Start With Friction: Why Discovery Has to Come Before Deployment

The most common mistake nonprofits make with AI is skipping the listening phase. This is the third article in The Nonprofit AI Playbook series. Go back to Part 2: Where Does Your Nonprofit Stand? The 4-Rung AI Adoption Ladder.

There’s a number that should stop every nonprofit leader before they sign a single software contract. Across digital transformation projects of all kinds, roughly 70% fail to meet their objectives — and when researchers trace those failures back to their root cause, about 70% of them come down to problems with requirements. Put those two figures together and you get a sobering inference: nearly half of all technology initiatives fail because the organization never properly understood what it actually needed before it started building or buying.

That is the entire argument for this article in a single statistic. The right technology deployed against the wrong problem is still a failure. And the only way to know the right problem is to listen — systematically, structurally, and before you deploy anything at all.

This is the foundation of the whole crawl-walk-run progression. You crawl by listening first. Everything else in this series depends on getting this step right.

What Facilitated Discovery Actually Means for Nonprofits

Discovery is not a survey. It’s not a suggestion box, and it’s not an hour of brainstorming in a staff meeting. Facilitated discovery is a structured process designed to surface what’s actually true about how work gets done inside your organization — the real friction, the real workarounds, and the real opportunities — at every level, from frontline staff to leadership.

The word that matters most is facilitated. Unstructured conversation produces opinions. Structured facilitation produces actionable intelligence. The difference is the design: who’s in the room, what gets asked, how answers get captured, and how raw input becomes a prioritized map you can act on. A skilled facilitation process pulls out the things people don’t think to volunteer and creates the safety for them to say what they actually experience.

It also has to run in two directions at once. Discovery has to be both top-down and ground-up. Leadership holds the strategic context — where the organization is trying to go. The ground floor holds the operational truth — where the work actually snags. Talk only to leadership and you’ll build for problems that don’t exist. Talk only to the front line and you’ll optimize tasks that don’t matter to the mission. You need both.

This is the methodology behind the Liberating Facilitator, the free facilitation and project-planning platform we built specifically to structure this kind of work. It’s adapted from established facilitation practice, and it’s free precisely because good discovery shouldn’t be something only well-resourced organizations can afford. You’re welcome to use it whether or not you ever work with us — that’s the point of building it the way we did.

The Three Things Nonprofit Discovery Must Surface

Every effective discovery process is hunting for three specific things. Miss any one of them and the resulting roadmap will be incomplete.

First, where friction exists. These are the daily pain points people have learned to live with — the tasks that take longer than they should, the places where information gets lost or delayed, the parts of the job that drain time without advancing the mission. Staff have usually stopped complaining about these because they’ve accepted them as just how things are. Discovery’s job is to make the invisible visible again.

Second, how people are already working around that friction. This is the most important thing to surface and the one most assessments miss entirely. Somewhere in your organization, people have built unofficial shortcuts — a personal spreadsheet, a clever use of a free tool, a workflow they invented because the official process was too slow. These workarounds are a map. They show you exactly where the friction is most acute and, increasingly, where AI is already being used without any organizational awareness. The shadow systems people build are the truest signal you have.

Third, where the opportunities are to operationalize what’s working. Once you can see the friction and the workarounds, the opportunities reveal themselves: take the best of what people have already figured out informally and build it into official, supported, organization-wide practice. The smartest solutions are often already in the building. Discovery finds them.

What Nonprofit Friction Looks Like in Practice

Across the discovery work I’ve done with organizations in different subsectors, the friction is remarkably consistent. The specifics vary, but the patterns repeat.

Documentation is almost always the number one drain — frontline staff routinely losing the equivalent of a full workday each week to notes, reports, and required paperwork. Manual document triage is common: teams sorting by hand through enormous intake or referral packets to find the handful of pages that actually matter. Duplicate data entry shows up everywhere, with the same information typed into three or four systems that don’t talk to each other. Manual report compilation consumes whole roles, especially where every funder wants the same data formatted a different way. And paper-based tracking persists in surprising places, with staff recording information by hand and then transcribing it later for compliance.

The thread connecting all of them: most of this work exists because of legitimate regulatory and funder requirements that can’t simply be eliminated. The opportunity isn’t to make the requirements disappear. It’s to dramatically reduce the time, friction, and exhaustion of meeting them.

The Human Case for Reducing Administrative Burden

That exhaustion isn’t a soft concern. It’s the central workforce crisis of the sector, and it’s why this work matters beyond efficiency.

In 2025, 95% of nonprofit leaders cited burnout as a major challenge. The Center for Effective Philanthropy found that 76% of leaders say staff burnout is at least slightly affecting their organization’s ability to achieve its mission. Among burned-out staff, the overwhelming majority — more than nine in ten in one analysis — attribute it to excessive workload. And the cost of losing people to that burnout is brutal for organizations that can least afford it: turnover in the sector runs well above the rate in other industries, and replacing an employee can cost a meaningful fraction of their annual salary.

Here’s the connection that makes discovery a mission activity rather than an IT activity. When you ask staff what they’d do with the time if the administrative burden were reduced, they don’t say they’d work less. They say they’d spend more time with the people they serve — more direct contact, more of the work they came into the sector to do. Every hour reclaimed from documentation is an hour available for the mission. Discovery is how you find those hours.

Why Nonprofits Can’t Skip the Discovery Step

The temptation is always to move faster — to buy the tool everyone’s talking about and start using it. The failure data is the answer to that temptation.

Remember that roughly 70% of software implementations fail because of poor user adoption, and that the deepest root cause is a lack of clarity about what users actually needed. Technology deployed without discovery doesn’t remove friction; it adds a new layer of it on top of the old. Staff revert to their workarounds, maintain their shadow systems, and quietly route around the expensive new tool. The organization has now paid for software and gotten resistance in return.

There’s a second reason you can’t skip it, and it’s specific to AI: discovery is also how you surface the risks. You cannot write a meaningful AI policy until you understand how people are actually using AI and where they’re thinking about using it. The unsanctioned workflows that discovery reveals are exactly the practices your policy will need to address. Listening comes before governing, which is why discovery comes before the policy article that follows this one.

How to Structure Discovery Sessions at Your Nonprofit

The mechanics matter as much as the intent. A few principles that make discovery actually work:

Separate the levels. Run frontline staff, mid-level managers, and leadership in separate sessions, not together. People are far less candid about what’s broken when their supervisor is sitting across the table. Mixing levels produces a sanitized version of the truth.

Engineer for safety. Anonymity, explicit no-judgment ground rules, and a clear statement that there are no wrong answers. People need to feel safe saying “this is broken” and “I built a workaround the official process doesn’t allow” without fear of consequences.

Consider an outside facilitator. Candor improves measurably when the person asking the questions isn’t the person who controls budgets and reviews. In regulated or hierarchical environments especially, an external facilitator surfaces things internal facilitation never will.

Ask the questions that surface real friction. A few that consistently work:

• What takes far more time in your week than it should?
• Where do things fall through the cracks?
• If you could fix one thing about how information moves around here, what would it be?
• Are you using any tools or shortcuts that aren’t officially part of your workflow?
• What do you do by hand that you suspect a computer should be doing?

One of the most useful moments in a session is building something together in real time — taking the group’s own description of their work and generating a live opportunity map from it. It turns an abstract conversation into something concrete the room can react to, and it shows people the pace at which their input can become action.

From Raw Sessions to a Roadmap: The Discovery Engine

Running the sessions is only half the work. The other half is parsing what came out of them, and that’s a real task — discovery generates a large volume of qualitative input that has to be turned into a prioritized, defensible plan.

This is why we built the Discovery Engine, a companion tool that processes the data and transcripts from discovery sessions to identify patterns, surface technology opportunities, and produce prioritized recommendations. It takes the raw intelligence from the room and turns it into a roadmap — a ranked map of friction, risk, and opportunity that leadership can actually budget against. Having this capability changes the economics of doing discovery well, because the analysis that once took weeks of manual synthesis can be done quickly and consistently.

Turning Nonprofit Discovery Findings Into Action

The output of good discovery is a prioritized map: here’s where the friction is worst, here’s where the risk lives, here’s where the fastest and highest-value opportunities are. That map feeds directly into the next three things this series covers — policy, training, and deployment — in that order. Discovery first. Then governance. Then the people. Then the tools.

And discovery is not a one-time event. Friction changes as organizations grow, as staff turn over, and as tools evolve. The organizations that stay ahead build a regular discovery cadence into how they operate, revisiting what’s working and what’s snagging on a predictable rhythm rather than only when something breaks.

The next article takes the output of discovery and turns it into the thing most nonprofits get backward: an AI policy that actually fits the organization, written after the listening rather than before it.

This is the third article in a nine-part series on how nonprofits are leveraging AI and technology to advance their mission in 2026, produced by Xponent21. Statistics cited are drawn from research by the Center for Effective Philanthropy, the Givebutter and Gitnux nonprofit burnout analyses, and industry data on digital transformation and software adoption failure rates.

Click here to read Part 4: How to Write a Nonprofit AI Policy That Actually Works.