AI Turns Recorded Sales Calls Into Coaching for Reps and More Accurate Forecasts for Leaders

Sales call transcripts can run much of a revenue operation once AI analyzes them. Paired with win/loss records, those recordings tell reps their next move in each opportunity and give leaders a clear view of which customers convert and where the quarter will land. Most teams still limit AI to account research, call summaries, and CRM updates. Pointed at the sales process itself, it gives growing companies coaching, targeting and forecasting before they can staff a full RevOps team.

Brad Habansky is the Chief Sales Officer at Pipe17, a commerce orchestration platform that helps ecommerce brands and third-party logistics providers manage their order operations. He oversees sales, pre-sales, sales development and revenue operations, drawing on more than 25 years in technology sales. As the fourth U.S. employee at commercetools, he helped build the company’s go-to-market across North and South America, and he later led global sales at Elastic Path and served as Chief Sales Officer at inriver. Over the past year, he and his team have restructured how Pipe17’s go-to-market organization works with AI.

“Teams are using AI for all the easy, obvious stuff. You really have to start thinking about how you use it to sell better,” says Habansky. Sales leadership has changed quickly over the past 18 months, and the most effective teams now apply AI to how deals get won. At Pipe17, that shift starts with the company’s archive of past customer conversations.

A coach on every call

Every recorded sales call shows how a deal moved forward or stalled. Across enough calls, those recordings reveal which behaviors separate wins from losses, and AI can turn those patterns into guidance reps can apply to open deals. The results depend on training it with the right examples, meaning well-run wins and the losses a team should have won.

Pipe17 built a separate AI coach for each of its three sales motions, since each one has its own buyers and cycle length. “We have transcripts for every call, for everything we’ve ever done. I have all the data I need. I just have to drop it in and do the analysis,” says Habansky.

With that foundation, an AI coach can map out an ideal sales cycle, including the activities, metrics, exit criteria and assets each stage requires. Teams can then add their ICP and preferred sales methodology. Because the model learns from deals the company has already won, the framework reflects what works with its own buyers and spares the team months of internal debate over what each stage should include.

After each call, reps upload the transcript and get feedback on gaps, where the opportunity stands and what to do next. When the system is also trained on company messaging, it can help reps tailor a presentation to each prospect’s problems and business case.

It also gives reps a sounding board when their manager isn’t available. “Reps can spitball and whiteboard with the coach. What if I do this? What if they said this? It’s like talking to me about deal strategy,” Habansky notes. That access matters more as a team grows and leaders have less time for one-on-one coaching. New hires ramp faster as well, because they learn the team’s proven approach while working live deals.

Narrowing the ICP

For early-stage companies, Habansky recommends recording customer calls from day one, even before the team has closed enough deals to know what a good sales cycle looks like. Companies don’t need an expensive tech stack to start. A basic CRM can serve as the system of record, with more of the budget going toward AI, and any tool that produces call transcripts will work.

“All you need is transcripts. You have to have something the AI can learn from, and then it can help you,” says Habansky. A useful coach starts with four to seven well-run wins and three to five losses for each sales motion, and early-stage companies rarely struggle to find the losses. With that base in place, teams can iterate quickly and shorten the roughly 18 months it has traditionally taken to build a new market.

Win and loss data also shows where a company’s product fits today, which can be a narrower market than leadership expects. Grounding the ICP in which deals close and why gives the whole team a shared, evidence-based view of where to focus. “We could probably sell this to anybody, but not at a 35 percent win rate. At a 6 or 7 percent win rate, that doesn’t scale,” he adds.

Calculating how many leads each segment needs to hit the revenue target makes that tradeoff concrete. At a low win rate, a team needs far more leads than most marketing budgets can generate, which points the team toward the segments that convert best. The ICP can expand later as the company adds customers.

Metrics that sharpen coaching

The data behind an AI coach also gives sales leaders a clearer view of each rep. Alongside win rates and pipeline, it helps to track how long qualified opportunities survive and whether they move through stages in order. Those signals show which reps let weak opportunities through and which screen out too many.

Call recordings can then explain what a metric flags. In one case, two reps converted marketing-qualified leads at very different rates, and their calls showed one rep stopped asking questions after the first problem a prospect raised. “Once the customer gave him one pain point, he would go with that rather than digging to find three or four more. What’s important to one person isn’t important to everybody in the organization,” says Habansky.

That habit is costly for companies selling at a premium, where reps need several problems with a dollar value attached to justify the price. The rep needed to ask more questions earlier. AI shows leaders where a rep is struggling, but figuring out why still takes a manager’s judgment.

Annual plans usually start as spreadsheets built backward from board targets, and they rarely change once the year begins. Turning that spreadsheet into a live tracker of leads, conversion rates, win rates and average order value by source lets targets adjust as numbers shift. Pricing can draw from the tracker too, keeping quotes tied to the value each customer stands to gain.

“Our forecast has been within 10 percent accuracy every quarter since we started using it. Win rates are up over 10 percent, average order values have doubled and sales cycle times are down 30 percent,” says Habansky. Planning also takes less time. With the data behind each target already documented, a plan that once took a month can come together in a couple of hours.

Where leaders focus next

With AI handling much of the operational work, leaders who came up as top sellers no longer have to become forecasting experts on their own. “Someone who was a great salesperson but had a weakness in sales ops, AI fills that gap for you. The stuff you need now is the sales skills, the EQ, the gut feel, the things you learn from being in a lot of deals,” Habansky notes.

Handoffs between teams remain a weak spot. The information is usually already captured, and a closed deal can produce a customer success brief in one click. Steps that depend on people, such as how quickly a rep follows up on a new lead, are harder to standardize, and experienced reps don’t change how they sell overnight.

The next step is linking those systems across departments. “When you start rolling AI out, you do it in silos. Each group does their own thing, and you don’t look at the full customer journey. That’s probably the next phase, trying to figure out how we connect the whole organization,” says Habansky.