Home · Business · Marketing & Content · Email & CRM campaigns

PARTLY

As of 13 August 2026, AI can only partly find invalid email addresses in your customer database.

This still needs a person who signs their name to it.

Can you do it?

15 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ityou

What the alternative costsThe available tool data does not provide a price for a specialist email-validation service.

If this goes wrong: you suppress a valid customer or keep sending to an address that should have been removed, and the database requires manual correction.

What to actually do

  1. Hand it to a person

    The route this page recommends

    A person who owns the outcome does this end to end, worth it when the failure is dear.

  2. Use a tool built for this

    Second choice

  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Export the customer database to a CSV containing a record ID and email address, and remove fields that are not needed for this check.
    2. Open a chatbot, paste the prompt, and attach or paste the cleaned CSV.
    3. Save the returned table and filter it to the rows classified as invalid format, incomplete or duplicate.
    4. Compare every flagged address with the original customer record, correcting clear typing mistakes only when the correct address is confirmed from a trusted source.
    5. Check the model's totals against the number of rows in the export and inspect its stated rules for false positives such as unfamiliar but correctly formed domains.
    6. Use your email platform's own suppression or validation workflow for any address whose mailbox status cannot be established from the file.
    7. Keep the original export, the checked results and the final change log before applying suppressions or corrections to the live database.

    Prompt

    Analyse the attached customer email export. Use only the columns needed for this task and do not infer or reveal personal information beyond the email value and the supplied record ID.

For each row, classify the address as one of:
- valid format
- invalid format
- incomplete
- duplicate
- cannot determine

Flag clear problems such as missing text before or after the @ symbol, spaces, multiple @ symbols, malformed domains, invalid characters, and obvious transcription errors. Do not claim that a mailbox exists or does not exist unless the file contains verified delivery information. Do not call an address invalid solely because the domain is unfamiliar or because you cannot check it live.

Return a table containing only: record ID, email address, classification, reason, confidence, and recommended action. Keep the original email unchanged. Recommend "check manually" where the evidence is not conclusive. At the end, give totals for each classification and list the exact rules you applied. Do not change, delete or invent any customer records.

Customer export:
[PASTE THE CSV OR SPREADSHEET DATA HERE]

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • It cannot prove that a correctly formed address has a live mailbox without a separate verification process.
  • It cannot know whether a flagged address belongs to the right customer or whether a corrected address is genuine.
  • It cannot safely decide your organisation's suppression, retention or consent policy from the email column alone.
  • It may treat an unusual but legitimate address as an error unless you check the row against your customer records.
  • It does not apply changes to your live CRM unless you separately authorise and execute that workflow.

What caps this at PARTLY: verification cost, private data access and stakes of error.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs1
Verification1
Liability1
Effort delta2
Total7 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT check if an email address is valid?
It can check whether an address follows common formatting rules and flag obvious errors. It cannot reliably confirm that the mailbox exists or that the address belongs to the intended customer without a separate validation process.
Can AI clean my email list?
Partly. AI can identify malformed, incomplete and duplicated addresses in an export, but you should confirm each change before suppressing or replacing a customer record.
How do I find bad email addresses in a CSV?
Export the record ID and email columns, then ask a chatbot to classify formatting errors, duplicates and uncertain rows without changing the original values. Compare the results with the source file and use your email platform's validation or suppression workflow for mailbox status.
Is it safe to upload my customer email list to AI?
Only use a service approved by your organisation, minimise the fields you upload and check its data-handling terms before sharing the file. Not professional advice: for a serious data-protection concern, speak to your organisation's data-protection lead or a solicitor.

Nearby answers

All Email & CRM campaigns rulings

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.