Real-time call coaching

Live call coaching is a glanceable cue on the AE's screen, small enough to ignore, never a script in the buyer's ear.

The quality job is narrow. Surface a missed discovery question, a competitor mention, or a next-step ask while the conversation is still on that topic. The AE glances, uses it or mutes it, and keeps listening. Talking over the buyer to read the card is a failed prompt. A buyer who can tell someone is feeding lines is a failed rollout. Mute has to be one control, usable mid-call, and using it cannot be a performance flag.

Cresta, Balto, Gong, and Chorus are a class of conversation-intelligence and live-guidance tools, not a ranking. Some are sold as overlays during the call, some as record-and-review, some as both. Do not buy on a promised win-rate lift. You cannot isolate that number from the product, the market, and who sat on the call. Judge whether AEs still sound like themselves, whether prompts fire on the right moment, and whether managers actually open the recording.

A pre-call account briefing is the work before the meeting. Live coaching does not replace that page. It only covers what the AE forgot to ask once the buyer is already talking.

Every extra live prompt competes with the buyer for attention. Load three trigger types, each one line the AE already knows how to say.

Missed discovery question. Fire only after the buyer named a pain, process, or constraint and the AE moved on without a probe. The buyer names a manual charge-entry handoff and the AE never asks who owns it or what it costs. The prompt is one question, not a methodology sheet: "Who owns that handoff today, and what breaks if it stays as-is?" Score coverage after hang-up with a discovery gap analyzer. That pass is slower on purpose.

Competitor mention. Fire when a named competitor appears. Show the one claim enablement has approved and the one question to ask instead of a feature dump: "They named Acme. Ask what they liked in that evaluation, then stop." A battle-card wall of pricing and differentiators is how AEs stop hearing the next sentence. Two lines, already rehearsed. If they have not rehearsed it, the overlay will not teach it live.

Next-step ask. Fire late, when remaining time is short and no owner, working session, or follow-up meeting has been named. One line: "Who else needs to see this, and can we put time on the calendar before we hang up?" Do not fire it in the first minutes or while the buyer is still answering a pain question.

Objections are a separate overlay. If you want live help there, treat live objection-response prompts as its own product, not a fourth card on the same screen. After the call, a methodology scorecard against your published play beats stacking more in-call text.

Mute-able means the AE can silence one trigger type or the whole overlay for this meeting without a ticket. First-name intros, audio checks, and weekend small talk are not triggers. If the system prompts on "Hi, this is Sarah," you trained the AE to ignore the rest.

The following is a made-up but realistic trial design, not a case study and not reported results.

Meridian sells revenue-cycle software to independent clinics. Eight AEs, two enablement partners, conversation intelligence already recording discovery (the Gong and Chorus class). New hires were skipping the physician-owner question and freezing when an incumbent clearinghouse came up. Enablement added a live overlay in the Cresta and Balto class for the four AEs with under six months on the floor. The overlay was visible only to the AE.

Week one they loaded the full battle-card library, every methodology gap, and a rapport cue on first names. The overlay lit up on "Hi, I'm Priya." One AE started reading competitor bullets while the practice manager was still explaining why they had evaluated that vendor. Another talked over a physician-owner to ask the prompted impact question because the card was still on screen. A third muted everything after a wall of cards covered the notes pane. Managers had not watched a recording.

What they changed for the next month:

1. Three triggers only: abandoned pain, named competitor, missing next step. Everything else off.
2. Competitor card cut to one approved sentence and one question. Full battle cards stayed in the wiki, not in the overlay.
3. Mute expected. Using it is not a coaching note.
4. No prompt in the first two minutes. No prompt on greetings, names, or audio checks.
5. After every recorded discovery, manager and AE pick one ignored prompt and one used prompt from the tape. Prompt-accept rate is not a KPI.
6. Recording disclosure stays in the calendar invite and the first thirty seconds. The overlay never runs on a call that is not disclosed as recorded.

They still ran a post-call summary and CRM auto-update. The overlay is not the CRM clerk.

If the AE is reading, the buyer is not being heard. Stop the trial for that pattern, not for a low adherence score.

Watch the recording for three tells:

• The AE jumps the buyer's sentence to ask the prompted question. The cue was late, or they are treating it as a script. Fix timing, or teach ignore. Do not coach them to hit every card.
• Battle-card walls. A paragraph of differentiators is a second tab, not a prompt. Cut it until it fits in one glance.
• Greeting and rapport cues. Prompting on a first-name intro trains ignore. The useful cue gets ignored next.

Tenured AEs will mute faster. Value concentrates on new hires and unfamiliar competitors. If tenured AEs start sounding scripted, take it off those seats.

Latency is the same class of problem. A competitor card after the AE has moved on, or a next-step prompt mid-story, is noise. If a vendor trial cannot keep cues inside the live moment, you still have the recording. Use that.

Anything that listens to the call in order to prompt is recording, or close enough that legal and the buyer should treat it as recording. Disclose it.

Put the notice in the calendar body and say it in the first thirty seconds: the call is recorded for training and notes. Follow two-party or all-party consent rules where they apply. If a buyer declines, the overlay and the recorder both go off. Do not relabel the overlay as "AI notes" while a model is transcribing in the background. The buyer should never hear a prompt, a whisper, or a manager.

After hang-up, the recording is the coaching surface. Sit with the AE and the tape:

• Which prompt fired, and was the moment real?
• Did they ignore it because the buyer had already answered, or because the cue was wrong?
• Did they use it and talk over the person?
• What should the next meeting cover that this call left open?

That review is where quality actually moves. The overlay is a reminder. The tape is the lesson. Pair it with the methodology scorecard and the discovery gap analyzer. Neither belongs in the buyer's attention during the call.

Enablement should score the trial on whether new-hire discovery still sounds like a conversation, whether mute is used without stigma, and whether managers are watching recordings. Do not score it on win rate, prompt-accept counts, or talk time you cannot attribute to the overlay. If AEs sound more scripted, or buyers ask who else is on the call, turn the overlay off and keep the recorder.