The Sales Calls Your Agency Records and Never Reviews

The calls nobody hears again

Somewhere on a server your agency pays for, there is a recording of the call where your newest producer fumbled the same objection your best producer handles without thinking. There is another one where a caller asked a question about network coverage and got a shaky, half-right answer instead of the confident one that closes. There is a third where the disclosure language ran fast and mumbled, the kind of thing that reads fine on a compliance checklist and sounds bad the moment CMS's own reviewer plays it back.

All three recordings exist because CMS requires them to. All three will sit untouched until the day, if it ever comes, that a complaint or an audit pulls them back up. Nobody at your agency is going to sit down this week and listen to them for what they actually are: a free, already-paid-for record of exactly how your team sells.

Meanwhile, the same producer keeps fumbling the same objection, call after call, because nobody ever told them it was happening. A caller who might have enrolled hangs up confused instead. A manager who could have caught the pattern in five minutes of listening never gets the chance, because five minutes of listening is not the part of the job anyone tracks, rewards, or schedules. The problem is not that your producers are bad at their jobs. It is that the one piece of information that would make them better already exists and nobody has built a path from the recording to the coaching conversation.

That is the gap this guide is about. Not whether you record calls, you already have to. Not how long you have to keep them, that is a retention question this site has covered elsewhere. The gap is between the compliance archive you are required to build and the coaching system you never got around to building on top of it, even though the raw material for both is the exact same file.

Why the review never happens

The honest answer is capacity, not neglect. A manager who wants to review calls has to carve real, uninterrupted time out of a day that is already full of their own selling, their own service work, and every other fire an agency owner has to put out. Listening to a call carefully enough to score it, not just skim it, takes longer than the call itself once you account for pausing to take notes.

Work the arithmetic plainly. Say a manager spends 12 to 15 minutes listening to and scoring one call well. A small team running 60 sales calls a week, which is a modest volume across three or four producers during a normal month, would need 12 to 15 hours of a manager's undivided attention to review every single one. That is close to two full workdays, and it assumes the manager does nothing else that week. It is illustrative arithmetic, not a published statistic, but the shape of it holds at almost any agency size: full manual review does not scale with a growing book.

So agencies do what capacity allows. A manager overhears part of a call while walking past a desk. A new hire gets shadowed for their first week and then is on their own. Feedback becomes whatever a manager happens to catch rather than a system that looks at the whole picture. None of that is a knock on any specific agency. It is what happens when the only tool for review is a person's limited attention applied to a growing pile of recordings.

There is a second, quieter reason review never happens, and it has nothing to do with time. Nobody at most agencies actually owns it. Compliance owns producing the recording if CMS or a plaintiff's attorney ever asks for it. IT or the CRM vendor owns the fact that it gets stored somewhere for the required window. Sales management owns hitting the month's numbers. None of those three roles has "turn this archive into a coaching system" written into their job, so it falls into the gap between them, which is exactly where things that are nobody's explicit job tend to die.

Compare that to how a strong sales floor actually runs. The best sales organizations do not treat call review as an afterthought squeezed into whatever time is left. They build it into the week on purpose, with an owner, a cadence, and a rubric, precisely because they know that without an assigned owner, "we should really be listening to more calls" stays a sentence people say in a meeting and never becomes something anyone actually does.

What the industry data actually shows

A 2026 national survey of independent agencies by the Big "I" Agents Council for Technology found that just 8.29% of independent agencies use AI regularly and strategically in their operations, even though 68% plan to increase their AI use over the next twelve months¹. That gap between intent and actual practice is the whole story of this guide in one number. Agencies know something needs to change. Almost none of them have closed the loop yet.

The same report found 56% of agencies have no written AI use policy at all, and 44% still rely on informal, peer-to-peer training to bring staff up to speed on new tools and new approaches¹. Peer-to-peer training is exactly how bad habits spread as easily as good ones. If your best producer never gets recorded and studied, and your newest producer only learns by watching whoever happens to be free that week, the agency has no actual mechanism for making everyone sound like the best person on the team.

It is not only an agency-side gap. On the carrier and enterprise side of the same industry, adoption is much further along. A Deloitte Center for Financial Services survey of 200 US insurance executives, fielded in June 2024, found that 76% had already implemented generative AI in one or more business functions, with life and annuity insurers slightly ahead at 82% against 70% for property and casualty². Forty-five percent of leaders in that same survey said the benefits of gen AI already outweigh the risks².

What a scored call actually catches

A rubric only works if it looks for specific, repeatable things rather than a vague sense of whether a call "went well." Across the calls we have listened to while scoping builds for agencies, the same handful of patterns show up again and again, and every one of them is something a reviewer, human or automated, can catch on a first pass.

None of these require a compliance background to spot. They require someone, or something, to actually listen for them on a consistent basis, which circles back to the capacity problem from the last section. A rubric like this is also exactly the kind of structured, repeatable check that a scoring system can apply to every call rather than the handful a manager gets to.

What the review gap costs

Put a real dollar figure under the illustrative math from earlier. The US Bureau of Labor Statistics reports the median annual wage for insurance sales agents was $60,370 in May 2024³, which works out to roughly $29.02 an hour using the standard 2,080-hour work year. That figure is a stand-in for the value of an hour of selling or managing time at a typical agency, not a claim about any specific role's salary.

Now run the worked example from the section above through that hourly figure. A manager who spends 12 to 15 hours a week trying to keep up with full manual call review, on top of their actual job, is spending roughly $348 to $435 a week of fully loaded time on a task they can only ever complete a fraction of. And that is the cost of attempting full review. Most agencies never attempt it, so the real number is not the hours spent. It is the hours never spent, and the bad habits, missed disclosures, and inconsistent scripts that keep compounding because nobody caught them.

How to fix it yourself

You do not need software to start closing this gap. You need a repeatable habit, which is worth more than any tool if the tool never gets used. Here is the method, in full, so an agency that would rather run this manually has everything it needs.

Run that five-step loop for a quarter and you will already be ahead of most agencies, because you will have turned some fraction of your compliance archive into an actual coaching signal. An agency owner who reads this and decides to run it by hand is making a reasonable choice. The honest limit is the same one from the cost section: a human reviewer can sustain a sample, not full coverage, and the calls that never get sampled are the ones where the real problems hide.

One more thing worth planning for if more than one person ends up reviewing calls: two reviewers using the same rubric will still drift apart over time unless they calibrate against each other occasionally. Once a quarter, have both reviewers score the same three calls independently and compare notes on where their scores diverged. Otherwise a producer's score ends up depending on which manager happened to review them that week, which defeats the point of having a rubric in the first place.

Why this matters more heading into AEP

Medicare's Annual Enrollment Period runs October 15 through December 7 every year, and it is the single window when most agencies run the highest call volume, add the most temporary or new-to-industry help, and put the most pressure on every producer to move fast. It is also exactly the window when an unreviewed bad habit does the most damage, because whatever a producer is doing wrong in September gets repeated across the largest number of calls they will make all year.

A gap that costs you a handful of missed appointments in a quiet month costs you dozens of them across seven weeks of AEP volume, and a disclosure line that runs sloppy on one call in September is the same disclosure line running sloppy on hundreds of calls once volume peaks. The agencies that build a review habit before AEP starts catch and correct these patterns while the stakes are still small. The ones that wait find out what was wrong from a complaint, a low quality score, or a CMS inquiry, after the volume has already made the problem bigger.

How we build it instead

This section sticks to what is stated on our own live pages and in our own service descriptions, verified while writing this. Among the custom AI builds we scope for agencies is call analysis and producer coaching, built alongside the other AI agents in an agency's actual workflow: an AI receptionist that answers after hours, conversational follow-up that keeps context on every contact, database reactivation, and appointment automation. Everything is built on your agency's own accounts, your own domain, and your own CRM, and you own it, and it keeps working whether or not you keep working with us⁴.

The mechanism is the same recordings you already keep for CMS. Instead of a manager sampling three calls a week by hand, an automated call analysis layer transcribes and scores every recorded call against a rubric, then surfaces the ones that actually need a human's attention, a missed disclosure line, a fumbled objection, a producer who is drifting from the script that closes. The manager's limited coaching time goes to the calls that need it instead of whichever ones they happened to overhear.

That same context layer connects to the rest of the pipeline, not just the phone. If a producer's follow-up texts and a caller's plan questions already live in one memory per contact, the calls you review are not isolated data points, they are one more channel feeding the same picture of how a specific relationship is going⁵.

For a Medicare, ACA, or health-focused agency, there is a compliance layer under this that a bolt-on transcription tool rarely thinks about. A system that touches recordings with health details connects to your CRM and other systems, and every connection is a place protected information can leak. Our builds are engineered so that handling is HIPAA compliant by design, with coded, safe information passing between systems and raw protected data never held where it should not be. We say HIPAA compliant plainly, and we do not promise a compliance outcome, because the responsible way to say it is to build it correctly and let the architecture speak.

We are not going to promise you a close rate, a retention number, or a revenue figure, because those depend on your market, your book, and the people having the conversations. What we will say, and what is on the live page, is that you describe it, we build it, and you own it, on your accounts and your domain⁴.

What you actually get

Concretely, moving from occasional manual spot-checks to a build like this gets you three things. Every recorded call gets scored, not just the handful a manager had time for. The calls that actually need a human's attention rise to the top instead of getting buried in the ones that were fine. And the pattern across a producer's calls becomes visible over weeks, not something you notice for the first time when a complaint arrives.

Underneath those three, you get the quieter benefit that matters most as your book grows: the recordings CMS already requires you to keep stop being a cost center that only exists in case of an audit, and start being an asset your best producers are effectively teaching from, whether they know it or not.

There is also a training benefit that compounds beyond any single quarter. Once a year or more of scored calls exists, a new hire's first weeks stop being guesswork built on whichever senior producer happens to have time to shadow them. You can point them to actual examples of a well-handled objection, a clean disclosure, and a real appointment ask, pulled from your own agency's own calls, not a generic training script written for nobody's specific book. That library is worth more the longer it exists, and it only exists once review stops being occasional.

This is not the retention rule

It is worth being precise about what this guide is not, because the two topics get mixed together easily. CMS's marketing rules already require TPMOs, including independent agents and brokers, to record the full audio of Medicare sales and enrollment calls and retain them for a period CMS sets, and that retention requirement recently changed for contract year 2027. That is a separate, fully sourced topic covered in our guide to the call recording rule change, including exactly what shifted and when.

This guide assumes you are already compliant with that requirement, because most agencies are, and asks the next question: given that the recordings already exist, why do they almost never get used for anything besides sitting in storage. The retention rule tells you how long to keep the file. It says nothing about whether anyone ever listens to it, and that gap is where the actual coaching opportunity, and the actual compliance risk, both live.

Questions agencies ask