Most maternity services are not short of feedback. Surveys, the Friends and Family Test, complaints, PALS contacts, MNVP engagement, birth reflection conversations, incident reports, claims, staff observations, the messages that arrive informally and never quite get logged — the volume is considerable. Ask a board whether the organisation listens to women and families, and the honest answer is usually "constantly."

And yet the same services are often surprised by what a review, an inquest or a national report later reveals. The information was frequently already present, in someone's words, months earlier. The problem was rarely that no one spoke. It was that no one could hear it as a signal.

The gap between hearing and listening

Feedback becomes intelligence only when three things happen: it's gathered from the people who actually have something to tell you, it's connected to everything else you know, and it reaches someone with the authority and curiosity to act. In most organisations, each of those steps leaks.

The routes that collect feedback tend to favour people who are already confident, digitally included and inclined to trust services. The women whose experiences would tell you the most — those who were frightened, not listened to, discharged early, traumatised, or who simply gave up — are frequently the least likely to fill in the form. The sample is quietly skewed toward the people least likely to have been harmed.

Then the sources sit apart. Complaints live with one team, incidents with another, survey data with a third, birth reflection notes in a clinic, claims with the legal team. Each uses its own categories. A woman who describes a frightening induction, a delayed escalation and a dismissive conversation afterwards may appear — fragmented — in four systems, none of which knows about the other three. The pattern that would be obvious if you read her account whole is invisible when it's been chopped into codes.

And where the information does surface, it often arrives at the board as a number: satisfaction up two points, complaints down four. A metric can tell you that something moved. It almost never tells you why, or whether the thing that moved is the thing that matters.

What experience reveals that metrics miss

The reason to take this seriously isn't compassion for its own sake, important as that is. It's that people describe the mechanics of unsafe care with remarkable precision, in ordinary language, before those mechanics produce a recordable harm.

They tell you that no one explained what was happening. That they raised a concern three times and felt brushed aside. That the handover between teams dropped something. That they didn't understand what they were consenting to. That they were left alone when they were frightened. Communication, escalation, continuity, consent, compassion — these are exactly the failure modes that sit underneath the most serious maternity incidents. Women often narrate the near-miss version of an event that, in another woman, becomes a catastrophe.

Treated as a satisfaction score, that account is noise around a number. Treated as safety intelligence, it's an early warning.

What it takes to close the gap

Turning experience into intelligence is a design problem, not a values problem. Nobody needs persuading that listening is good. What organisations need is an architecture that makes listening lead somewhere:

  • Deliberately hear the people who are missing. Assume your existing routes under-represent the most affected women, and build ways in for those who are traumatised, neurodivergent, digitally excluded or from communities that services have earned less trust from.
  • Read across the sources, not just within them. Someone needs a mandate to look at complaints, incidents, narrative feedback, birth reflection and claims together, and to work with the accounts as accounts — not only as codes.
  • Preserve the whole story. The meaning is usually in the sequence and the context. Flattening a distressing experience into a sentiment tag or a theme count throws away the part that would have told you what to fix.
  • Give it somewhere to land. Themes need an owner, a governance route and a visible line to action — and the loop has to be closed back to families, staff and the board, so people can see that speaking changed something.
  • Support the staff who do the listening. Hearing difficult accounts, especially after harm, is skilled and costly work. Without time, confidence and psychological safety, defensiveness fills the gap, and the signal is lost at the point of collection.

None of this replaces conventional safety measurement. It makes it more honest. Incident data tells you what your systems noticed. Experience data tells you what they missed. An organisation that can hold both — and connect them — knows far more about its own risk than one relying on either alone.

The question worth asking

The useful question for a maternity leader isn't "are we listening?" You are; everyone is. It's sharper than that: can we show what we heard, what it revealed about safety, and what changed as a result? If the honest answer is uncertain, the feedback isn't the problem. The architecture around it is — and that can be redesigned.