Nonprofit marketing analytics: a sheet of metric definitions

What to take away

Decision first: one line at the top

Most nonprofit analytics projects begin with a tool and end with a report nobody opens. Reverse the order. Write down the decision you are trying to make, the person who makes it, and the date it is due.

A decision has a shape. "Should we move the year-end email send from December 28 to December 5?" is a decision. "Improve donor engagement" is a wish. Only the first one tells you what data to pull. Useful decisions for a marketing lead tend to fall into four buckets:

Each bucket needs a different number. Acquisition needs cost per acquired donor and first-year net revenue. Donor contact needs recency, frequency and giving history. Message testing needs a controlled split. Program funding needs a reconciliation between the fundraising report and the finance system.

Write the decision, the owner and the review date into one line at the top of the report. Everything below that line exists to serve it.

The metric sheet: seven definitions

Vanity metrics feel good and decide nothing. Page views, impressions, follower counts and email opens all rise when you send more, and none of them pays for a program. The seven metrics below each answer one question and carry a denominator and a period.

Two rules keep this table honest. First, every rate needs a denominator and a period written beside it. "Retention is 42%" means nothing; "42% of donors who gave in the 2025 fiscal year gave again in 2026" is a fact.

Second, use the same donor ID across systems. If the email platform and the CRM disagree on who gave, your retention figure is measuring your data entry, not your donors.

Then write the counting rule into the file, so the next analyst gets the same number from the same records.

Keeping a definition from drifting

A definition sheet only works if it is kept. When someone changes how lapsed is defined, or moves a cost line in or out of cost per acquired donor, the new figure is not comparable with last year's. Date the change on the sheet.

Put the sheet where the people who build reports can find it, and give it one owner. A report that quotes a metric should name the row of the sheet it follows, so a reader can check the counting rule in seconds.

If two reports show different figures for the same metric, compare the rows they follow before you compare the data. The gap usually sits in a numerator, a denominator or a period, not in the donors.

The record under the email rows: consent

Email metrics rest on a consent record. Canada's anti-spam law sets the rules for commercial electronic messages, including many fundraising emails. Express consent does not expire on its own, and implied consent has a window that depends on how the address was obtained.

The statute prohibits sending a commercial electronic message unless the recipient has consented, whether the consent is express or implied. Do not guess the window. Confirm the current rule with a qualified Canadian charity lawyer, because the answer changes with the circumstances.

Keep a consent record that shows the date, the source, the exact wording the person agreed to, and the withdrawal route. When someone unsubscribes, honor it across every system that can send, not just the one they clicked. Quebec Law 25 adds its own requirements for Quebec donors.

Where a workflow uses automated scoring or segmentation, the NIST AI RMF Playbook lists voluntary suggestions grouped under govern, map, measure and manage. It is a management aid, not a certification and not legal clearance.

The record under the claim rows: program sources

A metric that ends up in an appeal becomes a public claim. Every statement about impact, need, urgency, eligibility or cost should trace back to a document: the program source, the population, the method, the period and the approving role.

A single case story can show what one person experienced. It cannot establish that the program causes that outcome for everyone. A financial total reports funds received; it does not show what changed for participants.

Urgency must come from a real deadline or a documented need. A countdown timer, an implied shortage or an emotional image the facts do not support is manufactured pressure, and donors notice eventually.

Response rows: count only what was served

A campaign that produces responses it cannot serve damages trust faster than a campaign that produces none. A donation, a program inquiry and a volunteer sign-up each count only when the next step was served, so write that condition into the definition.

Test the normal path and one realistic failure before you increase reach. A restriction field that confuses donors, a form that fails an accessibility check, or a story subject who withdraws permission are all ordinary events, not edge cases.

Net revenue per donor and mission contribution

Gross revenue flatters a campaign. Net revenue tells you whether it was worth running.

Net revenue per donor equals total gifts, minus payment processing fees, minus receipting and acknowledgment cost, minus any refunds, minus the staff hours attributed to that campaign. State each of those as a named variable and let the reader substitute their own figures.

Attribution assigns credit under a rule. It does not prove the communication caused the outcome. A donor who gives after an email may have decided weeks earlier at an event.

Choose an outcome window that fits the action. A same-day click rate can diagnose a donation page. Donor retention, service completion or a program result may take months to move.

When the review shows that attributed response has been mixed with program impact, correct the statement even if revenue rose. Trust is the asset; the campaign is not.

Give the board a one-page file

Boards do not need campaign screenshots. They need a decision file they can read in ten minutes.

Name the role that can stop publication and the role that can accept an exception. Time-limited exceptions should expire unless a fresh review renews them.

Worked example: a year-end email decision

A mid-sized charity sends four year-end emails to a list of 40,000. The marketing lead wants to know whether to add a fifth.

Define the variables. Let L be the list size, O the average open rate, C the click rate, D the donation page conversion rate and G the average gift. Expected gross revenue from one extra send is roughly L × O × C × D × G.

Now subtract. Add the platform cost per thousand sends, the staff hours for copy, review and sign-off, and the expected increase in unsubscribes multiplied by the lifetime value of a lost donor.

Run the arithmetic twice, once with your observed rates and once with rates 20% lower. If the campaign only works at the optimistic figure, you have your answer.

The decision record should show the inputs, the date, the owner and the threshold at which the team would stop. That record is what makes the next year's planning faster.

Common questions

What should a nonprofit track first?

Start with cost per acquired donor, first-year net revenue per donor and retention rate by cohort. Those three tell you whether acquisition is paying for itself and whether the donors you already have are staying. Everything else is refinement.

How do we handle consent for fundraising email in Canada?

Name the authority and follow it. CASL sets the rules for commercial electronic messages, express consent does not expire on its own, and implied consent runs on a window that depends on how you got the address. Confirm the current position with a qualified Canadian charity lawyer.

When should a campaign be paused?

Pause when the response outruns the service or stewardship capacity behind it, when consent records are missing, or when a public claim cannot be traced to a program record. Protect the affected people first, then correct the statement or workflow.

What should the analytics file contain?

The decision, its owner and due date. The source data with dates. The consent and rights position. Gross and net cost. The outcome definition. Complaints, corrections and unresolved risks. Close it with the next review date so the file stays alive.