An Altru database rarely gets messy all at once. It happens a little at a time. A guest buys tickets online with a personal email and then joins at the desk with a work email. A seasonal cashier creates a new constituent because the search didn’t find “Bob” under “Robert.” An old import left a few hundred records with the city typed in the state field. Five years later, your membership counts don’t match finance, renewal mailings bounce, and nobody trusts the reports.
Cleaning up is mostly a project-management problem. The data fixes themselves are usually simple. What makes them hard is deciding what “clean” means, doing the work in the right order, and keeping the mess from coming back. Here’s how we’d approach it.
Start by measuring, not fixing
Before you change a single record, find out how big the problem is. Build a handful of constituent queries that each surface one kind of issue:
- Constituents with no email address and no mailing address (you can’t reach them at all)
- Records sharing the same email address
- Records sharing the same last name and postal code
- Addresses missing a ZIP code, or with obvious formatting problems
- Constituents with no revenue, membership, or ticket activity in many years
- Households with only one member, or individuals who should be in a household and aren’t
Write down the count for each. Those numbers become your baseline, and they’re how you show leadership that the project did something. They also tell you where to spend time first. A few thousand duplicate emails is a bigger problem than a few dozen bad ZIP codes.
Agree on rules before anyone edits
Cleanup projects go sideways when three people fix the same kind of problem three different ways. Settle these questions in writing first:
- Name formatting. Title case? Nicknames in the first-name field or a separate field? How do you handle “Mr. and Mrs.” records?
- Which record survives a merge. Usually the one with the longest giving or membership history, but decide ahead of time.
- When a record is “inactive.” Pick a rule (for example, no activity and no valid contact information after a certain number of years) and decide whether you mark them deceased, inactive, or leave them alone.
- Who is allowed to merge. Merging is hard to undo. Limit it to one or two trained people.
Work in this order
- Back up and test. If you have a sandbox or test environment, try any bulk change there first. If you’re unsure what your Altru subscription includes, ask Blackbaud support before running large updates in production.
- Fix formatting. Standardize names, states, and phone formats first. Duplicates are much easier to spot once “CA,” “Calif.,” and “California” all look the same.
- Resolve duplicates. Work from your highest-confidence matches (same email plus same last name) down to the fuzzy ones. Review fuzzy matches by hand. Two people with the same name at the same address may be a parent and child, not one person.
- Clean up households. After merges, check that households still make sense and that the right person is the primary contact. Membership mailings often depend on this.
- Update addresses. Address hygiene services such as NCOA processing catch people who have moved. Whether and how that runs for your Altru database depends on your agreement with Blackbaud, so confirm with them rather than assuming.
- Retire what’s dead. Apply your inactive rule last, so you aren’t marking records inactive that were about to be merged into an active member.
A note on duplicates in Altru
Altru has some duplicate-checking when records are added, and administrators can search for and merge possible duplicates. How well that catches real-world duplicates depends on how the records differ. Matching tends to miss the hard cases: different emails for the same person, nicknames, or a typo in the last name. If you rely only on what gets flagged automatically, expect to find more duplicates through your own queries. Run the “same email” and “same last name plus ZIP” queries above and compare them against what Altru suggests.
Keep it clean afterward
The cleanup itself is a one-time project. Staying clean is a habit. A few things that help:
- Train the desk to search before adding. Search by email or phone first, then by last name. Most new duplicates are created at point of sale.
- Look at your web forms. Online purchases are a common source of duplicates, because the guest types their own information however they like. Check how your online sales create or match constituents.
- Put your audit queries on a calendar. Rerun the baseline queries monthly or quarterly. If a number starts climbing, you’ll know which process is leaking.
- Review imports before they go in. Group sales lists, event registrations, and partner files should be checked against existing constituents before import, not after.
Where Social Good can help
Duplicates are usually the biggest and most tedious part of a cleanup. Deduper finds and fixes duplicate constituent and email records in Altru, so you can work through likely matches without building and cross-checking queries by hand. The rest of this guide, setting rules, fixing formatting, and training staff, still applies either way.
Quick answers
How long does an Altru cleanup take?
It depends on database size and how many people can help. Measuring first gives you a realistic estimate. Treat duplicates and households as the slowest part.
Should we delete old constituents?
Generally no. Records with any revenue or membership history are part of your financial and reporting record. Marking them inactive is usually safer than deleting them.
