FreshBooks and AI Assist Pro: 5,000 calls reviewed in 4 months
What happens when you have one manager responsible for manual processes to review and evaluate calls? Coaching that is difficult to be applied at-scale. Frustration, usually. While the whole team tries to maintain quality in conversations with prospects, manual feedback can often inhibit rep development. FreshBooks has always prided itself on quality in sales calls, but with the volume of calls needing review, managers often found themselves unable to scale their coaching efforts. So they turned to Aircall’s AI Assist Pro to help automate the administrative side of the process and free up managers to focus on their team in the moment.
Introducing FreshBooks
FreshBooks is a leader in small business management software, helping its customers simplify the financial complexities of running their businesses. The company’s core market is North America, and they serve customers in over 160 countries.
We caught up with Simon Kelly, Director of Customer Success at FreshBooks. He leads a team of Product Consultants and Onboarding Specialists, providing direct coaching to improve the quality of conversations on every call with prospects and customers.
"Our rep calls are consultative. They cover a lot of ground, and the quality of the conversation genuinely determines whether that person ends up with the right FreshBooks product for their business. That's why call quality is something we take seriously and built a robust program around."
- Simon Kelly, Director of Customer Success, FreshBooks
FreshBooks reps mostly speak directly to business owners, so every conversation carries real weight and impact.
The challenge: Reaching consistent quality call coaching at scale
Before AI Assist Pro, call quality assurance was entirely manual; managers conducted live call shadowing, pulled a handful of recordings for ad hoc review, and provided feedback during weekly one-on-ones. Every rep had to write their own post-call recap, log notes in Salesforce, and draft follow-up emails largely from scratch.
"FreshBooks has a high-velocity sales team, and we do high-volume outreach, but we couldn't coach at the scale that we're selling at."
Low-volume coaching wasn’t the only constraint. The team was experiencing several other challenges with this model:
Coaching relied on random samples. Managers used to review a handful of calls per rep, with no way to know if that snapshot reflected a real pattern.
Standards varied by manager. Each manager weighted factors like discovery questions and objection handling differently, so reps received inconsistent feedback.
Call feedback took too long. By the time a rep heard feedback, they could have forgotten details of the conversation due to the number of daily conversations they have.
Tenured reps develop blind spots. Experienced reps settle into habits that work for them. Without unbiased scoring, managers needed an efficient way to point out improvements.
Admin swallowed too much time. Managers and reps lost hours to call review and post-call admin that could have gone toward prospecting instead.
This model also meant they were being reactive to their coaching gaps. Sometimes, product areas would get skipped in calls, or reps would struggle with objections, but issues like these wouldn't surface until they had already hit the pipeline.
And for customers, inconsistency in call structure and content was a common flaw for newer reps. Some reps would gravitate towards specific product areas, while others applied a more broad approach.
The turning point: Pilot testing with fast results
FreshBooks was already using Aircall as its telephony platform, but the team researched other coaching tools before settling on AI Assist Pro.
“The more we looked, we ran into the same issue: a separate layer sitting on top of our existing system, another tool reps would have to log into, another app to maintain,” said Simon. “Our reps were already comfortable with Aircall and making all their phone calls from the platform.”
Additionally, Simon noted that a full conversation intelligence platform is a significant commitment to make before knowing whether it would actually change rep behavior.
“An add-on to a platform we'd already deployed let us test it out,” he said. “We were able to move from hypothesis and a few different testing stages to actually coming to a confirmed, confident answer that we were making the right call using AI Assist Pro.”
The solution: End-to-end AI-powered call coaching
Before rolling AI Assist Pro out, FreshBooks had a small group of reps try it firsthand to see what value it could offer. They used this pilot to gather data and shape scorecard configuration, based on what the reps themselves considered a good or bad call example.
Since rolling out completely, AI Assist Pro has made an impressive impact in a number of areas:
Instant call scoring: Automated scoring gives reps a clear read on their own calls the moment they hang up. For example, flagging the exact spot where a yes/no question should've been open-ended, so they can self-correct immediately.
Live in-call prompting: Real-time guidance surfaces right in the middle of a conversation, so reps know how to handle common objections.
Standardized scorecards: FreshBooks configured scorecards using feedback from the pilot group, shaping the criteria around what reps and managers actually considered a strong call.
Summaries synced to Salesforce: AI-generated call summaries flow automatically into Salesforce, giving reps a head start on next-step notes and follow-up emails instead of writing recaps from scratch.
These features equip the team with a coaching layer built right into the phone system they already use, within Aircall. As a result, the team started seeing results almost immediately, rather than waiting on a lengthy platform rollout.
The results: Sharper coaching, full volume call coverage
Coaching scaled to nearly 5,000 evaluations in 4 months
FreshBooks went from manually listening to a handful of calls per week to full coverage of every call, whether through a formal manager evaluation or a rep pulling their own immediate scoring.
"Since April of this year, we ran close to 5,000 scored call evaluations, almost 2,000 on inbound and just over 3,000 on outbound,” said Simon. “This volume was never available to us when we were doing manual call scoring, simply because there's only so many evaluations a manager can do in a day."
Faster rep improvements from proactive self-coaching
Reps can automatically view evaluations on their own, which launched a genuine culture shift toward self-coaching. Instead of waiting for their managers to flag an issue, reps now review their own calls to address any weak spots and apply adjustments to their next calls.
Higher quality manager feedback
One-on-ones are now grounded in evidence instead of relying on memory. Managers can pull an exact moment from a transcript and show a rep how it was scored. This specificity allows reps to see where improvements can be made in the conversation, making the most granular changes far more likely to stick.
Reps send follow-ups immediately after the call ends
Post-call admin time has significantly dropped. With reps making 40-60 touch points a day, follow-ups had to wait until the rep had free time, which could be delayed during busy periods of back-to-back calls. Now, call summaries feed straight into templates, letting reps fire off a next step immediately after the call ends.
Consistent coaching for every rep
Every rep now has the same standard of scoring across the board.
More efficient coaching from the beginning for new reps
With live prompting creating a built-in structure to lean on, new reps have a sense of security and confidence while they’re still learning to navigate conversations on their own.
FreshBooks hasn’t replaced its live call shadowing. AI Assist Pro is weaved into a human process that gives managers and reps the ability to cover more call evaluations and make strategic changes faster.
Looking ahead: More AI Assist Pro
FreshBooks plans to continue iterating on scorecard design as it builds up a larger data set over time. They also want to dive deeper into conversational trend analysis to build out sales enablement content.
The team is looking forward to expanding their use of Ask AI Assist. Currently in beta, Ask AI Assist lets supervisors and admins ask questions about their call data, transcripts, and team performance without building reports or running manual searches. FreshBooks wants to use this feature beyond sales, so teams like marketing and product can validate customer feedback across a larger, statistically significant set of calls.
Advice to others: Start with the problem, not the tool
"Teams that buy an AI tool first and go looking for the value afterward will probably be disappointed,” said Simon. “Start by understanding the problem you’re trying to solve.”
He also advises piloting small. FreshBooks trusted a handful of reps to run the testing period with AI Assist Pro, then used real data from the group to justify the purchase. So they brought real, human experience and expertise to the build of their AI platform. Simon notes it’s the ideal time to invest in scorecard design to make sure criteria is fair and actually guiding reps toward the right behaviors.
Lastly, Simon encourages others to bring reps in early to the process and remain transparent about intent. He reassured his team that AI scoring wouldn’t mean they were on their own: reps would have immediate, in-the-moment takeaways they can act on with the very next customer, but real-human feedback wasn’t going anywhere.
“Don't take the human out of it,” he said. “Live call coaching is always going to be part of our culture and our sales process here."
Ready to help your team coach at scale? See how Aircall's AI Assist Pro can bring instant, consistent call scoring to every conversation your team has.