How much revenue do missed calls really cost?

The short answer

A missed call costs revenue only when three things are true at once: the caller never reaches you again in time, the call was a real sales opportunity, and it would have turned into an order. Skip the first two checks and count calls instead of callers, and in the example below the estimate comes out more than nine times too high.

An honest estimate multiplies five numbers, and you can measure each one or at least give it a range. They are unique missed callers, the share you never recover, the share that were real opportunities, your conversion rate and your average order value. Service calls, with their repeat contacts, complaints and customers who leave, belong on a separate line and deserve more caution.

In the example business below, where every figure is an assumption, that method gives €9,000 to €49,500 of lost revenue a month, with €24,000 in the middle. Counting every missed call as a €1,500 sale would give €900,000 a month. The distance between those two answers is the reason to do the arithmetic.

The formula, step by step

Lost revenue per month = unique missed callers × share never recovered × share that were sales opportunities × conversion rate × average order value

1. Unique missed callers. People who called and reached nobody who could help: the call rang out, went to voicemail or got a busy signal, or the caller hung up in the queue. Count people, not attempts, and remove spam and internal calls.
2. Share never recovered. Callers who did not get through on a later call and whom nobody called back within a window you define, for example the same working day for a quote request. A caller who reached you later, or ordered online instead, is not a lost caller.
3. Share that were sales opportunities. New inquiries, quote requests, orders and bookings. Existing customers with a service question, suppliers, job applicants and wrong numbers are not sales opportunities, even when they matter for other reasons.
4. Conversion rate. The share of answered phone inquiries of the same type that became an order in the same period, taken from your CRM. Use the team's rate, not your best seller's.
5. Average order or deal value. Revenue per converted inquiry. For lost gross margin instead of lost revenue, multiply the result by your gross margin percentage. Add repeat purchases only if your CRM shows them for customers who first came in by phone.

A worked example with three sets of assumptions

The business below is invented to show the method. Every number is an example assumption, not a benchmark and not DRING data. The company sells and installs heating and cooling systems, and one phone number takes both sales and service calls.

In one month the phone system logs 3,000 inbound calls, and 600 of them are missed. Of those 600, 180 are repeat attempts from callers already missed earlier that month, and 20 are spam or sales pitches to the company. That leaves 400 unique missed callers. The average order is €1,500 and the gross margin is 30%.

Each column moves every assumption in the same direction, so the low and high cases are outer limits, not likely results. Over a year, the middle case comes to €288,000 of revenue, or €86,400 of gross margin.

The high case is five and a half times the low case. Most of that spread comes from two inputs: the share never recovered and the share that were real opportunities. Both can be measured rather than guessed, and that is where your own data helps most.

The service side, kept separate

Not every important call is a sale. Suppose, in the same example, 120 of the 400 missed callers were existing customers with a service question. Their cost shows up in three ways, and only the first is easy to price.

• Repeat contacts: the extra calls, emails and chats they make before the issue is handled. If they generate 150 extra contacts at a handling cost of €5 each, that is €750 a month (example figures).
• Complaints: count the complaints and reviews that mention not getting through. Track the count, but do not convert it into revenue.
• Churn risk: in the middle case, 40% of the 120, or 48 customers, never got through and were never called back. Follow that group in your CRM for a fixed period, such as 90 days, and compare their cancellations or reorders with customers whose calls were answered. Put a value on churn only if that gap appears in your own data.

Keep these lines next to the sales estimate, not added into it. A finance reviewer is more likely to trust a model that labels its weakest inputs.

How to get your own numbers

Most of the data already exists, usually in four places:

• Phone system or PBX call logs: every call with its time, caller number, duration and result, such as answered, no answer, busy or voicemail.
• Queue or abandoned-call reports: calls that hung up while waiting, and how long they waited first.
• CRM: which numbers belong to customers or open opportunities, the conversion rate for phone inquiries and the order values.
• Callback records: who was called back, when and with what result. If callbacks live in notebooks or on personal phones, that is a finding in itself.

Use at least four full weeks, ideally a normal month without a holiday or a one-off campaign, and note whether your business has a season. Then clean the list:

• Count unique numbers per period, not attempts. Five tries from one person count as one missed caller.
• Remove spam, robocalls, internal calls and wrong numbers.
• Do not drop after-hours calls as a group. Remove only those that are not business, such as wrong numbers at night. Keep the customers who called after closing; they are real demand, and you will want them as a separate segment.
• Put very short hang-ups of a few seconds in a separate bucket. Many are misdials, but check a sample before you exclude them.

Missed calls leave no recording, so you cannot hear what those callers wanted. Use answered calls from the same hours and weekdays to estimate the reason mix, and read voicemails where they exist. Then match each missed number against later answered calls, callbacks and orders through other channels within your window. Whatever is left is your never-recovered share.

The evidence is often already on file. In one human-run logistics operation that DRING analyzed, six months of the company's own switch data showed that roughly 40% of calls were never picked up.

Common mistakes that distort the number

• Counting every missed call as a lost sale. In the example, 600 missed calls × €1,500 is €900,000 a month. Even with the 25% conversion rate applied, it is €225,000, more than nine times the middle case. Many missed callers try again, reach you another way, or never wanted to buy.
• Ignoring repeat callers. In the example, 180 of the 600 missed calls were repeat attempts. Even after the 20 spam calls are removed, counting calls instead of callers starts the model at 580 instead of 400 and inflates every later step by 45%.
• Ignoring the time of day. Missed calls are rarely spread evenly. Look for clusters around opening time, lunch, the last hour before closing and evenings. A monthly total hides that pattern, and the pattern decides the fix. Missed calls at 10:00 on a weekday point to staffing and queue design; missed calls at 20:00 point to after-hours coverage. Chart missed calls by hour and weekday before you choose anything.
• Using revenue where margin belongs. €24,000 of revenue is €7,200 of gross margin at 30%. Any fix is paid for out of margin.
• Ignoring the cost of the fix. No fix recovers every lost caller. If a fix wins back half of the callers you now lose, the middle case recovers 8 sales, €12,000 of revenue and €3,600 of gross margin a month. That €3,600 is the most the fix can cost per month to break even, including the team time needed to run it. If won-back callers buy less often than callers who got through the first time, use a lower conversion rate for them.

How to close the gap

Missed calls fall into three situations, and each has its own fix: peaks during business hours, calls after closing, and the calls that are still missed.

Peaks during business hours

When calls arrive faster than the team can answer, callers wait and some hang up. On one human-staffed B2B distribution line, DRING measured a median wait of 54 seconds before anyone spoke, across a random sample of 70 calls.

The options are staffing to the peak, routing overflow to another team, or a first answer that does not depend on a free person. Your abandoned-call report shows how long callers waited before they hung up; that is the wait your peak plan has to beat.

After hours

Decide what "covered" means before you choose a tool. It can mean answering every call, capturing the reason and a callback number, booking an appointment or resolving a known request. Each is a different promise with a different cost. Our guide to after-hours call center coverage walks through where to draw those boundaries.

Fast callback

Some calls will still be missed, and a callback recovers them only if it is fast, owned and recorded. Set a callback window for each call type, give the callback list an owner, cap the attempts and write each outcome to the CRM. Next month you can then measure the never-recovered share instead of estimating it.

When callbacks run at volume, Reach, DRING's outbound calling product, places the calls inside the calling windows and attempt caps you set. Each result, such as no answer, busy or reached, follows its own retry rule.

Where an AI agent fits, and where it does not

An AI agent is one way to answer at peaks and after hours without staffing for the busiest hour of the week. On a customer support line, a DRING agent answers, finds out why the customer is calling and handles what it is allowed to handle. When someone has to decide, it passes the call to a person with a written summary.

It also closes the data gap described above. A call that used to leave only a number in the log now leaves a reason, an outcome and a next step. Call analytics turns those records into the reason mix and recovery rate that you otherwise have to estimate.

It does not fit everywhere. Price negotiations, sensitive complaints and decisions that need judgment belong with a person, and the agent's job there is a clean handover. DRING's current production workloads average about 72% resolution. That is a platform-wide average, not a forecast for your line.

So model the AI case with its own assumptions for how many lost callers it wins back and how often they buy. Start with conservative values and replace them with measured results after the first weeks. And if one question drives most missed calls, such as where an order is, fixing the cause may be worth more than answering it faster.

Whatever you choose, the test is the same: the margin you recover each month against what the fix costs each month, both calculated from the numbers you measured.