Before sounding the alarm on a rising AHT, it's worth checking whether it reveals a real problem, or simply that selfcare is finally doing its job.
When simple requests shift to selfcare, the math is unavoidable: Average Handling Time (AHT), the average time an advisor spends resolving a customer request end to end, calculated only on contacts that remain with a human, goes up, even though nothing has actually gotten worse. A rising AHT almost automatically triggers alarm bells for a manager used to reading this figure as a drift signal. But a rising AHT has several possible causes, and some of them are a sign that a setup is improving, not deteriorating. This guide breaks down the most common causes of a rising AHT and gives you a method to quickly tell a real warning signal apart from a harmless mechanical effect.
AHT is an aggregated figure. It summarizes hundreds or thousands of very different interactions into a single average. The problem with the instant-alarm reflex is that a rising average never explains, on its own, why it's rising. It can go up because advisors are handling requests less well. It can also go up because the mix of requests handled by a human has changed, without any advisor handling anything worse than before.
This distinction isn't a technical detail. It completely changes what action to take. If AHT rises because advisors are struggling, the answer is training or process work. If it rises because selfcare has absorbed the simple requests and only complex cases are left for humans, launching a corrective action on AHT would mean penalizing a setup that's working exactly as intended.
Selfcare absorbs the simple requests. Simple, repetitive requests typically make up between 60% and 80% of a customer service team's total volume, according to several industry studies. When a well-built selfcare system captures a growing share of them, it's precisely the fastest contacts to handle that disappear from the human scope. What's left, more complex by construction, mechanically pushes the average up. That's a positive signal, not a warning sign: it means advisors are now only handling what genuinely requires human intervention.
The effect is purely statistical, and it's worth visualizing with a simple example. Imagine a center handling 1,000 contacts a day: 600 simple ones at 3 minutes and 400 complex ones at 10 minutes, for an average AHT of 5.8 minutes. If a well-built selfcare system absorbs half of the simple contacts, only 300 contacts at 3 minutes and 400 at 10 minutes remain in human handling. The average AHT calculated on this new scope climbs to 7 minutes, a rise of more than 20%, without a single advisor having handled a single request worse than before. The overall setup improved. The human AHT, read in isolation, tells the opposite story.
Customer context gets richer, and that's intentional. An advisor who now has the customer's full history, a 360° view of their contracts or previous contacts, can handle each request with more depth. That sometimes takes longer than a quick, generic answer, but the extra time invests directly in personalization and perceived quality, two dimensions no AHT figure captures on its own.
Concretely, an advisor who sees the last three contacts before picking up can anticipate the likely reason for the call, avoid making the customer repeat information already given, and offer a solution that accounts for context the customer didn't need to re-explain. This extra reading time, which mechanically adds to AHT, reduces in return the chance the customer calls back for the same reason. It's a trade of time for quality, not a loss of efficiency.
Average case complexity genuinely increases. A regulatory change, a new product launch, a period of heavy news coverage on a sensitive topic: these events increase the share of cases that require long, careful handling. A rise in AHT linked to this factor is temporary and fades naturally once the period passes, as long as teams aren't pushed to rush handling just to bring the number down artificially.
A concrete example helps. When an insurer changes its terms and conditions or an energy provider adjusts its pricing grid, the following weeks bring a wave of questions that require the advisor to explain the change, sometimes reassure a worried customer, before even addressing the original request. AHT climbs during this specific period, then gradually comes back down once customers get used to the change and advisors' knowledge base stabilizes on the new topic. Reacting in the moment by pushing teams to go faster often prolongs the period of confusion rather than shortening it.
AI increases handling time for some contacts, not all. An AI tool that surfaces more context or suggests more complete actions can slightly lengthen certain interactions, especially during the adoption phase when advisors are still learning to use this new information efficiently. This rise, if it stays temporary and affects the topics where AI genuinely adds value, isn't a sign of a problem but a normal learning phase.
A process or training issue, the only real warning signal. An outdated knowledge base, a clunky tool, a team undertrained on a new product: these causes push AHT up with no upside at all. It's the only one of the five situations that genuinely calls for quick corrective action.
The good news is that telling these causes apart doesn't require a sophisticated tool. Three simple checks are enough in most cases.
Compare AHT by contact reason, not as a global average. If AHT rises overall but stays stable or falls within each reason, that's a composition effect: the contact mix changed, not performance on each type of request. If it also rises within each reason taken separately, that's a real signal worth digging into.
Systematically cross-check with selfcare volume over the same period. A rise in human AHT that coincides with a rise in the self-resolution rate on digital channels almost always confirms a transfer effect rather than a decline in performance. It's the fastest check to put in place, and the one that prevents the most false alarms.
Look at FCR before concluding anything. FCR, or First Contact Resolution, measures the percentage of requests resolved on the first contact, without the customer needing to call back for the same reason. An AHT that rises with a stable or improving FCR is never a problem: advisors are simply taking more time to resolve things better. That's exactly the logic behind the AHT-FCR cross-matrix detailed in our article on balancing AHT and FCR [link to be added to the pivot article "Beyond AHT" once its publication URL is known]. An AHT that rises alongside a falling FCR is, by contrast, the signal that justifies a genuine alert.
| What you observe | What it means | Action |
|---|---|---|
| Global AHT rising, AHT stable by reason | Composition effect, contact mix changed | No corrective action needed |
| AHT rising, selfcare rate rising | Successful transfer of simple requests | Document and communicate the positive result |
| AHT rising, FCR stable or rising | Deeper handling, not unnecessarily slower | Monitor, no immediate correction |
| AHT rising, FCR falling | Real signal of declining process or training quality | Audit causes before correcting symptoms |
In a relationship with a BPO provider, a rising AHT is often the first point raised in a steering committee, because it shows up immediately on a dashboard. Without this reading grid, a company can end up blaming its provider for a rise that actually reflects a successful transfer of load to selfcare, or an upgrade in the customer context advisors are working with.
A good provider anticipates this reading by systematically presenting AHT alongside its context: how the reason mix has evolved, the transfer rate to digital channels, and the associated FCR. This discipline of presentation, more than AHT itself, is what lets a steering committee make the right call instead of reacting to the first number that moves. For the full structuring of these indicators in an outsourcing contract, our guide on how to measure the performance of an outsourced contact center covers the complete method.
No. It can reflect a successful transfer of simple requests to selfcare, richer customer context being used by advisors, or a temporary spike in case complexity. The only situation that justifies a real alert is a rising AHT combined with a falling FCR (First Contact Resolution, the rate of requests resolved on the first contact).
By comparing AHT by contact reason rather than as a global average, and cross-checking it with how the selfcare resolution rate and FCR evolved over the same period. If AHT by reason stays stable and selfcare is growing, it's a composition effect, not a decline in performance.
Not systematically, and not permanently. Some AI tools reduce AHT on repetitive tasks like post-call summaries. Others, which enrich the context shown to the advisor, can temporarily extend it during the learning phase, before stabilizing or decreasing once teams are fully trained.
It's better to set AHT targets by contact reason rather than a single global ceiling. A single target applied to every type of request mechanically pushes teams to rush complex cases just to hit a number that doesn't make sense for that type of contact.
By systematically presenting AHT alongside its context: how the handled reason mix evolved, the transfer rate to selfcare, and the associated FCR rate over the same period. This presentation avoids reacting to the raw number and allows a discussion about real causes rather than the average alone.
A rising AHT doesn't automatically call for corrective action. It first calls for a question: has the contact mix changed, or has performance on each type of contact actually declined. Three simple checks, AHT by reason, the selfcare transfer rate, and the associated FCR, answer that question in a few minutes, and prevent you from correcting a setup that's actually working very well.
At Armatis, AHT trends are always presented alongside their contact-mix context and associated FCR, so every variation reads as an understandable signal rather than an isolated number. Discover how Armatis manages contact center performance.
Sources
Armatis is a European specialist in customer relations and business process outsourcing (BPO), operating across multiple continents with thousands of employees serving companies of all sizes and sectors. The company designs and manages end-to-end customer service operations: multichannel contact centres, complaints handling, technical support, back-office and digitised processes. Backed by integrated technology infrastructure and the ability to adapt to any sectoral and regulatory context, Armatis helps its clients combine operational performance, quality of experience and cost control, wherever they need it.
Contact our teams to discuss your challenges and find out how we can support you
Join the leaders who trust our multilingual and technological expertise.