Check before you enrich: where the credits in a Clay table really go
Bad providers get the blame. A lot of the money goes to rows that were never going to send. Four of them, the checks that catch them, and a list that decides what is worth a lookup.
A lot of the money in a Clay table isn't lost on bad providers. It's lost on rows that were never going to send: info@ addresses, the same person pulled from two providers, “Founder | Ex-Stripe” in the last-name field, a +49 number on a US contact. You find out after the credits are gone.
So we built a step that runs before enrichment: more than 120 checks on any list. They run when the list comes in and tell you which rows are worth a lookup before a single provider is called. The second half of the idea is that the list itself decides what gets enriched: a score, and a rule whose action is the enrichment. Both halves are below.
Four rows that were never going to send
Each of the four is a real check in ListPlus, and the names below are the ones the Analyze panel shows. Put the rows into the checks and these are the findings that come back.
info@: Role-Based Email
info@brightpath-software.com is filled in, correct in format and nobody's address. Role-Based Email flags info@, sales@, office@ and similar mailboxes. A sequence sent there lands in a shared inbox, and a reverse lookup that is supposed to turn it into a person has nobody to find; ListPlus does not run one on a shared mailbox at all. The trigger template Clean up your list tags freemail, disposable and info@ addresses and marks them ignored, so they leave the working view without being deleted.
The same person twice: Duplicate Name + Company, Duplicate LinkedIn
Two providers, two rows: Daniel Brooks at “Ironclad Manufacturing Inc” with one address, Daniel Brooks at “Ironclad Manufacturing” with another, one LinkedIn URL with www. and one with a trailing slash. Duplicate Name + Company ignores the legal form, Duplicate LinkedIn ignores how the URL is written, and both flag the pair. Enrich both and you pay twice for one person, then email him twice. Fix merges each group: the most complete row stays, takes over the values only the other row had, and the other row is marked ignored. Similar Name + Company finds the near misses, a typo or an accent, for you to review.
“Founder | Ex-Stripe” in the last name: Invalid Characters in Name
Scraped profiles carry headline text into the name fields. A last name with a pipe in it is flagged by Invalid Characters in Name, and a lookup by name and company would start from a person who does not exist: at best it finds nothing, at worst someone else. Its sibling, LinkedIn Status Text in Name, catches “#OpenToWork” and “We're hiring” and removes them with one Fix. A pipe and a headline need a look from you, because no rule knows which half is the name.
+49 on a US contact: Phone-Country Mismatch
Phone-Country Mismatch compares the dial code with the country column: a +49 mobile on a contact in the United States gets “expected +1”. Either the country is wrong, the number is an old one, or the person really works abroad. It is a note, not an error, but it belongs in front of a mobile lookup at 12 credits or a call list, not after it.
The same report has more of the kind: Email Domain Typo (gmial.com), Disposable Email Domain, and Domain Cannot Receive Mail, a domain that does not exist and bounces every message.
What the checks are
More than 120 checks, across names, emails, phone numbers, companies, job titles, addresses, LinkedIn URLs, duplicates, compliance and completeness; which ones run depends on the columns your list has. All but eleven are rule-based. They run on their own when you first open the Analyze panel, and they are included in every plan without spending credits. The eleven AI checks, such as a fake-looking name, a person's name in the company field or a department instead of a job title, are optional and billed by usage, with the price shown before you start.
Checks only read. Nothing in the list changes until you click Fix, and Fix exists only where the right value is obvious: casing, stray spaces, a missing dial code when the list has a country, a duplicate merge. Click a finding to see its rows; tick several and Show only these filters the table to them.
Then the list decides what gets enriched
A clean row is not automatically worth a lookup. The second step is a score and a rule, and both live on the list itself, so every record that comes in later, from an import, a webhook or a form, passes the same gate.
The score
+ Column → Score gives every record points per criterion: decision maker +40, business email +20, personal email −10, or criteria of your own. A criterion is a rule on a column, built in the same editor the triggers use, and it can ask whether a record is already in another list, say your customers (−50). Tiers A, B and C can come as a column of their own. Rule criteria spend no credits; a criterion the AI judges does, once per distinct value.
The rule
Enrichment in ListPlus is a rule's action, not a separate job you start and forget. Open Tools → Triggers, start from the template Find work emails and add an AND condition: ICP Score at least 40. Conditions can also read the checks' results, for example the Cold Email verdict (Ready, Enrichable, Unusable) or “email is a business address”. Under Then, the Enrich action carries three settings: Fields to reach (work email, mobile, job title …), At most per record (5, 10, 25 or 50 credits) and what happens to cells: Fill empty cells only, or Replace empty and invalid values.
In the sample list from the video, the template alone matches 10 of 28 contacts; with the score as a brake, 3 are left. Seven lookups drop out before a credit is spent. That is a sample list, not a forecast for yours.
From then on the rule runs by itself. A step that would go past the budget is not started for that record, an email is charged only when one is found, and a record the rule already acted on is not enriched again when it changes later.
- Import the list. The checks run on their own.
- Fix what has a fix and take out what will never send, with the template Clean up your list.
- + Column → Score, with tiers if you like.
- Tools → Triggers: Find work emails plus “ICP Score at least 40”, a budget per record. Save.
For GTM engineers: the checks as an MCP tool
If your rows live in Clay, a sheet or a workflow of your own, you do not need a ListPlus list to use the checks. They are an MCP tool, check_records, that your agent can call. In Claude Code:
claude mcp add --transport http listplus https://listplus.ai/mcp --scope userSend the rows as JSON with whatever key names they have (first_name, Vorname, Mail …); the answer says how each key was read and returns, per record, the check id, its name, the field, the value and a proposed fix where there is one. Nothing is stored or changed. Without signing in, a call takes up to 25 records; signed in, up to 1,000. In n8n it is one HTTP node, so the check can sit in front of the enrichment step of a workflow you already run.
With the Set up right on its connection, an agent can also add the score column and the enrich rule to a ListPlus list (add_column, create_trigger) and set the rule up paused until you have agreed to the budget.
What the checks do not tell you
Rule-based checks read the value, not the mailbox: a well-formed business address can still bounce. Verification asks the mail server and costs 0.3 credits for a clear answer; catch-all and unknown are not charged. A Phone-Country Mismatch is sometimes correct. Duplicates are found inside one list or one call; two lists are combined with Consolidate Lists. And a clean row with a high score is a verdict on the data, not on whether you may contact the person. How many credits the step saves depends on your list; we do not claim a number.