A prospect who was rated “low likelihood” three years ago has since joined at the patron level, given to the spring appeal, and bought a table at the gala. Their constituent record still says low likelihood. Nobody changed it because nothing in the database ever told anyone to.
Stale scores like this are the usual reason people go looking for how to “update” model ratings in Altru or Raiser’s Edge NXT. The answer is less about a setting and more about understanding where those ratings come from. Below is how they work, how to refresh them, what to do about ratings that show up as “undefined,” and how to manage all of it when your organization runs both systems.
Model ratings are snapshots, not live calculations
Model ratings, such as major gift likelihood, annual giving likelihood, planned giving likelihood, or a target gift range, are produced by a screening or modeling run. Blackbaud’s analytics services and third-party wealth screening vendors take your constituent and giving data, score it against their models and outside data, and send results back. Those results are then loaded onto constituent records.
Once loaded, the scores sit there. Neither RE NXT nor Altru re-runs the model when a constituent makes a new gift, upgrades a membership, or moves. That’s the core thing to understand: a model rating is only as current as the last time someone screened and imported.
If your organization has an analytics or screening service that refreshes automatically on a schedule, that’s a feature of the service contract, not of the database. If you aren’t sure what your contract includes, ask your Blackbaud account manager or the vendor directly before assuming a refresh is coming.
Three realistic ways to update ratings
1. Re-screen your file
The cleanest option is a new modeling or screening run. Before you send data out, clean it up: merge duplicate constituents, fix addresses, and mark deceased records. Screening results are matched back by constituent, so duplicates mean split scores and wasted screening fees.
Many organizations re-screen every one to three years, with a smaller rescreen of new donors and members in between. Pick a cadence your budget can sustain and put it on the calendar.
2. Import results with a clear rule for old ratings
When new results arrive, decide before the import whether you are replacing old ratings or adding new dated ones alongside them. Both are defensible:
- Replace keeps records tidy and avoids staff reading the wrong score.
- Add with a date preserves history, which helps if you ever want to see whether a prospect’s score climbed after cultivation.
Whichever you choose, use the same rating source and category names every time. Small variations (“Target Analytics” one year, “TA Model” the next) are what make ratings hard to report on later. Run the import against a small test group first and check a few records by hand before loading the full file.
3. Build an internal score from data you already have
Between screenings, you can maintain your own engagement score. A simple recency, frequency, and amount approach works well: when did they last give or renew, how many times have they given, and how much in total. In Altru, queries and smart fields can surface those values; in RE NXT, queries and lists do the same job. Store the result in a constituent attribute or a custom rating category clearly labeled as internal, so nobody mistakes it for a vendor model.
An internal score won’t replace wealth data, but it will flag the patron-level member who started giving, which is exactly what a three-year-old model misses.
When RE NXT shows a rating as “undefined”
Ratings that display as “undefined” in the RE NXT web view usually point to a mismatch between the imported value and how the rating source and category are configured. In our experience the common culprits are:
- A rating category set up as a table type, with an imported value that isn’t one of the table entries.
- A source and category combination that exists on records but was never defined in configuration, often after a vendor changed its naming.
- A data type mismatch, such as a numeric score loaded into a category expecting text, or the reverse.
To troubleshoot, pull a handful of affected records and compare the source, category, and value against your ratings configuration in database view. If one category accounts for most of the problem, fix the configuration or re-import that category with corrected values. Because display behavior can differ between database view and the web view, and can change with Blackbaud updates, confirm the cause with Blackbaud support before mass-editing records.
Running both Altru and RE NXT
Plenty of museums, zoos, and gardens use Altru for admissions and membership and RE NXT for major gifts. That split creates a question worth settling explicitly: where do ratings live?
Usually the answer is RE NXT, since that’s where gift officers work their portfolios. But the strongest signals for a cultivatable prospect, like frequent visits, membership upgrades, and event attendance, often sit in Altru. A few practical rules help:
- Make sure the constituent file you send for screening includes Altru members, not just RE NXT donors, matched by a shared ID where you have one.
- Load vendor ratings into one system only, and say which in your data standards document.
- If Altru staff need to see a prospect flag, pass a simple indicator rather than duplicating the full rating history.
How Altru itself stores rating or screening data can depend on your configuration and what services you’ve added. If you’re unsure whether your Altru database has a home for vendor ratings, check a test constituent record in a sandbox or ask Blackbaud support rather than inventing a workaround that conflicts with later imports.
Quick answers
Will ratings update on their own after new gifts?
No. Model ratings change only when new modeling results are imported. Internal scores built on queries can reflect new activity, because they read current data.
Should I delete old ratings after a rescreen?
Only if your policy says replace. If you keep them, make sure the dates are clear and that reports and lists pull the most recent rating.
