MenuStart a projectStart

AI & data

Charity supporter data clean-up and deduplication

Supporter data clean-up brings spreadsheets from events, appeals and forms into one place, standardises the details and suggests which records are the same person. A person approves each merge, and consent and Gift Aid records from both originals are kept.

An example of how the studio would build it. No live services are connected.

From scattered supporter lists to one record per person.

  1. Step 01

    Lists imported

    Spreadsheets from events, appeals and online forms are copied into a staging table, with the original files kept unchanged.

  2. Step 02

    Details tidied

    Names, email addresses, phone numbers and postcodes are put into one consistent format, and postcodes are checked against a postcode lookup.

  3. Step 03

    Matches suggested

    Exact matches are found by rules, and OpenAI judges the harder cases, such as a nickname, a typo or a changed surname, giving a reason for each.

  4. Step 04

    Merges approved

    A person reviews each suggested pair side by side in Airtable and chooses which details to keep.

  5. Step 05

    Records merged

    Approved merges are applied, keeping the history and every consent and Gift Aid record from both originals.

The result

One record per supporter, with nothing lost along the way.

No record is merged or deleted without a person’s approval, and the originals are kept so any merge can be undone. Consent and Gift Aid declarations are always carried over, never overwritten.

Before we connect it.

Where this is useful

Useful for charities, community groups and membership organisations whose supporter lists have grown across spreadsheets, so people get duplicate letters or their preferences are recorded in two places. A sensible first version cleans one historic list and produces a review queue of likely duplicates, before any new imports are switched on.

What we need to connect

  1. 01Copies of the spreadsheets and exports you want to combine, with a note of where each one came from.
  2. 02An Airtable base, or the database you plan to use, with agreed fields for contact details, consent, Gift Aid and history.
  3. 03Your rules for which details win in a merge, who approves merges and how consent records must be kept.

What needs careful handling

Two different people can share a name and address, such as a parent and child, so a match on those alone is never treated as certain. Consent recorded at different times can conflict, and the safest reading, usually the most recent and most restrictive, is applied and noted on the record. Supporter data is personal data, so only the fields needed for matching are sent to OpenAI, under terms that keep it out of model training where that choice is offered.

Questions people ask.

Why use AI at all rather than simple matching rules?

Rules handle exact and near-exact matches well, and they do most of the work. AI helps with the awkward cases, such as nicknames, a changed surname or a typo in an email address, and it explains each suggestion so the reviewer can judge it.

Can we move to a dedicated CRM later?

Yes. A clean, consistent Airtable base is a good starting point for moving to a charity CRM, and the same matching rules can check new imports in either system.

Question asked → Answer given
Automation 116

Ask your business data in plain English with MCP

Failure received → Final step taken
Automation 39

Failed Direct Debit recovery automation

All automation examples