Which Signup Form Sends Bad Email Addresses Into Your CRM?
Updated November 20, 2026
7 min read
Your newsletter form is collecting subscribers. Your webinar page is bringing in registrations. Your demo form is generating requests for sales. The CRM keeps growing, and every dashboard makes that growth look encouraging.
Then someone checks the email addresses and finds a problem. Some can’t receive mail. Others need a closer look. You clean the list, but the next export brings the same problem back.
Before you repeat that cycle, look at where those addresses entered your system. One form might account for most of the undeliverable records, while another barely contributes any.
Here’s the rule: compare verification results by their recorded collection source, then investigate the sources with the clearest problems.
An AI agent can make that comparison for you. With the Emailable CLI, our command-line tool, an agent like Claude Code or Codex can verify addresses and analyze saved results alongside your source information, turning a recurring cleanup into a decision about which form to fix first.
As we noted in our post on verifying AI-sourced leads, source labels make recurring data-quality problems visible. Here’s how to use them.
Keep the Collection Source Attached
Start with an export from your form platform or CRM. Three details make the comparison useful:
1. The submitted email address. Keep it exactly as collected so its result can be matched correctly.
2. The form identifier. Use a recognizable label, such as newsletter, webinar registration, or demo request, plus the page or version if several forms share a name.
3. The submission date. This lets you compare addresses collected during the same period.
Make sure your source field records the collection event. A CRM field called “latest campaign” may describe something that happened after the contact first signed up.
The Emailable CLI supplies verification results; your export supplies the form history. Installing the CLI doesn’t give an agent access to your CRM or tell it where an address came from.
Have the agent match results to the original export by email address, ignoring capitalization and stray spaces. Matching by row position can attach a result to the wrong contact once rows are reordered or deduplicated.
Count each address once per form, keeping both appearances if it appeared on two forms. But verify it only once: have the agent check each unique address a single time, then attach that result to every form where it appears:
emailable batch verify unique-emails.csv --field email --wait --json
Put records with no source label in a separate “Source missing” group. Guessing a form from an address or company name makes the report look complete but less reliable.
Report blank email fields and unmatched records as visible data gaps, too, or an incomplete export could make one form look healthier than it is. A missing result also isn’t an unknown result, so don’t merge them.
Separate Undeliverable From Uncertain
Before ranking forms, decide what you’re measuring. Emailable distinguishes deliverable, risky, undeliverable, and unknown results. Risky addresses may still receive mail, while unknown means verification couldn’t reach a confident conclusion. Neither should automatically count as undeliverable.
This matters most for business audiences. A form attracting people from accept-all domains may collect more risky addresses without collecting more fake signups. Preserve the result and its flags instead of turning uncertainty into an accusation.
In our sample test, an agent verified 1,923 business addresses with the Emailable CLI: 878 deliverable, 728 risky, 294 undeliverable, and 23 unknown. The reasons add detail: 290 undeliverable results were rejected_email, and 16 unknowns were unavailable_smtp, meaning the mail server couldn’t be reached. That’s one list, not a benchmark, and it can’t say which form caused the problem.
The source column answers the next question: how are those results spread across your forms?
With saved results and source records in place, the request can be short:
Compare these saved verification results by form; show counts and percentages only, without verifying again.
Reorganizing saved results needs no new verification and uses no credits. If the results are still on Emailable, the agent can retrieve the existing batch instead of resubmitting the list.
Read the Reasons Before Blaming the Form
A high undeliverable rate tells you where to look. It doesn’t tell you why.
Suppose the webinar group shows many invalid_domain results. Typos like “gmail.con” are one possible cause. Other possibilities include misspelled company domains, expired or unregistered domains, fake domains entered during sign-up, and domains with missing or improperly configured email records. Another possibility is the integration itself: if the form export holds complete addresses but the CRM contains altered, truncated, or incorrectly mapped values, the problem sits between systems.
If the addresses match and rejected_email results dominate, examine the collection period, traffic sources, and any older records in the export. A reason describes what the check found; it doesn’t prove someone faked an address.
A rise in disposable addresses raises a different question: the offer may attract people who want a download, not a relationship. Review the offer and your collection policy first.
Look across forms, too. If unknown results rise everywhere in the same verification window, check the reasons before touching any signup page; unavailable_smtp points to temporary mail-server trouble, not the form.
Where labels exist, compare campaigns or page versions before changing the form. If a webinar form gets both advertising traffic and existing subscribers, a problem concentrated in one campaign points to targeting or the offer, not the form.
Your agent can summarize these patterns and suggest questions for the form owner. A useful finding sounds like this: “The webinar group has the highest undeliverable rate; compare its original submissions with the CRM records and review the campaign dates.”
That’s specific enough to act on without pretending the investigation is finished.
Fix the Collection Point That Needs Attention
Once you understand the pattern, choose a change that addresses it. Three responses are especially useful.
1. Help People Correct Mistakes While Signing Up
If the evidence points to input mistakes, let people fix their address before they finish the form. That’s why verification belongs in your signup flow: the person is still there to correct a typo. Emailable’s Widget verifies website email fields in real time, and the API supports custom signup workflows. Neither is part of the CLI analysis.
A suggested correction should stay a suggestion. When Emailable returns a did_you_mean value, let the person confirm the intended address, then verify it. Silently replacing an address can attach someone else’s inbox to the record.
Review the Widget’s settings and messages, too. Its allowed states default to deliverable, risky, and unknown, so it won’t block uncertain addresses unless you change that.
2. Check the Path From Form to CRM
If addresses change between the two exports, review the field mapping or import process. Confirm that the integration reads the intended email field and preserves the form identifier.
One older webinar page using a different field name can keep producing broken records. Fixing that mismatch beats cleaning up after it every month.
3. Give Uncertain Results Their Own Follow-Up
If a demo form attracts many risky business addresses, review the reasons before tightening its rules. Treating every risky result as fake could shut out legitimate prospects.
Route risky and unknown results to a separate review queue, following the rules in our Agentic Email Verification Playbook, and keep undeliverable addresses out of send ing. A deliverable result still doesn’t prove marketing consent or mailbox ownership.
Check Whether the Change Helped
After a change, compare the next collection period using the same definitions: the same form identifier, verification timing, and treatment of repeated addresses.
You can ask:
Compare this week’s saved form results with last week’s and show the changes in counts and rates.
Look beyond a falling percentage. A stricter form might cut its undeliverable rate while useful signups disappear, too, so review form completions and the number of deliverable addresses collected.
Say the webinar page drops from 13% undeliverable to 4%. Before celebrating, check whether traffic volume, targeting, or the offer changed at the same time. A different audience can change results even when the form doesn’t.
Share totals, not contacts. Form labels, counts, reasons, and percentages are enough for a team discussion; names, email addresses, and company domains don’t belong in the report.
Turn the Comparison Into a Reusable Skill
Once the report works, save it as a skill your agent can rerun.
Trigger it monthly or whenever a form, field mapping, or campaign changes. Its inputs are the source export, saved results, and your rules for repeated addresses and missing labels. Its output is the form-by-form table, reasons behind any outliers, data gaps, and an owner for each finding.
Keep its permissions narrow: it reads saved files and writes reports. Fresh verification needs approval and a credit budget, and the skill shouldn’t edit forms, CRM records, or Widget settings.
Before relying on it, test it on a copy with one source label removed. The report should show a “Source missing” group, not a quietly smaller total.
Give the Finding a Clear Next Step
The most useful output is a short report that names the form, shows the affected volume and rate, explains the strongest clue, and assigns someone to investigate.
Start with one recent export and the source information you already have. Let your agent organize the evidence, then use it to improve how addresses enter your CRM.
The best cleanup is the one that stops the same problem from coming back.