Support has never been short of metrics: response time, resolution time, ticket volume, CSAT, deflection, and now, increasingly, AI adoption. These numbers tell us a lot about how efficiently a support organization operates. But they don’t always tell us what happens when a customer issue gets complicated, when the answer isn’t in the ticket, when another team has to step in, when a rep has to hunt across systems before making a decision.
Looking beyond the usual numbers
From a decade of working closely with high-stakes B2B support teams, we kept noticing the same thing: teams were confident about their operational numbers, far less certain whether the understanding a rep builds about a customer actually survives everywhere it needs to go. The observations became the starting point for our latest benchmark study.
We surveyed over 500 B2B support professionals, from senior support managers to VPs of customer support, across companies of every size. We asked how teams define resolution, what a rep can see the moment a ticket lands, how support collaborates across functions, how its work connects to revenue, where AI is headed, and what’s changing for the people doing this work. Our goal wasn’t another leaderboard of who responds fastest. It was to see what happens to customer understanding as it moves through people, tools, and teams.
When we pieced the answers together, the picture that emerged was of an industry that is surprisingly not short of customer information. Only 9% of support leaders we surveyed say a rep sees nothing about an account when a ticket lands, without opening another tool. Context, in other words, is rarely missing. It’s almost always in motion, and how well it survives that motion turned out to be the real story.
The number every team tracks
Start with the number every B2B support team already tracks: deflection rate. We found that 67.5% of teams track it in some form, but only 24% call it their most important metric.
The clearest evidence that deflection alone isn’t enough showed up when we asked a different question: is a deflected ticket the same as a resolved one? 37.5% of the leaders we surveyed said no, it can simply mean the customer stopped asking. A conversation not reaching a human doesn’t reveal whether the customer actually got what they needed, solved it themselves, gave up, or quietly took the issue somewhere else.
However, that does not make deflection the wrong number to track. It’s an important signal of how well automation is working, just an incomplete one on its own, for organisations where resolution regularly depends on people and processes the ticket system was not built to capture. The metric should not simply be whether a ticket was deflected, it should be whether the customer’s problem was resolved.
What reps see and what they have to go find
The gap gets more interesting once we look at what a rep can actually see when a ticket needs attention. Close to 40% can see conversation history and open issues the instant a ticket lands. That layer is fairly mature. But 66.8% of complex tickets require a rep to check three or more tools before they can act with full information, because the pieces that actually tell you how much a ticket matters [account health, CRM details, contract value] sit elsewhere. The ticket lives in one system, and the account, and everything that makes it worth understanding, lives across several.
So a rep assembles what they can, on a live call, often in seconds. The harder question is what happens to that picture the moment it has to leave their hands. B2B support rarely resolves anything alone. A billing dispute needs finance. A tricky feature request needs product. A frustrated customer is often already on customer success’s radar. When a case does need another team, we found that 37.6% of B2B support teams say context gets lost frequently or almost always in that handoff. Only 3.1% say it never happens. Coordination itself, more often than not, happens through Slack, email, or a phone call rather than the systems built to keep a record of it. The ticket has a clean, trackable path from one team to the next. The understanding behind it doesn’t have the same guarantee.
Does this reach the people making business decisions?
Which raises the obvious question: if that intelligence struggles to survive a handoff between two teams, what happens when it needs to travel all the way up to the people making decisions about the business? Support sees trouble before almost anyone else does, a pattern of complaints, a shift in tone, a customer gone unusually quiet, often weeks before it becomes a renewal conversation or a churn number. Most organisations already sense this is valuable: we found that 79.1% either already connect support outcomes to revenue, or are actively building toward it.
But we also found that revenue impact of support, the metric that ties support directly to commercial outcomes, is what leadership asks about least, at just 9%. The groundwork existing and the habit of using it, it turns out, are two different things.
Someone is carrying the context
Behind every one of these gaps sits the same quiet fact: somebody is currently closing them by hand. Someone is assembling the scattered account picture on a live call. Someone is deciding what to type into a Slack message so the next team doesn’t start from zero. It’s probably not a coincidence, then, that 29.2% of the support leaders we surveyed name retaining experienced reps as their single biggest challenge, ahead of AI upskilling.
What that judgment actually looks like, what it takes to build it, and what it costs an organisation to lose it, is where the story gets more interesting than a single statistic can capture.
That question of what’s carrying B2B support right now, and whether it can keep carrying it as AI takes on more of the work, is the one we set out to answer in full.
Context in Motion goes deeper into all of it. It looks at what resolution really means once deflection stops being the whole answer, and what survives a handoff and what doesn’t. It also gets into how support can make its case to leadership, and what the next few years are likely to ask of AI, and of the people still making the final call.
