Coaching scorecards: turning call recordings into sales coaching

Thursday, 9:40 a.m. Before the weekly pipeline meeting, a sales manager opens her dashboard: six reps, 214 calls last week, three recordings she actually listened to end to end. She already knows the sentence she will say — "let's push a little harder on objections this week." The sentence is not wrong, but nobody's Friday morning will look different because of it. Nowhere is it written who backed off, on which type of objection, at which minute of the call. The team leaves the room with the information it walked in with: work harder.

This article is about the one document that closes that gap: the coaching scorecard. We start with why a scorecard says something different from "that was a good call," then which categories are worth measuring. From there: the reasoning behind a score, single signals such as talk-to-listen ratio, turning objections into observable behaviour, separating weekly noise from a real trend, running the scorecard inside a one-on-one, reading it at team level, and where the line sits between coaching and performance management. A three-week rollout order and an FAQ close the piece.

What does a scorecard say that "good call" cannot?

A manager's impression of a call is a function of the last five minutes she heard and the mood she was in. Impressions are not reproducible: play the same recording two weeks later and the verdict shifts. A scorecard records criteria that were published in advance instead of impressions. When a rep opens it, she sees which category lost points and which behaviour cost them. The argument moves off "you have it in for me" and onto "why was discovery scored low on this call."

The difference grows when you move from one call to many. Anyone can hold an opinion about a single conversation; forty calls scored against the same rubric produce a pattern. If a rep's discovery score has sat in the same band for three months, that is not a bad day, it is an unlearned skill. Quality assurance built on call recordings has enforced this discipline in support teams for years; what sales lacked was the translation of that rigour into coaching language.

The scorecard's second job is memory. What a rep was being developed on six months ago usually disappears when the manager who ran that plan moves on. A scorecard detaches the development topic from the person holding it: the next manager opens the history and picks the conversation up where it stopped.

Which categories belong on the card, and what stays off?

The most common design mistake is measuring everything measurable. A twenty-category card becomes a report nobody opens. Rocketly's Coaching Scorecards feature tracks a rep profile across four headings: communication, objection handling, closing and empathy. That set is not arbitrary — each maps to behaviour you can observe directly in a recording, and each can be rehearsed separately.

The single score given to a whole call sits in a different layer. Performance Score produces a mark out of 100 across communication, product knowledge, needs analysis, objection handling and closing. The scorecard keeps the movement of those marks over time; a single call's score is a data point, the card is the line through those points. Managers should read the line, not the dot.

What stays off matters as much as what goes on. Call duration alone is not a competency: a long call is not a good call, and a short one is not a failure. Closed-won count does not belong either — that is an outcome, not a behaviour. A scorecard measures behaviour precisely so coaching is possible. The moment outcome measurement replaces developing reps from conversation data, the card turns into a pressure instrument.

Why does a score without reasoning start an argument?

A rep's first reaction to a scorecard is almost always the same: "where did that number come from?" It is a fair question. A figure that does not carry its evidence reads as a verdict and triggers defence. Score Reasoning answers it directly: next to each sub-score sits the reason the mark was given and the supporting sentence pulled from the transcript. The discussion drops from the fairness of the number to the sentence that was actually said.

The practical value of that quoted line is that it fixes where the coaching conversation starts. Instead of "your discovery was weak," the rep's own answer to the budget question sits in front of both of you, word for word. In most sessions the rep spots the problem before you have delivered any feedback.

This is the largest contribution of systems that analyse conversations automatically: not that they speed up review, but that they make review auditable. When the source of a score is visible, the manager's own mistakes become visible too, and the card becomes a two-way document.

Is talk-to-listen ratio a skill or an excuse?

Talk-to-listen ratio is the most misused signal on a scorecard. Taken alone it degrades into "talk less," and the rep simply goes quiet. The Talk-to-Listen Ratio measure compares rep and customer speaking share against a healthy range and flags the dominant side. What matters is not the ratio itself but the stage at which it breaks.

A rep dominating a discovery call is a warning sign; the same share during a demo or a technical walkthrough is normal. So the card has to read the ratio together with the call type. A ratio that collapses in discovery usually reflects a thin question repertoire — a ready-made frame such as SPIN's situation, problem, implication and need-payoff questions gives the rep material that opens the conversation instead of filling it.

Monologue & Silence Flags complete the picture: the longest uninterrupted stretch of speech, the share of silence and long pauses are all marked. A four-minute monologue is a far more workable coaching item than a sixty-percent talk share, because it points to a specific minute of the recording that the two of you can listen to together.

How do objections turn into behaviour on the card?

The objection-handling category should not run on "did an objection come up" but on "what happened next." Objection Detection drops price, location, competitor comparison and decision blockers onto the timeline as objection moments. What the card needs is the thirty seconds after each of them: did the rep ask a question, change the subject, offer a discount, or say nothing.

A card that can make that distinction lets you aim coaching at a single objection type. Working only on price objections for a quarter beats touching five categories at once. Rehearsal gets concrete too: replaying three real recorded price objections and acting out the same moment three different ways leaves more behind than a classroom session.

There is a boundary to draw here. The card should push reps toward the concern behind an objection, not toward cornering the buyer; the ethical limit of persuasion technique belongs in the design of the rubric itself. If persistence becomes a scoring criterion, the team closes more this quarter and sees more refunds and more churn the next.

At what threshold does a category become a coaching session?

For a card to be useful, every category needs an action threshold — a simple rule that says which conversation happens when a value crosses a line. The table below shows example thresholds for the four core headings plus empathy; the numbers should be reset against your own team's distribution.

Category	Signal on the card	Example threshold	First move
Communication	Talk-to-listen ratio, longest monologue	Rep dominant in three consecutive discovery calls	Listen to the longest monologue together, stop it twice
Discovery	Customer turn count, open questions asked	Customer turns under half the rep's turns	Run a fixed question set on the next three calls
Objection handling	Objection moments and the response after each	Half of objections close without an answer	Pick one objection type and rehearse it from recordings
Closing	Closing attempt, clarity of the next step	No date agreed in a third of calls	Standardise the closing sentence of the call
Empathy	Direction of sentiment change, interruptions	Sentiment lower at the end than at the start	Mark the interruptions and discuss them the same week

Written thresholds stop coaching from running on a manager's mood. They also create predictability for the rep, who knows in advance which behaviour becomes a topic at which point.

How do you separate weekly noise from a real trend?

One week's score is noise. A difficult account mix, a holiday period or a new product launch will move a mark by a few points, and reacting to that movement makes coaching erratic. Coaching Scorecards track a rep profile on 7-to-90-day trends; decisions belong on the 30 and 90-day lines, while the 7-day line exists for early warning only.

A workable rule: a dip on the 7-day line raises a question, a dip on the 30-day line starts a conversation, and a dip on the 90-day line calls for a development plan. That three-step split breaks the reflex of walking over to a rep's desk after every wobble.

Scorecard data should not sit on its own either. Read alongside pipeline movement, it shows which behaviour change paid off at which stage; the revenue intelligence approach that joins conversation and pipeline data describes exactly that join. If closing scores rise while deals pile up in the proposal stage, the new skill has not converted into revenue yet.

How do you open the scorecard in a one-on-one?

How the card is presented matters as much as what it contains. Turning the screen around and reading out numbers triggers defence instantly. The sequence that works is usually: the rep narrates her own call first, then the card is opened, then one category is chosen. Touching three categories in one session changes none of them.

That meeting needs its own rhythm. A one-on-one template with a fixed agenda makes the card a normal part of the weekly cycle rather than a document that only appears when something is wrong. When the card is not seen as a carrier of bad news, arguments about the data drop sharply too.

A final step keeps the outcome from evaporating. Follow-up Automation turns highlighted risks and coaching items into a task or an appointment in one click and attaches them to the linked customer. "Let's work on that objection next week" then comes back as a trackable record at the next one-on-one. Bookmarks and Notes do the same job on the recording side, pinning a supervisor note to the exact second.

What appears when you read scorecards at team level?

Stack individual cards and the team's shared weakness becomes visible. If five of six reps score low on closing, the problem is not in the people but in the process or the message. Redesigning the closing sequence and rehearsing it once with the whole team beats six separate coaching conversations.

The team view also decides where comparison should stop. Turning scores into a public ranking creates movement in the short run and makes the data defensive in the long run. The motivational pull of a leaderboard works better on output metrics than on behaviour metrics; coaching categories should stay private to the rep.

Lost deals are the richest source for this reading. Rescoring the recordings of lost opportunities against the same categories shows which skill gap actually cost revenue. When a debrief run after a lost deal is fed by card data, what enters team memory is behaviour rather than excuses.

Where is the line between coaching and performance management?

If that line stays vague, the card loses its function. The moment a rep believes the card feeds compensation or exit decisions, she starts playing to the card: raising question count instead of asking better questions, going quiet instead of listening. A measure that becomes a target stops being a measure.

The healthy split is usually drawn like this: the card is for development, outcome metrics are for evaluation. Card data enters a performance conversation only as supporting evidence, in a form the rep can see herself. Drawing that line is one of the core duties of running a sales team, and unless it is stated in writing the team will assume the opposite.

The second requirement is scope. Scoring a sample rather than every call reduces the feeling of surveillance and keeps the manager's attention intact. Three calls per rep per week is a sufficient sample for most teams.

How do you stand the system up in three weeks?

The first week runs without scores. Recordings are collected, turned into text, and the current distribution is observed. Automatic Transcription converts audio in the background while the Bulk Analysis Queue pushes an existing archive through in one pass; the aim at this stage is to learn the team's current band, not to grade anyone. Setting thresholds against that band is far more realistic than copying an off-the-shelf standard.

In the second week the categories and thresholds are written down and announced. Which four headings are tracked, which threshold starts which conversation, and the fact that the card does not feed performance reviews are all said out loud. Every scorecard project that skips this announcement is read by the team as covert monitoring.

The third week brings the first one-on-ones. One category per rep, one recording listened to together, one rehearsal agreed for the following week. From week four the routine carries itself: the card joins the weekly cycle and trends start to accumulate. If you want to try that routine with your own team, you can open a free Rocketly account and send your first recordings through analysis.

FAQ

How many calls before a scorecard means anything?

One call gives you an idea but not a decision. In most teams a category-level trend becomes readable once fifteen to twenty scored calls have accumulated per rep. Before that, use the card to start conversations, not to set targets.

What if a rep disputes a score?

A dispute is a good sign, because it means the rep is looking at the data. Open the evidence sentence in the score reasoning and listen to the surrounding context in the transcript. If the criterion genuinely fails to capture the behaviour, correcting the criterion strengthens the card rather than weakening it.

Do we need to analyse every call?

No, and for most teams it is not efficient. A sample of three calls per rep per week is enough to surface a trend and to build a coaching agenda. Including calls at critical pipeline stages in that sample is usually sufficient.

Should the card work differently for a new hire?

Yes. For a new rep you read absolute level and rate of learning rather than trend, and you pick a single category for the first six weeks. For an experienced rep the absolute level is already high, so the real information is the direction of the 90-day line.