How we score
Clarity and energy are easy to fake. Getting a call back on track after it goes wrong is not. So we do not grade the performance. We build hard moments into every call and watch what the candidate does with them.
The rest of this page shows you exactly how that works. It takes about four minutes, and there is nothing technical in it.
Play the tape
Two people took the same seven-minute call. Same customer, same complaint, and the same hard moments at the same seconds.
Filled dots are the moments we wrote into the call. They happen at these exact seconds for every applicant.
What they said
“Good afternoon, thank you for calling Meridian Broadband. My name is Alex — how may I help you today?”
dashed line = how this candidate sounded in the warm-up
What we heard
Nothing to fault
Steady and quick, with no hesitation anywhere.
What they said
“Hi, good afternoon — you've reached Meridian, um, this is support. How can I help?”
dashed line = how this candidate sounded in the warm-up
What we heard
Two ums
Slower than A, and that is simply how this person sounds. It costs them nothing.
What we measured here
The first minute is easy on purpose and nothing in it is scored. We use it to learn how this person sounds when nothing has gone wrong. Later, when the call gets hard, we compare them to this. Not to a native speaker, and not to your best agent. Only to how they sounded a minute ago.
Four things we watch, not one
A support call is not one performance you grade once. Several different things happen inside it, and each one needs measuring in its own way. Here are ours, in the order a candidate meets them.
measures — How this person normally sounds
An easy opening that carries no score. It tells us how somebody sounds when nothing has gone wrong yet, so everything later in the call is measured against them rather than against anyone else.
This is why a strong regional accent costs a candidate nothing here. We never hold two candidates' voices up against each other. We only compare a candidate with how they sounded a minute earlier.
measures — Composure
The customer loses their temper. Same words, same second, same volume, for everyone who takes this pack. We measure how far the voice moves and how long it takes to settle again.
Sounding calm and being calm are not the same thing. Somebody whose voice never moves might be steady, or might be reading from a script. How they settle afterwards tells you which.
measures — Coachability
The customer tells the candidate they have a fact or a step wrong. We watch the next fifteen seconds. Do they take it and carry on, or argue and dig in?
Handling an angry customer means handling someone else's temper. Being told you are wrong means handling your own. They are different things, and this is the one a team leader cannot train out of somebody later.
measures — Resolution
Did the candidate deal with what the customer actually asked for? A fix, a hand-off, or a real next step with a name or a date attached to it.
This one is pass or fail. Fail it and no amount of clarity, warmth or composure will reach Take the call. It is the rule that stops a good phone voice beating a good agent.
And four things we deliberately do not do
We do not score how native someone sounds.
We only ask whether they are easy to understand for the customers this particular role serves. Never how close they are to any one accent.
Sounding good is never enough on its own.
Anyone can hold it together for thirty seconds. Polish moves a candidate up the list. It cannot get them past the one thing that decides it.
Every number points at a moment you can play.
Nothing is scored on a feeling. Each line on a report comes with the second it happened, and you can listen to it yourself.
We suggest. You decide.
We rank the candidates and show our working. A person makes every hiring decision, and the bottom group is grey, never red.
Scenario packs
A scenario pack works like a driving test. Every learner drives the same route and meets the same trouble at the same corners, but the examiner reacts to how you actually drive.
Three candidates and three different calls, all of them running into the same three moments at the same second.
Same customer, same complaint, same trouble at the same corners, for every person who applies. That is the only reason 431 candidates can sit on one ranked list. It is one test taken 431 times, not 431 different tests.
In between those fixed moments the customer reacts to what the candidate actually says. Calm them down and they soften. Argue and they get worse. So no two calls run the same way, and there is nothing useful to pass on to the next applicant in a WhatsApp group.
We record which pack and which version produced each score. So you can compare someone you assessed in March with someone you assess in October and know it is a fair comparison. If we change a pack, older scores stay readable instead of quietly shifting.
You pick the route — candidates never see the map
When you set up a role you choose the pack, and you see everything in it first: the calls a candidate walks into, what each one is really testing, and the traps built into them. Your applicants get the calls. They never get the list of what is being watched.
Free pilot batch · 25 assessments · no card, no seat licenses
Run HearMost beside your current process for one real hiring batch. Your candidates get a link — no logins, no IT, works with whatever you use today. You get a ranked shortlist you can press play on. If it doesn’t predict who survives training, throw us out.
After the pilot: pay per completed assessment, from ₹199. Abandoned calls cost you nothing. One avoided mis-hire pays for the whole batch.