What Funders Actually Look For in a Home Visiting Fidelity Report (and How to Keep It Clean)

Reporting season has a particular feel to it. Somebody exports the calendar. Somebody else opens the spreadsheet where visits were meant to be logged and finds three weeks of blanks. A supervisor spends two evenings reconciling the two, and the number that finally goes out is defensible but nobody is entirely proud of it.

It does not have to work that way. A fidelity report is mostly arithmetic done honestly. The programs that dread it are the ones doing the arithmetic afterwards.

Here is what funders are actually reading, and how to keep the underlying record clean while the year is still happening.

What a fidelity report is really asking

Strip away the format differences between models and funders and almost every fidelity report answers four questions.

1. Did the families get the visits the model promised?

This is expected versus delivered. For each family, for a defined period, how many visits should have happened under the model and stage they were in, and how many actually did. Everything else in the report is context for this number.

2. Did the visits look like the model?

Duration, who was present, where the visit happened. A ten-minute doorstep conversation and a ninety-minute visit with the primary caregiver are both "a visit" in a calendar and are not the same thing in a model.

3. Did families stay?

Retention and attrition. When families left, at what stage, and — where you know — why.

4. When the number fell short, do you know why?

This is the question programs most often fail, and it is the one that most changes how a funder reads the rest. A program reporting eighty-two percent with a clear account of a vacancy, a flood, and four families who moved out of the service area reads as well managed. The same eighty-two percent with no explanation reads as a program that is not watching.

The number that gets you into trouble is the denominator

Almost every fidelity dispute is really an argument about the denominator.

If expected visits are calculated from a fixed figure typed in at enrollment, the denominator is wrong the moment a family changes stage. If they are calculated from the model''s stage rules — so many weeks weekly, then biweekly until birth, then weekly postpartum, then stepping down — the denominator moves with the family and the ratio means something.

Three rules keep the denominator honest:

• Define the period and stick to it. Monthly, quarterly, whatever your funder asks for. Do not mix.
• Prorate enrollment and exit. A family enrolled on the twentieth of the month did not owe you a full month of visits. A family who exited in week two did not owe you the rest of the quarter.
• Write down your rules once, in plain language, and apply them identically everywhere. A rule that lives in one supervisor''s head is not a rule.

Log the visit, not the intention

The single biggest source of dirty fidelity data is the gap between what was scheduled and what happened. A scheduled visit that the family cancelled is not a delivered visit. A visit that ran fifteen minutes because the baby was at the paediatrician is a delivered visit with a note attached.

So the record needs three states, not two: scheduled, delivered, and did not happen — with a reason on the third. Attempted-but-not-completed contacts matter too; several models count them separately, and they are the evidence that a home visitor was doing the work even when the door did not open.

Capture all of that at the point it happens. Same day, in the car, in under a minute. Anything that has to be reconstructed on Friday will be reconstructed generously, and generous reconstruction is exactly what an auditor is looking for.

Keep the exceptions where the number lives

Every program has a quarter with a story in it. A home visitor on leave for six weeks. A caseload transferred mid-period. A snowstorm. A family who moved three towns away and stayed enrolled by phone.

Those stories are legitimate and funders expect them — but only if they arrive attached to the number rather than added in the covering email afterwards. Note the exception in the record when it happens, on the family or the caseload it affects. Then the narrative section of the report is a matter of reading your own notes back.

Look at it monthly, not quarterly

A quarterly report reviewed quarterly is a post-mortem. The same report reviewed monthly is a management tool.

A monthly rhythm gives you two useful things. First, a home visitor who is drifting has eight weeks of runway to recover instead of none. Second, the quarterly number stops being a surprise to anyone, including the person who has to explain it.

The supervisor''s monthly review is short: expected versus delivered per home visitor, the families in the at-risk bucket, anyone with a stage transition coming, and any exception worth writing down. Fifteen minutes per caseload, and reporting season becomes an export.

For the home visitor''s side of that rhythm, we wrote a companion piece on keeping an NFP caseload on cadence week to week.

Report what you can actually evidence

Two small disciplines protect you more than any formatting effort.

Do not report a metric you cannot reconstruct. If someone asks how you arrived at a percentage and the answer involves a spreadsheet nobody kept, leave it out.

Do not round in your own favour. A program that reports seventy-eight percent and explains it keeps credibility. A program that reports eighty-five percent and cannot show the working spends its next site visit defending arithmetic instead of talking about families.

Mileage and cost data belong in the same routine

Many funders also want travel and cost-per-family figures, and those come from the same place: the visits you actually drove to. If your visit record is clean, mileage is a by-product rather than a second reconstruction job. Our guide to mileage tracking for home-visit professionals covers what counts.

Making it a by-product instead of a project

The goal is simple to state: by the time the report is due, there is nothing to assemble.

That happens when the system holding your schedule also holds the model rules, the stages, the delivered visits, and the reasons a visit did not happen. Casedaisy does this for home visiting programs — expected versus delivered per family and per caseload, for any period, exported as CSV or PDF, built from the same visits your home visitors already logged. It reports what you scheduled and delivered; it does not replace your model developer''s data system and it does not submit anything for you.

If you are starting from a spreadsheet, keep it. The free caseload template has the columns worth carrying across, and it imports in a couple of minutes.

The short version

Get the denominator right, log what happened rather than what was planned, note the exceptions where the number lives, and read the whole thing monthly. Do that and the fidelity report stops being a project. It becomes a description of a year you already know went well — or a year you already fixed.