Every fortnight, we run an open Digital Q&A session for charity and nonprofit teams – a chance to bring your questions and challenges: the things you’re stuck on, the decisions you’re not sure about. You can find out more and sign up for a future session here.
This fortnight’s session had a brilliant mix in the room, writes Matt Saunders. We had a family support worker struggling to recruit volunteers, a communications team from an international education charity experimenting with A/B testing, a helpline charity that had just launched a chatbot, a carers charity thinking about AI-powered navigation tools, a housing association trying to stay connected with tenants, and several others listening in and contributing. Four distinct conversations emerged. Here’s what came out of them.
Q: We get lots of interest from potential volunteers but almost nobody follows through to training. How do we fix the dropout problem?
This is a familiar pattern, and the frustrating thing is that it often isn’t a marketing problem – it’s a process problem. Getting people interested is one challenge; keeping them engaged through to trained and enrolled is a different one entirely.
A few things came up. The first is clarity of expectation. Vague volunteer appeals – “support families through isolation,” that kind of thing – attract vague interest. The more specific you can be about what the role actually looks like day to day, the more likely you are to attract people who will actually commit. A bullet-pointed breakdown of exactly what someone will be doing, how often, and for how long does more than an emotional appeal. One person in the room had just redesigned their volunteer poster along exactly these lines – moving from a broad description to specific, concrete examples. That tends to be a more effective approach
The second is the commitment ask itself. If your volunteer role requires weekly visits to the same family for up to a year, that’s a significant commitment – and being transparent about it upfront, rather than revealing it once someone’s already engaged, will reduce dropout at later stages even if it reduces initial interest. Better to attract ten genuinely committed candidates than thirty who disappear at the point of training.
The third is tracking where dropout happens. If you’re not mapping the volunteer journey – from initial expression of interest through to full enrolment – you can’t see where you’re losing people. A simple CRM or even a shared spreadsheet that tracks candidates through each stage will show you whether the drop-off is happening at first contact, at application, at interview, or at training. Each of those points has different causes and different solutions, and without the data you’re guessing.
The fourth is asking people directly. When someone drops out, or when you get someone on a call before they do, ask them what might get in the way of committing. People are often surprisingly open about this, and their answers will tell you which objections to address proactively in your recruitment communications.
And on social media for volunteer recruitment specifically: flip the framing. Instead of “we need volunteers,” lead with what volunteers get out of it – the skills, the experience, the community, the purpose. Especially for younger recruits who may be volunteering partly for their CV, what’s in it for them matters as much as what they’ll be doing.
Q: We’re doing A/B testing on our email campaigns but the results barely differ. Are we doing something wrong?
Probably not wrong – but possibly testing the wrong things, or measuring the wrong outcomes.
If you’re testing subject line length and seeing a 0.1% difference in open rates, that’s partly because subject line length is a relatively small variable, and partly because open rates are a notoriously unreliable metric. Security software on email servers frequently opens emails automatically to scan for malicious links – which registers as an open without any human ever seeing the email. Click-through rates have similar problems. These metrics can point in a general direction, but they’re too noisy and too easily distorted to be the primary thing you’re optimising for.
The more useful question is: what outcome are you ultimately trying to drive? If it’s donations, are you tracking whether email recipients donate, and at what rate? If it’s event sign-ups, are you tracking sign-ups back to specific email campaigns? If it’s partnerships, are you tracking enquiries? Once you’re measuring the end result rather than the intermediate steps, you can test variables that actually influence that result – and the differences between variants will be more meaningful.
A practical example: a subject line that generates curiosity might get a higher open rate than one that’s directly relevant to the recipient’s situation – but the latter is more likely to get the people who actually care to take action. Testing “which gets more opens?” gives you a different answer to “which gets more of the right people taking the right action?”
The other thing worth doing is making the gap between variants bigger. If you’re testing short versus long subject lines, try very short versus very long, with meaningfully different content, not minor variations. Small differences between variants will always produce small differences in results.
Q: We’ve just soft-launched a chatbot on our website to take pressure off our helpline. It’s working – 88 people used it in the first week. What should we do next?
First: the fact that you’ve done this at all puts you ahead of most organisations in the sector. Launching a chatbot trained only on your own website content – rather than a general AI that could hallucinate or pull from unreliable sources – is exactly the right approach for a health and advice context where accuracy matters.
A few thoughts on what comes next. The most important thing right now is to add a feedback mechanism. If someone uses the chatbot and it doesn’t answer their question well, and there’s no way for them to tell you that, you won’t know – and you won’t be able to fix it. A simple “did this help?” prompt at the end of a conversation gives you qualitative data that is far more useful than usage numbers alone. Without it, you risk a scenario where usage drops after a few months because people tried it, found it unhelpful, and gave up – and you never knew why.
The second is the language question, which you’re already thinking about correctly. People seeking help with drug or alcohol issues won’t necessarily use clinical terminology. People asking about sexual health may use informal or slang terms. Building in the language that your actual users use – rather than the language your organisation uses internally – will significantly improve how well the chatbot understands and responds to real queries. Your teams who handle these calls already know this language; work with them to build it in.
The third is using the chatbot’s failure log. Most platforms will show you questions the chatbot couldn’t answer. That list is a direct window into what people are actually looking for – and where your website content has gaps. Review it regularly and treat it as a content brief.
On measuring impact: it will be genuinely hard to know whether the chatbot is replacing helpline calls or reaching people who wouldn’t have called at all. Both outcomes are valuable, and they’re not mutually exclusive. The key is having a clear baseline – what call volumes and topics looked like before – so you have something to compare against in three to six months.
Q: We communicate with tenants mainly through social media and email, but we want to do more two-way engagement. What works?
This came through the chat from someone at a housing association working with tenants in rural areas with high levels of deprivation, so we couldn’t go as deep as we’d have liked – but a few thoughts are worth sharing.
Email remains the most reliable direct channel, especially for an audience that may not be active on social media or may move between platforms over time. If you have an app – which it sounded like this organisation does – push notifications and in-app messaging are valuable because that’s already where tenants expect to interact with you. Meeting people in the channels they’re already using is always more effective than asking them to come to you.
The thread running through everything: know your audience before you pick your channel
Different topics, different organisations, different challenges – but the same thing kept surfacing. Whether it was volunteer recruitment, email testing, chatbot development, or tenant communications, the conversations that got furthest were the ones that started by asking: who exactly are we trying to reach, and what do they actually need from us?
That question sounds obvious, but it’s harder than it looks. It’s easy to spend a lot of time optimising channels, testing subject lines, and scheduling posts without ever sitting down to really understand the person on the other end. The organisations making the most progress in these sessions are consistently the ones who’ve done that work – who know their audience well enough to say, this is what they want, this is where they are, and this is what we can offer them that nobody else can.
Everything else follows from that.
These sessions are always a reminder that the challenges facing charities aren’t really about size – they’re about resource, confidence, and finding the right place to focus. If any of the questions above resonated with you, we’d love to see you at a future fortnightly session. You can find out more and register here.
And if you’d rather have a one-to-one conversation about what any of this means for your organisation specifically, feel free to get in touch with me at All Things Equal – this is exactly the kind of work we do.