The phone isn't dying. It's concentrating.
Self-service is absorbing the easy contacts, and what's left rings the phone: complex, urgent, emotional. Voice volume is falling while voice difficulty climbs — and most teams still staff it like 2019.
Every support budget review has the moment: someone points at the cost-per-contact line, notes that phone is the expensive row, and floats the obvious question — can we kill it? The forecast on the slide seems to agree.
Read quickly, that says the phone is on its way out. Read carefully, it says something more interesting: executives expect the ranking of channels to change as self-service and chat absorb volume. Nothing in it says the calls that remain will be like the calls you have today. They won't be — and that difference is the whole operational story.
Self-service is a filter, not a destination
Start with what self-service actually accomplishes. Customers try it, genuinely and in growing numbers. Then:
Fourteen percent fully resolved. The rest of those journeys don't evaporate — they continue, hop by hop, toward a human.
So picture the funnel honestly. Self-service strips out the simple, patient, well-documented problems — password resets, order status, the questions a decent help centre should absorb. What passes through the filter and eventually rings the phone is the residue: the issue too tangled for an article, the customer too anxious for an email thread, the account worth too much to risk on a chatbot. Every call that connects now arrives pre-loaded — a self-service attempt that failed, a customer already two channels deep, a problem that has survived every cheaper way of solving it.
Self-service didn't take the phone's job. It took the phone's easy tickets.
Voice volume can fall for years while voice difficulty rises every quarter. If your average handle time on the phone is creeping up, that may not be a performance problem. It may be arithmetic: the denominator lost its easy calls.
The trust paradox
Meanwhile the channel itself is in a strange state. Customers still rate it as essential:
And yet the same research shows the inbound ring tone has become a threat signal:
Hold both facts at once and the shape of modern voice becomes clear. Customers want to be able to reach a person by phone for the things that matter — and they will not pick up when an unknown number calls them. The channel is inbound-healthy and outbound-broken. A callback your customer requested, at a time they chose, from a number they recognise, is service. The same call out of the blue is, statistically, presumed fraud.
Staff the channel you actually have
If voice is concentrating into the high-stakes tail, running it like a volume channel is a category error. The adjustments are concrete:
- Re-baseline handle time on purpose. A rising AHT on a shrinking, harder call mix is expected physics, not slacking — AHT is a budgeting metric, and the budget just changed. Punishing agents for the filter's arithmetic teaches them to rush your most fragile customers.
- Put experience on the phone. The easy calls that juniors used to learn on have been deflected. What remains needs your best people — judgment, product depth, de-escalation — which also means the phone team is a destination, not an entry-level penance. For the genuinely stuck cases, pull in help rather than passing the parcel: the case for that is in tiers vs. swarming.
- Carry the journey across the handoff. Nine in ten of these customers already tried something else. Making them re-explain from zero is how a hard call becomes a hostile one. The self-service trail — what they searched, what article failed them — should arrive with the call.
- Fix outbound before you rely on it. If callbacks are part of your queue strategy, they must be scheduled, expected, and recognisable. An unrecognised number is a 74%-probability dead end, whatever your intentions.
- Defend the line item. Yes, the phone is your most expensive channel per contact — and it now handles the contacts where the relationship is actually decided. Cutting it to save cost-per-contact optimises the metric and forfeits the moment. The maths of what a contact really costs is in the real cost of a ticket.
The phone isn't a legacy channel with a sunset date. It's becoming the escalation path of last resort and the recovery room for every failure upstream — smaller, harder, and more consequential per minute than it has ever been. Staff it, measure it, and price it like that. Where your handle times and service levels should sit against everyone else's, the benchmarks hub has the reference points.