AI for membership organisations, used where members feel it.

02/ member service and chat

Member service and chat.

Members already use chatbots elsewhere. Yours answers in your name, so it must answer from your own pages.

The public is used to them. The government’s public attitudes tracker found 60% of UK adults had used an AI chatbot in the previous three months and 44% at least monthly — while seven in ten said they knew only a little or nothing about how AI systems are trained. Members will try yours; most will not know what it can and cannot know.

An organisation answers for what its chatbot says. In a Canadian tribunal case, Moffatt v Air Canada, the airline argued it could not be held liable for its chatbot’s wrong answer about bereavement fares. The tribunal called that “a remarkable submission”: the chatbot “is still just a part of Air Canada’s website”, and “it makes no difference whether the information comes from a static page or a chatbot.” That is not a UK ruling, but the Charity Commission makes the same point in its own terms: generative AI “can confidently produce inaccurate … or biased results”, and it expects human oversight to prevent material errors.

So the safe first job for member-facing AI is narrow: answer routine questions — renewal dates, CPD rules, event logistics, where to find a benefit — from content you already publish, say plainly that it is automated, and hand anything about a member’s status, fees or eligibility to a person.

We run one ourselves. Signed in on membership.quest, the membership advisor is a chat panel that asks about your organisation one question at a time — type, size, the biggest challenge, what success would look like, timeline — saves your answers to your account so you are not asked twice, suggests which of our services might fit, and offers to set up a call. Who you are comes from your sign-in, never from what you type, and nothing is saved for a visitor who is not signed in. It is new; we make no claims for what it achieves.

04/ personalisation and churn

Personalisation and churn prediction.

Prediction is only as good as the member record under it, and the obvious target is not always the right one.

Start with the record. A MemberWise practitioner piece on data and AI in membership organisations puts it bluntly: clean data before algorithms, connected systems before predictions — and warns that AI built on incomplete data “fills in the gaps” without telling you. If your CRM, events and learning systems do not share a member ID, fix that before buying a model.

Then aim carefully. Eva Ascarza’s field experiments, published in the Journal of Marketing Research, found that the customers a churn model rates most likely to leave “are not necessarily the best targets” for a retention programme. The better rule was to target people by their sensitivity to the intervention, “regardless of their risk of churning” — which you can only learn by holding back a test group.

Personalisation follows the same logic. Segment by need and career stage, show members the benefits that match why they joined, and check that it changed what they did. The personas and journey stages to personalise against are on the member journey; the engagement score and signals are on member engagement; the renewal numbers to move are on membership retention rate.

06/ an AI policy for your board

An AI policy for your board.

A one-page policy settles most of the questions staff are already asking. Copy the template and make it yours.

The Charity Commission’s statement on charities and AI is the clearest UK regulator view for membership charities, and a sound model for any board. It suggests considering an internal AI policy “so it is clear how and when it can be used in governance, by employees in their work, or in delivering services”. It is firm that trustees “remain responsible for decision making”: it is “vital this process is not delegated to AI or based on AI generated content alone.” And it expects human oversight to prevent material errors.

The government’s stated approach, in its February 2024 response to the AI regulation white paper, is context-based and “avoids unnecessary blanket rules”: five principles for existing regulators to apply — safety, security and robustness; appropriate transparency and explainability; fairness; accountability and governance; and contestability and redress. They make a good spine for a board policy.

Two public sources fill in the practice. The government’s own AI Playbook sets ten principles for civil servants, including knowing AI’s limitations, using it lawfully and responsibly, and keeping “meaningful human control at the right stage”. The National Cyber Security Centre’s guidelines for secure AI system development cover design, development, deployment and operation, and ask suppliers to be transparent about where users’ data is used or stored, including for retraining. Put that question to every AI supplier before member data goes in.