Before You Add AI, Name the Decision

AI can help a nonprofit move faster. But if the question is unclear, it can also help the team move faster in the wrong direction.

It is Monday morning, and the leadership team is trying to understand whether year-end fundraising is on track.

Development has a report from the donor database. Finance has the latest budget numbers. Communications has email results. Someone opens an AI tool and starts pasting in information.

Within minutes, there are summaries, suggested messages, and a list of possible next steps.

There is just one problem: the team still has not agreed on the decision it is trying to make.

Are we deciding whether to change the campaign goal? Are we looking for donors who need personal attention? Are we trying to understand why responses have slowed? Or are we deciding whether another appeal would help?

Those are different questions. They require different information, different judgment, and different actions.

This is where many conversations about AI quietly go wrong. The conversation starts with what the tool can do instead of what the organization needs to decide.

A faster answer is not always a better decision

AI is useful at many things. It can organize information, compare documents, prepare a first draft, summarize a long discussion, and help someone see possibilities they may not have considered.

But it cannot rescue a question that was never clear in the first place.

Ask, “How is our fundraising going?” and you will probably receive a broad answer.

Ask, “Which expected year-end gifts are now outside their usual timing, and what context should we review before reaching out?” and the work becomes much more useful.

The difference is not better wording for the sake of better wording. The second question points toward a real decision. It helps the team understand who may need attention and what should happen next.

Before choosing an AI tool, it helps to name that decision.

Start with the moment that keeps repeating

The most useful starting point is usually not a grand AI strategy. It is an ordinary moment your team knows well.

Maybe the executive director walks into each board meeting unsure which number will be questioned.

Maybe the development team discovers a changing donor relationship later than it would have liked.

Maybe finance and fundraising each have a correct number, but nobody has time to explain why those numbers look different.

Maybe the team spends Friday afternoon rebuilding an update that will be out of date by Monday.

These moments are useful because they are specific. The people involved already understand the frustration, the information being gathered, and the consequence of getting the answer too late.

Instead of asking, “Where can we use AI?” try asking:

What decision do we repeatedly make with too little time, incomplete context, or too much manual preparation?

That question gives the technology a job.

Write the decision in one sentence

A simple sentence can bring surprising clarity:

“At the end of this review, we need to decide whether…”

For example:

“At the end of this review, we need to decide which five donor relationships deserve personal attention this week.”

“At the end of this review, we need to decide whether the campaign is behind because of timing or because something meaningful has changed.”

“At the end of this review, we need to decide what the board needs to understand before discussing next year’s budget.”

If the team cannot finish the sentence, it probably is not ready to evaluate a tool.

That is not a reason to avoid AI. It is a reason to spend a little more time on the problem first.

Decide what the technology may prepare

Once the decision is clear, the next question is what the technology should do and what must remain with a person.

Technology may be able to:

• Gather relevant information
• Compare recent activity with familiar patterns
• Point out something that changed
• Prepare questions for review
• Draft a possible next step
• Reduce the work required to assemble an update

A person still needs to understand the context, check the result, consider the relationship, and own the final decision.

That line matters in nonprofit work.

A donor relationship is not simply a sequence of transactions. A community member is not a data point. A board decision may carry history that no software can fully understand.

The goal is not to remove people from the work. It is to give them better preparation for the parts of the work that require their judgment.

Bring the relevant information into the same conversation

Even a clear question can produce an incomplete answer when the supporting information is scattered.

A fundraising system may show that giving is behind. Finance may know that a grant arrived later than expected. The email platform may show that supporters received two appeals close together. A program leader may know that an important story was delayed.

Each piece changes the interpretation.

This is why adding AI inside one system does not automatically give leadership a complete view. The tool can only reason from what it can see.

Before relying on an answer, ask:

• Which information was included?
• Which information was missing?
• Does the person reviewing the result understand both?
• Would another team see the situation differently?

Sometimes the most important thing an AI system can do is admit that the available information is not enough.

Try one small, reversible use

A responsible first step does not need to be a large rollout.

Choose one recurring question with a clear owner and relatively low risk. Decide what information may be used, who reviews the result, and what a useful outcome would look like.

Then run the experiment long enough to answer a few practical questions:

Did it actually save time?

Did it improve the decision or simply produce more material to read?

Was the answer accurate enough to use?

Did it create extra work for someone else?

Would the team choose to use it again?

The option to stop is important. A tool should earn a place in the organization’s routine rather than remain there because time has already been invested in it.

Where Donorlytics fits

Donorlytics is built around the space between information and action.

It works across the systems a nonprofit already uses, helps leadership see what changed, explains why it may matter, and prepares the next step for review.

The purpose is not to give a nonprofit one more place to ask broad questions. It is to help the organization bring the right context into a decision before the moment passes.

Leadership still decides what deserves attention. The team still owns its relationships. Human review remains where it matters.

The technology does the gathering and preparation so people can spend more of their time exercising judgment.

A better place to begin

Before booking another demonstration or adding another feature, ask one question:

What decision would we like to make earlier, with less manual preparation and more confidence?

That answer will tell you far more about the technology you need than a list of AI capabilities ever will.

Explore Donorlytics
Talk to our team →

Donorlytics helps nonprofits bring their data together, understand what is changing, and move on what matters. Because too much important work still depends on piecing it together by hand.

‍