Measuring Reassurance: Assessing Sales Conversations in a Confronting Decision
Measuring Reassurance: Assessing Sales Conversations in a Confronting Decision
You cannot score empathy. You can score whether the consultant asked about the person rather than only the condition, whether they acknowledged the situation before moving to process, and whether they said what would happen next and then did it. Those are observable facts, and together they account for most of what a customer means when they say someone seemed to care.
This is the central problem in measuring any emotionally difficult purchase: aged care, retirement living, funeral services, disability support, serious health decisions. The thing that determines the outcome is the thing that looks least measurable. This article sets out how it is done without turning the result into opinion.
Why can’t you just ask whether the consultant was caring?
Because the answer would be an assessor’s opinion, and a programme built on opinion fails in three predictable ways.
It cannot be defended
A site told that its consultant was insufficiently empathetic will dispute it, and they will be right to. There is nothing behind the finding except one person’s judgement, and the finding will be withdrawn. A programme that withdraws findings loses authority quickly.
It cannot be aggregated
If two assessors apply different internal standards, the scores are not comparable. You cannot compare sites, track change over time, or distinguish a real pattern from assessor variation.
It cannot be acted on
Even where the finding is accepted, nobody knows what to do with it. “Be more empathetic” is not a corrective action. It describes a desired outcome without naming a single changeable behaviour, which is why this kind of feedback produces defensiveness rather than change. Our article on training staff using real feedback covers why specificity is what makes feedback usable.
What do you measure instead?
Observable behaviours that produce the impression. The method is to work backwards from the impression to the things a person actually did.
Reassurance in a confronting conversation is generally produced by four behaviours, each of which can be observed and recorded factually.
Acknowledgement before process
Whether the consultant responded to the situation the customer described before moving to information, eligibility or next steps. This is binary and easily observed. A caller who explains that their mother has had a fall and is immediately asked for a postcode has experienced something specific and recordable.
Person-centred questioning
Whether the consultant asked anything about the person at the centre of the decision beyond their requirements. What they are like, what matters to them, what they enjoy. In care and retirement settings this is the behaviour families most often cite when describing a provider that felt different, and it is entirely observable.
Language matched to the listener
Whether scheme names, technical terms and fee structures were translated or recited. An assessor can record whether a term was used and whether it was explained, without forming any view about the consultant’s manner.
Commitment and completion
Whether the consultant said what would happen next, with a timeframe, and whether that then happened. This is the strongest single measure of trustworthiness available and it is purely factual. It also extends beyond the interaction, which means the checkpoint runs across the enquiry and the follow-up as one linked case.
How are these checkpoints worded?
The wording is what makes the difference between a defensible finding and an argument. Three rules.
- Name the behaviour, not the quality. “Acknowledged the circumstances the caller described before requesting details” rather than “was warm and welcoming”.
- Make it answerable from the interaction alone. If answering requires the assessor to infer intent or state of mind, it is not a checkpoint.
- Avoid compound checkpoints. “Was attentive and informative” asks two questions and cannot be scored once. Split it.
A useful test is whether two assessors given the same recording of an interaction would record the same answer. If you can imagine them differing, the wording is not finished. Our guide to designing an evaluation questionnaire covers the mechanics in more depth.
Where does narrative fit?
Structured checkpoints tell you what happened. They do not tell you what it felt like, and in this category that matters.
The answer is to collect both, and to keep them doing different jobs. Structured fields carry the scoring, the aggregation and anything that may be used in a corrective conversation. A short free-text account carries the texture: what the consultant actually said, how the conversation moved, what the assessor noticed.
The narrative is frequently the most valuable part of the report for the people who have to change something, because it is what makes a finding recognisable to the person who was in the room. What it cannot do is carry the measurement, because it cannot be compared across sites. Our article on what a good report contains covers how the two are presented together.
How should these results be scored?
Carefully, and usually not as a single number.
Averaging reassurance behaviours into one experience score produces a figure that moves without anyone knowing why. A site can improve its language while getting worse at follow-through and show no change at all. Reporting the behaviours individually, tracked over time, is more useful than a composite, even though a composite is what most people ask for. Our article on what mystery shopping scores really mean explains why an improving average can conceal a worsening problem.
There is also a specific risk in this category. Behaviours that are scored become behaviours that are performed. A consultant who learns that asking about the person scores well may ask a scripted question about the person, which produces the checkpoint without producing the reassurance. Two design choices reduce this: measure completion as well as conduct, since follow-through is much harder to perform than a question is, and rotate which behaviours are examined in depth rather than testing the same checklist every cycle.
How do you keep assessors consistent?
Even with well-worded checkpoints, assessors drift. Two safeguards do most of the work.
The first is calibration before fieldwork begins. Give every assessor the same interaction and compare how they recorded it. Where they differ, the disagreement almost always points at a checkpoint that is still asking for a judgement, which means the instrument gets fixed rather than the assessor corrected. Running this before a programme starts is considerably cheaper than discovering it in the first set of results.
The second is a short submission window. In a conversation-heavy assessment, the specific things that make a finding usable are the exact phrasing, the order things happened in, and what was not said. All of that thins within a day. A report written promptly records the interaction, and a report written later records a summary of it.
Neither safeguard is exotic. Both are routinely skipped, and skipping them is the usual reason a programme produces data nobody trusts.
What this approach cannot do
Three limits worth stating.
- It measures the interaction, not the relationship. A single conversation cannot tell you how a customer would be treated over months of contact.
- It measures what the customer experiences, not what the staff member intended. A consultant having a genuinely difficult day may produce a poor interaction while being an excellent employee, which is an argument for reading patterns rather than individual visits.
- It does not establish cause. Knowing that acknowledgement was missing at two thirds of sites tells you there is a problem in the process or the training, not which.
What does this look like in practice?
A worked comparison. Both of these describe the same interaction.
Opinion-based: “The consultant was compassionate and made the caller feel supported. Score: 4 of 5.”
Behaviour-based: “The caller explained that her father had been admitted to hospital after a fall. The consultant acknowledged the circumstances before requesting any details. She asked what her father had done before retiring and what he enjoyed. She used the term used for the assessment process and explained it without being asked. She said she would email a summary that afternoon and confirm availability within two working days. The email arrived the same afternoon. The availability confirmation was not received.”
The second is longer. It is also defensible, comparable across sites, and immediately actionable, because it identifies precisely one thing that did not happen. The first tells a site it did well and gives it nothing.
Who should see these results?
Behaviour-based findings in an emotionally loaded category need handling with more care than a compliance result, because they describe how a person conducted a conversation.
The practical question is whether results are reported at individual level at all. Many organisations deliberately choose not to, so that the programme measures the experience the customer receives rather than the performance of named staff. That choice usually improves cooperation, and it removes the risk of the programme being read as surveillance of people doing emotionally demanding work.
Where results are reported at individual level, it should be a deliberate decision with a documented rationale, staff should know the programme exists, and the privacy implications should be settled first. Information collected about identifiable people in the course of their work engages obligations under the Privacy Act 1988 (Cth) and the Australian Privacy Principles, including how it is stored, who may access it and how long it is kept.
In care settings specifically, there is a further argument for reporting at team level. The behaviours that produce reassurance are strongly shaped by workload, staffing and handover quality, none of which the individual controls. A finding presented as a personal shortcoming when the cause is structural damages the programme and fixes nothing.
Getting the design right
Measuring a confronting conversation well is a design problem rather than a fieldwork problem. Once the checkpoints name observable behaviours, the rest follows. Once they describe qualities, no amount of careful fieldwork rescues the result. We work with organisations in retirement living and aged care and other sectors where the purchase is difficult to build instruments that measure what actually matters to the customer. To talk through how this would apply to your conversations, get in touch to discuss your requirements.
