There’s a number that almost no service business measures honestly.
It’s not the dashboard abandon rate – that one is sanitized, defined narrowly, and usually optimized to look acceptable. It’s the real number. The customers who tried to reach you and gave up. The ones who waited two minutes on hold and decided not to wait three. The ones who hit the IVR menu out of hours and put down the phone. The ones who called twice, got routed wrong both times, and went to a competitor on the third attempt.
Most operations teams can’t quote that number because their reporting doesn’t capture it. The phone system records calls that connected, calls that completed, calls that hit voicemail. It rarely records the customer who never made it through the front door. Those customers are invisible, and they’re the most expensive ones you have.
This post is about why that number is so consistently underestimated, what it actually costs, and what changes when the front door of your phone system can answer at scale.
The pattern, across industries
We’ve spent the last three years embedded in two very different customer environments — one in healthcare, one in telecom. The operations look nothing alike. The call abandon pattern looks identical.
In both environments, the pattern goes like this. A customer calls outside business hours, or during a peak that exceeds staffed capacity. They hit a menu tree. They navigate it imperfectly. They get routed to a queue. The hold time exceeds their patience. They hang up.
What happens next depends on the industry. A telecom customer calls back later, or moves the issue to chat, or churns. A healthcare patient calls another clinic, or ends up in the emergency room with a problem that should have been triaged earlier, or simply doesn’t follow up. A financial services customer escalates to social media, or calls their relationship manager directly, or accepts a degraded service experience and remembers it at renewal time. An insurance customer disputes the claim later instead of resolving it on the first call. A government services applicant misses a deadline because the call back never came.
In every case, the customer’s underlying need didn’t go away. It just got more expensive to serve — for the company, the customer, or both.
The four hidden costs
When operations teams do start measuring this seriously, four costs emerge that don’t show up on any standard contact center dashboard.
The first is escalation cost. A customer who can’t reach you on the phone often ends up in a more expensive channel. In healthcare, that’s the emergency room. In financial services, that’s the relationship manager or the branch. In telecom, that’s the retention team or the regulatory complaint queue. In insurance, that’s the claims dispute process. Every one of these channels costs more per interaction than a phone call would have. The cost is real; it just lands in a different P&L line, which is why operations teams rarely see it.
The second is staff opportunity cost. When the front door of your phone system can’t handle volume, the burden shifts to the people who can. In healthcare, senior nurses and clinicians spend a meaningful fraction of their shifts on telephone intake. In telecom, agents trained to handle complex retention conversations spend their day confirming bundle activations and answering balance inquiries. In any service business, your most expensive humans are the most likely to get pulled into the most routine calls. The opportunity cost is what they could have been doing instead.
The third is reputation cost. Customers who can’t reach you talk about it. They talk to their networks. They post reviews. They mention it in NPS surveys. They remember it at renewal. The reputation cost is invisible until it isn’t, and by the time it shows up in churn data or in customer acquisition cost, it’s already a quarter or more behind the operational decision that caused it.
The fourth is regulatory cost. This one is industry-specific. In healthcare, missed access can become a quality-of-care issue. In financial services, it can become a complaint that triggers regulatory scrutiny. In telecom, sustained access failures attract regulator attention in markets with consumer protection mandates. In insurance, claim handling timelines are often regulated. In every case, the cost of regulatory attention vastly exceeds the cost of answering the original call.
These four costs share a common feature: they’re real, they’re measurable in retrospect, and they’re almost never visible to the operations team that could prevent them. The team that runs the phone system is measured on cost-per-call and average handle time. The costs of the calls that didn’t happen show up somewhere else entirely.
Why the existing tools don’t help
If unanswered calls are this expensive, why hasn’t the contact center industry solved them already?
The honest answer is that the existing tools were designed for a different problem. Interactive Voice Response systems — the menu trees that dominate the front door of every phone system — were built in an era when their job was to triage customers efficiently into the right human queue. The success criterion was getting customers to the correct department; volume that exceeded capacity was a workforce planning problem, not a technology one.
That model worked for a long time. It started failing when customer expectations shifted. Modern customers compare your phone experience to a self-service mobile app, not to the contact center industry baseline. A 90-second hold is intolerable in a way it wouldn’t have been ten years ago. A menu tree that doesn’t understand “I want to schedule a follow-up appointment” feels broken in a way it didn’t when “press 2 for appointments” was the state of the art.
The industry’s response has been incremental. Better hold music. Callback options. Chat overflow. Workforce management software that smooths peak loading. Each of these helps at the margins. None of them changes the fundamental dynamic: the front door of the phone system is a fixed-capacity human bottleneck, and exceeding capacity costs you customers.
The voice AI products that came out of the 2023–2024 cycle promised to fix this, and many of them can — in a controlled demo. Where they’ve struggled is in the deployment economics. Replacing the IVR layer with a voice AI usually meant replacing the contact center platform around it, which usually meant a multi-quarter migration that the procurement committee killed.
The deployment model that actually works is one where the AI can answer the call without anything else having to change.
What “always answers” looks like in practice
OpenBrain handles the front door of the phone system. The customer calls the same number they’ve always called. They speak in their own language. The bot understands what they need — appointment scheduling, balance inquiry, claim status, service activation, account question, whatever the use case is — and either resolves it end-to-end or routes the customer to a human agent with full context already loaded.
In our healthcare deployment, the bot handles thousands of patient interactions a month, in the local language and in English, with other languages added as the patient population requires. It schedules appointments. It answers questions about procedures and recovery using a knowledge base sourced directly from the clinic’s medical documentation. It hands the call to a clinician the moment the conversation needs one — with the patient’s intent and history already loaded onto the clinician’s screen.
In our telecom deployment, the bot handles inbound volume across prepaid bundle activations, balance inquiries, simple service questions, and routine account changes. The agents who used to spend their day on these calls now handle the conversations where their expertise actually matters.
The metrics that matter, six months in:
- 100% answer rate. Every inbound call is answered, including overnight, on weekends, and during peak hours. There’s no abandon rate at the front door, because there’s no queue to abandon.
- Roughly 60% of calls resolved end-to-end by the AI. The other 40% reach a human agent — but with the customer’s intent, context, and conversation history already in front of them.
- Reduced staff time on routine intake. Specifically how much depends on the use case mix, but in our healthcare deployment, clinical staff time spent on telephone triage was reduced by approximately half.
- No infrastructure migration. The deployment ran on the existing PBX, the existing SIP trunks, the existing numbers. The supervisor team kept their existing console and added a parallel one for the AI conversations.
The customer experience that produces these numbers is straightforward: the customer calls, the bot answers in their language, the conversation goes one of two ways — fully resolved, or warm-handed to a human. That’s the entire user-facing change. Everything else is the operational work of getting from where most contact centers are today to a phone system that always answers.
What to do about it
If this post resonates — if the pattern of customers who couldn’t reach you sounds familiar — there are three things worth doing this quarter, regardless of whether you ever talk to us.
First, measure the real number. Most contact center reporting underestimates true abandon by a factor of two or three. Get your telecom provider to pull call attempt data — every inbound attempt, including the ones that didn’t connect to your queue. Compare it to the calls your contact center platform recorded. The gap is what you’re not seeing today.
Second, measure the escalation pattern. For a sample of customers who had a known service issue, trace what channel they ended up using to resolve it. The ones who escalated to a more expensive channel are telling you something about the cheaper channels’ capacity. That signal is in your CRM data already; it just isn’t usually pulled.
Third, look at the staff opportunity cost honestly. Pick the most expensive ten people in your customer-facing operation. Sample what fraction of their week is spent on calls that a well-designed self-service or AI flow could have handled. The number is usually surprising.
Once you have those three numbers, the business case for changing the front door of your phone system writes itself. We can help with the technology side. The measurement side has to start internally.
