Dennis Fois is CEO of Bloomerang, a giving platform for growing nonprofits, focused on technology that helps organizations scale impact.
The most useful market research I’ve seen in years wasn’t commissioned. It accumulated on its own, one typed question at a time.
Over the past year, fundraisers typed roughly 47,000 prompts into the conversational reporting AI inside our platform at Bloomerang. Each one is a small, unguarded record of what someone actually wanted from their data at the moment they wanted it. No survey design, no focus group. Just a person with a job to do, asking.
We read them. The pattern that emerged changed how I think about what our industry has spent the last decade building.
What People Actually Ask For
Here’s how a typical session goes. The request starts narrow and mechanical, such as: “Give me everyone who gave last year and hasn’t given yet.” A list, the kind of thing reporting tools have produced for 20 years. Within a few exchanges, however, the same person is somewhere else entirely. Why did the spring appeal underperform? Who should I call this week about a major gift? Which of the things we did last year actually got us to goal?
In 47,000 prompts, nobody asked for a dashboard. Nobody was trying to build a report. They were trying to answer a question.
The Dashboard Ceiling
That distinction sounds small. I don’t think it is. For a decade, the software industry’s answer to the analytics gap has been the better dashboard: more charts, more segments, more real-time totals. In fairness, dashboards do exactly what they were designed to do. They accurately and quickly show what happened. If the question is “where are we against goal,” a dashboard answers it well.
However, that isn’t the question. Our logs are a yearlong record of what people ask when they can ask anything, and they’re overwhelmingly “why” questions and “what next” questions. Why did this underperform? What should I run again? A dashboard can’t carry either of those. It was never designed to. We spent 10 years perfecting the answer to a question our users weren’t asking.
From Signals To Suggested Actions
What they’re asking for has a name: prescriptive analytics. Tools that attach a reason to a signal and a suggested action to the reason. The difference is concrete. A dashboard can show that the spring appeal missed its target. A prescriptive system can rank the likely causes, whether that’s the subject line, the send time or plain donor fatigue, and point to which donors deserve a call this afternoon. A lapsed donor stops being a row in a report and becomes a name, a likely reason they went quiet and a recommended next step.
The decision still belongs to a person, and it should. Giving is emotional. It’s personal to the donor and the story that moved them, and no model fully predicts it. What software can now predict, precisely enough to act on, is the behavior around the gift: who’s likely to lapse, who’s ready to give more and roughly how much. The system hands back a reason and a next step. The fundraiser decides and makes the call.
What This Means, Whichever Side Of The Software You’re On
The lesson applies twice, and it applies differently depending on whether you’re building the product or buying it.
If you’re building it, the same untouched research is sitting in your own logs. Look at what your users type into a search bar, a support ticket or a chat interface before you ask them what they want in the next release. The gap between what they ask and what your interface lets them ask is the road map.
Most enterprise software still answers what happened. The harder, more valuable question is why it happened and what to do next, and building that reasoning layer on top of a data store is a different technical problem than building another chart. It requires a system that can hold a hypothesis, rank causes and commit to a recommendation, not just a query that returns a filtered table.
If you’re buying it, first, the questions your team keeps asking are the right ones. When a fundraiser is frustrated that the reports never quite say why an appeal missed or who to call next, that isn’t a skills gap. Those were never questions a report could answer. The answers already sit in the data you have.
Second, and this is the one I’d act on this quarter: Change what you ask for in a software demo. Don’t ask to see the reporting suite. Ask the system why last year’s appeal underperformed and who your team should call tomorrow morning. If the answer is a chart, you’re looking at the last decade. If it’s a name, a reason and a recommended step, you’re looking at the next one.
None of this makes the dashboard obsolete. It makes the dashboard the starting point instead of the finish line because the finish line was always the decision that comes after it. Fundraisers have been asking for exactly that all along. We have 47,000 questions that prove it.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
