reorder, add or delete providers (Clay University, checked 18 October 2026).

Clay's work email waterfall adds a free first step called Infer Email, which guesses an address from the person's name and company domain. The doc says the default pattern returned a valid email "roughly 31% of the time" in internal testing on a software-industry dataset, and advises testing about 10 rows first. If the guess validates, no paid provider runs, and you pay only for the provider that finds a match (Clay University, checked 18 October 2026).

The last setting shows which providers earn their place: once you know which ones find the emails and which never do, you can prune the sequence and cut spend.

Clay's waterfall guide recommends starting with one field, usually work email, on a sample of 500 to 1,000 records that includes messy data, ordering providers cheapest-first among those that return trustworthy data, and setting a strict confidence threshold for cold outreach because a wrong address is worse than a missing one (Clay guide, checked 18 October 2026). The same guide reports a 2025 work email benchmark that Clay ran itself, finding that no single provider cleared both 95 percent quality and 90 percent coverage, and says its work email waterfall cascades across more than 100 email providers. These are Clay's own claims, and the page does not disclose methodology. An older University lesson says six email providers are available by default, so count what your workspace actually has before assuming the larger figure.

Verify before anything is sent

Finding an email and verifying it are separate jobs. Clay's work email waterfall offers four validation strategies. Conservative is the safest, includes all verified email types and is described as suited to cold outreach, where bounces affect sender reputation. Balanced accepts moderate risk and includes catch-all addresses. Aggressive prioritizes coverage and volume over precision, and Advanced gives manual control (Clay University, checked 18 October 2026). For cold lists we start at Conservative.

Catch-all addresses need their own handling. Hunter's verification guide says accept-all addresses are "27 times more likely to bounce than standard email addresses" and that most methods cannot confirm them as valid or invalid. It also recommends keeping bounce rates under 2 percent, verifying before sending rather than after bounces occur, and re-verifying older contacts before re-engaging them. Its data-decay estimate is about 2 percent a month, and a Hunter study found 2.3 percent of a sample list outdated after eight weeks (Hunter, checked 18 October 2026). Treat the 27x figure as Hunter's own claim; the direction is what matters.

Our working rules:

• Send the first wave only to addresses that pass the Conservative strategy.
• Hold catch-all addresses out until the first wave's bounce rate is known, then release them in small batches.
• Drop role addresses such as info@ and sales@ from cold lists.
• Verify a second time if the list is older than about eight weeks when it goes out.
• Treat 2 percent bounces as a ceiling. If a batch gets near it, pause and find out why.

The sending side of the same problem, including spam-rate limits, is covered in the deliverability playbook.

Segmentation and scoring in the table

Add columns for what the sequence needs to decide. Our usual set is an ICP score from 0 to 100, a segment tag, the signal and its date, verification status, and the provider that found the email. Filter to the rows that clear your score threshold and verification, and nothing else goes to the sequencing tool. How to build the score is in build an ICP for outbound.

Clay also has an Audiences feature that holds CRM, warehouse and enrichment data in one profile per contact and account, with segments defined by filters over CRM fields, enrichment data and signals, combined with AND and OR logic. The limits depend on plan. On Launch, Audiences imports only from Clay's own people and companies database and from CSV files. Growth supports up to 250,000 CRM or warehouse records with a daily refresh, and Enterprise supports up to 25 million with a 15-minute refresh (Clay University, checked 18 October 2026).

Dedupe before enrichment, because every duplicate costs credits twice. Use the company domain for companies and the verified email or LinkedIn URL for people. Merge subsidiaries that share a parent so two people at the same company do not land in the same sequence on the same day.

Our marketing agency list-building work produced 10K+ leads a month with 80% better data accuracy (case study). Our list building and CRM cleanup service builds and runs this process for clients.

CRM sync

The CRM should receive verified, scored, deduplicated rows and not much else. Two routes exist in Clay, and they behave differently.

For HubSpot, Clay's documentation says Look up object is the main matching tool and that matching on the record ID is "the most precise option available". The Domain filter reduces domains to the root domain unless you turn on subdomain matching. Update object has Ignore blank values on by default, so an empty Clay cell does not overwrite populated HubSpot data. Custom object permissions need a HubSpot Enterprise portal, and the sequence actions need sequence scopes that are disabled by default (Clay University, checked 18 October 2026). On the pricing page, CRM auto-sync appears under Growth and Enterprise only.

Our rules for CRM writes are short:

1. Look up before you create. A new record should only exist if the lookup on domain or email returned nothing.
2. Write to dedicated properties for source, verification date and ICP score, and leave fields that reps edit alone.
3. Sync only rows that passed verification and the score threshold.
4. Test with 10 rows and check the records by hand before running a full table.

Cleaning the records already in the CRM is a separate job, covered in CRM cleanup and data hygiene.

Questions

Does Clay charge for enrichments that return nothing?

Per Clay's FAQ, no. Data Credits are charged only when Clay successfully returns data, and a waterfall that finds nothing consumes no credits or Actions. Check your own table's run info on a sample before running a large column.

Which Clay plan do I need to sync with a CRM?

The pricing page lists CRM auto-sync under Growth and Enterprise, not under Free or Launch. The page's prices are inconsistent between its header and FAQ, so confirm with Clay. Checked 18 October 2026.

Which email validation strategy should I use for cold outreach?

Start with Conservative. Clay describes it as the safest option and suited to cold outreach, where bounces affect sender reputation. Balanced includes catch-all addresses, which Hunter says are 27 times more likely to bounce.

How often should I re-verify a list?

Re-verify before any new campaign that re-engages older contacts. Hunter estimates B2B email data decays about 2 percent a month, and our rule is to verify again if a list is older than about eight weeks when it goes out.