Communications, Donors, and Running on Almost Nothing

A nonprofit has the obligations of an organisation and frequently the staff of a small team, which produces a specific and recognisable pattern.

It has to fundraise, deliver services, account to funders, comply with charity regulation, manage volunteers, communicate publicly, safeguard people, keep accounts, and run a board. A commercial organisation of the same complexity would have departments. A small charity has four people, two of whom are part-time and one of whom also delivers the service.

The consequence is not that things are done badly. It is that some things are not done at all, and the things that get dropped are consistent: reporting that is not due yet, supporter communication, evaluation, and anything without an immediate deadline.

That pattern determines where tooling helps most, and it is not where a commercial organisation would look. A business adopts AI to do the same work with fewer people. A nonprofit's realistic gain is doing the work it was already failing to do.

That reframing matters for how success is judged. If a charity adopts tooling and the staff are equally busy afterwards, that is not a failure. The question is whether the newsletter now goes out, whether funder reports are on time, whether volunteers get a proper induction, and whether anyone has looked at the outcome data.

The rest of this lesson works through the specific places that applies, in rough order of value.

The highest-value application in most nonprofits is not fundraising. It is the reporting that happens after the money arrives.

Why it matters more than it appears. Funders decide on repeat and follow-on funding substantially on the basis of how an organisation reports. A charity that reports on time, honestly, with real numbers and a clear account of what happened, becomes an organisation a funder wants to keep supporting. Renewal is cheaper than acquisition in fundraising exactly as it is in commerce.

And yet reporting is the thing that slips, because it has no deadline pressure until it does, and because the money has already been received.

What the workflow looks like. Your delivery records, attendance data, case notes and feedback exist somewhere, usually scattered. Assembling them into the funder's reporting structure is aggregation and reformatting, which is precisely what tooling does well. The narrative sections draw on the same source library the grants lesson described.

The conditions. Numbers come from your records, counted the same way you promised to count them. Beneficiary information is anonymised or aggregated before it goes anywhere near a general tool, which the confidentiality step below covers.

And one point of substance. Report what happened, including what did not work. Funders are markedly more comfortable with an organisation that says we reached thirty rather than the forty we projected, and here is why, than with one that quietly reports success every time. Honest variance builds the relationship; unexplained perfection erodes it.

A tool makes the reporting fast. It does not make it honest, and honest is the part that earns the next grant.

This is the most serious risk in the sector and the one most likely to be breached by accident, because the material is exactly what you want help with.

Nonprofits hold information about people in difficulty. Case notes. Safeguarding concerns. Health and disability information. Immigration status. Domestic abuse disclosures. Criminal justice involvement. Children's records. Financial hardship.

Under data protection law much of this is special category data attracting heightened protection, and beyond the legal position, disclosure can cause serious harm to people who came to you for help.

The temptation is real and specific. A case worker with forty case notes to summarise for a report has an obvious use for a summarisation tool, and those notes are the most sensitive material the organisation holds.

The practical rules, in order of importance.

Identifiable beneficiary information does not go into general consumer tools. Not anonymised-by-removing-the-name, which is frequently not anonymous at all: a case described by circumstances can identify someone in a small community more reliably than a name would.

Where you need help with case material, aggregate before you process. Counts, categories and themes rather than individual accounts.

Where individual-level processing is genuinely necessary, it needs a lawful basis, a proper processor arrangement, and a decision recorded by whoever is accountable for data protection, which in a small charity is a named person rather than a department.

And safeguarding information should be treated as never leaving your systems, whatever the terms of any tool.

The asymmetry to hold on to. The convenience gained is an hour. The harm caused by a disclosure falls on someone who was already vulnerable and who trusted you.

A structural problem specific to under-resourced organisations, and worth understanding before it happens.

A charity cannot justify paid tooling, so staff use free consumer tiers. Free tiers are generally the ones whose terms permit the broadest retention and use of submitted content, because that is the trade. So the organisations holding the most sensitive data end up on the terms offering the least protection, purely because of budget.

That inversion is worth naming, because it is not a failure of care by anyone in the organisation. It is a structural consequence of resourcing.

What breaks it, in order of preference. Nonprofit pricing, which many vendors offer and which small organisations frequently do not know about or do not ask for. Donated licences through technology-for-good programmes. And an explicit organisational decision that certain categories of work are not done with tools at all if the budget will not stretch, which is a legitimate answer.

The worst outcome is the unexamined one: individual staff quietly using personal free accounts for work involving beneficiaries, with nobody having decided anything.

So the practical step for a small organisation is small and specific. Ask what tools people are actually using, without blame, then make one decision about what may be used for what. That conversation costs an hour and it is the difference between a considered position and an accidental one.

Charity communications have a recognisable register, and generated content lands squarely in the worst version of it.

The sector's default voice is earnest, abstract and slightly desperate: vital work, vulnerable people, making a real difference, now more than ever. It is well-intentioned and it is exhausting to read, and the reason it persists is that it is what everyone else writes.

A model asked for a fundraising appeal produces the average of all that, which is the least effective possible version.

What actually moves supporters is the opposite, and it has been understood in fundraising practice for a long time. One person's specific story rather than a category. Concrete detail rather than abstraction. What their money literally did rather than what it enabled. A clear ask rather than an implied one. And plain language rather than sector vocabulary.

So the same principle as the small business cursus applies with force. Start from something only you have: what happened last Tuesday, what a beneficiary said, the specific problem you hit and solved. The model turns your material into readable copy. It cannot supply the material, and material is the whole of what makes the appeal work.

Two cautions specific to this sector.

Consent and dignity. A beneficiary's story is theirs. Using it requires their informed agreement, and the version that raises the most money is often not the version that treats them with the most respect. That tension is a judgement about your values, and it is not one to hand to a tool optimising for engagement.

And never fabricate a beneficiary or composite them without saying so. An invented story in a fundraising appeal is a donation obtained by deception, and it is fatal to an organisation whose entire asset is trust.

Donor management has an obvious automation story and a limit that is easy to cross.

What works well. Thanking people promptly, which organisations fail at constantly and which measurably affects whether someone gives again. Segmenting communication so a monthly giver of five pounds and a major donor do not receive identical mail. Drafting updates tailored to what a supporter funded. Identifying lapsed donors. Preparing briefings before a conversation with a significant donor.

That first one deserves emphasis. Prompt, specific thanks is among the strongest predictors of repeat giving, it is entirely within your control, and it is the thing most often dropped when a small team is busy. Automating it is a genuine gain with no downside.

Where the limit is. Personalisation that is generated rather than real reads worse than no personalisation. A supporter who receives a message referencing their interest in a programme they never funded has learned that your system is guessing, and that is worse than a plain thank you.

So personalisation should draw only on facts you actually hold: what they gave, when, to what, and what you have discussed. Never on inferred interests or generated context.

And the relationship with a significant donor is not automatable at all. Major giving runs on trust built over years through actual conversations, and a donor who discovers that the warm personal correspondence was generated has had something taken from them.

The test from the supply chain cursus applies here too. If the supporter knowing this was sent without a person reading it would change how they feel about you, do not send it that way.

Two groups get overlooked in discussions of nonprofit tooling, and both are places where modest help has outsized effect.

Volunteers. Most organisations induct volunteers badly, not through indifference but because writing induction material, role descriptions, training notes and answers to recurring questions is real work that nobody has time for. Producing that material from what staff already know is straightforward, and the effect on volunteer retention is real: people leave because they felt unsupported far more often than because the work was hard.

The limit is that a volunteer relationship is a relationship. Automated management makes people feel like resources, and unpaid people who feel like resources stop coming.

Trustees and governance. Boards need papers, and papers get written the weekend before by someone who is already overloaded, which means trustees frequently govern on inadequate information. Assembling board papers from existing material, summarising finances into something a non-financial trustee can interrogate, and drafting the minutes are all genuine help.

The boundary matters here and is worth stating. Trustees carry legal duties, and in most jurisdictions those include duties of care and prudence that attach to them personally. A board paper may be assembled with tooling; the decisions are the trustees', made on information they have satisfied themselves about. A summary that smooths over a financial difficulty is worse than no summary, because it produces a board that believes it is informed.

And the recurring theme of this cursus applies to both groups. The tool produces the material that supports people doing something. It does not do the thing, and in a sector built on people choosing to give time, money and trust, that distinction is the whole business.

What a charity with a handful of staff can write down and actually maintain.

Beneficiary data. Identifiable information about people we support does not go into external tools. Not with names removed, because circumstances identify people in a small community. Safeguarding material never leaves our systems. Where we need help with case material, we aggregate first.

Tools. A named list, chosen for data terms rather than features. We ask every vendor for nonprofit pricing, because it usually exists. Personal accounts are not used for work involving beneficiaries or supporters.

Numbers. From our records, counted the way we promised to count them in the application. Never generated. What we report against is what we said we would measure.

Applications. The organisational material comes from our source library. The local evidence, the outcomes and what is hard are written by us. We do not apply to funders we do not fit, however cheap applying has become.

Communications. Every appeal starts from something that actually happened. Beneficiary stories are used with informed consent and told with dignity. We never invent or composite a person without saying so.

Donors. Personalisation uses only facts we hold. Significant relationships are handled by a person.

Governance. Trustees decide on information they have satisfied themselves about. Papers may be assembled with tools; the duty is personal and stays with them.

And the objective. We use this to do the work we were failing to do, not to do the same work with fewer people. If we are equally busy and the reports are now on time and the newsletter goes out, it worked.