Everyone's selling you AI. Almost no one's asking what you're trying to achieve.
We say that a lot at Betty and across the Blue Cypress family, and new data from GrowthZone shows why it matters.
Their 2026 Association Operations Benchmark Report surveyed 195 association, chamber and nonprofit leaders. Two findings sit side by side:
- 57% of associations already pay separately for AI tools, outside their core operations system.
- 76% still run at least one core function mainly through manual processes.
The Association sector has bought AI. The work, for the most part, hasn't changed - yet.
The budget was never the problem
It'd be easy to blame budgets. Half the organisations surveyed spend $15,000 or less a year on all their technology combined, and smaller associations carry roughly four times the relative tech cost of their mid-sized peers.
The report is clear that cost isn't why functions stay manual. The leading reason is bandwidth. Nobody has had the time, or a clear trigger, to change. In the report's words, an established spreadsheet process outperforms a new tool that requires a setup project, regardless of that tool's capability.
It's the classic "she'll be right" approach to operations. The spreadsheet works well enough, so it stays, right up until the person who built it goes on leave.
And associations clearly will spend on AI. More than half already have, often without any prompting from their existing vendors.
The gap isn't between having AI and not having it. It's between buying a tool and solving a problem.
Where the gap costs the most
The report shows where that gap hurts. Events are the top three cost for more associations than any other function. Sponsorship is named a top revenue source by 69% of associations, yet almost none budget for it, and it's the function most likely to be run entirely by hand.
Those are the functions carrying the most money and the most staff time. They're also the ones least supported by technology.
They're also where member questions pile up. The annual meeting runs for four days, but members' questions keep coming all year. And sponsors increasingly want proof of relevance and audience insight, which is hard to show when everything lives in a spreadsheet. Some associations are already turning their AI assistants into sponsored and premium offerings. That only works when the AI is pointed at a clear outcome.
Start with the outcome, not the tool
This is where "what are you trying to achieve?" earns its place.
A general AI subscription answers a different question: what can this tool do? That's a vendor's question. An association's question is narrower and more useful:
- Do we want members to get trusted answers without waiting for staff?
- Do we want to free up the membership team during renewal season for higher value work?
- Do we want to show sponsors what our members actually care about?
- Do we want to stop members from going to ChatGPT, which gets things wrong with total confidence?
Each of those is a measurable outcome. Each point to a specific first use case. Each gives your board something concrete to approve.
Make the project small enough to carry
Starting with the outcome also solves the bandwidth problem, because it shrinks the project.
AI isn't zero effort. There's some hard yakka involved as we would say back home. The effort has to be small, clearly owned and time-boxed.
Our own onboarding team has seen this across roughly 100+ implementations: the fastest teams weren't bigger or more technical. They named one internal owner, decided early which content is public and which is protected, and set a soft-launch date. Most launched in under 100 days, and some in under 50.
Building it yourself goes the other way. The hard part was never the model. It's who keeps it working, and earning its budget line, on day one hundred.
Five questions to ask before you sign
- What are we trying to achieve, and how will we know? If the vendor hasn't asked you this, ask yourself.
- What will it need from my team? Get the answer in staff hours, not project phases.
- Does it start with the knowledge we already have? Your standards, research and member guidance already exist. AI that needs all of it rebuilt first is a long project.
- Does it give time back before it asks for more? Taking routine member questions off staff frees up room for the next step.
- Can we show the board a result within a quarter? More than a third of associations need board approval at around $5,000, and roughly a quarter aren't sure where their threshold sits. Report impact, not activity.
Bolt-on AI doesn't fix bandwidth
Expect every core system vendor to add AI features over the next year, and I reckon some will be useful. But the 57% bought AI separately because they needed it before their vendors offered it. And AI bolted onto a system you already can't find time for is just one more module to configure. As I've argued before, the lowest-friction decision isn't the same as the safest one.
The report also found that member demand rarely triggers technology purchases. Growth and contract renewals do. That makes renewal season the moment to ask what you're trying to achieve, before the easy option gets signed by default.
The bottom line
Associations have the appetite for AI, and many have the budget. What they're short on is time, and a clear answer to what it's all for.
Everyone's selling you AI. Almost no one's asking what you're trying to achieve.
Start there, and you'll be right.
Source: GrowthZone, 2026 Association Operations Benchmark Report (195 responses; the sample skews toward organisations with budgets under $2M).
About this post: Rob Barnes wrote this in concert with Claude, Anthropic's AI model, over many rounds of debate, pushback and the occasional laugh, until it was worthy of publication. Claude challenged claims, checked every statistic against the source report, and caught at least one line our own blog would have contradicted. Rob kept the final say, and the Australianisms.
Rob Barnes CAE MBA AAiP is a Co-Founder and CEO of Betty, the AI System built for associations and nonprofits, part of the Blue Cypress family. He is the co-author of The Association Knowledge Cycle.
