Telesales ROI: from cost centre to revenue centre

Why telesales sits on the cost line of a budget

In many companies the telesales team lands in the budget as a cost: salaries, telephony, CRM licences, training. The revenue side appears on a different line — in sales. The connection between the two is written down nowhere, and as a result the team's value cannot be defended in an annual budget review.

That is a measurement problem rather than an accounting one. To show what telesales contributes you need two numbers: what one conversation costs the company, and what one conversation is worth on average. Without those two, any automation decision becomes a matter of taste.

This article gives the method for producing both. It deliberately does not give the numbers themselves — they vary so much by sector, city and product that borrowing an «industry average» leads to the wrong decision.

What the cost is actually made of

Telesales cost is usually calculated as salaries alone. The real structure is wider, and its invisible half is rarely smaller than its visible one.

• Direct salaries and taxes
• Telephony: per-minute or per-channel cost, number rental
• CRM and other per-seat tool licences
• Hiring and training: the time from a new operator's first day to being useful
• Turnover: finding and training a replacement for whoever leaves
• Management time: the hours a team lead spends listening to calls and reporting
• Idle capacity: time paid for in which no call is being made

The last two lines are the most commonly ignored. Quality control by listening to recordings takes a serious share of a team lead's day, and that time appears in no report as a telesales cost.

The revenue side: the chain from a call to money

To calculate revenue, write the chain stage by stage. Each stage carries a percentage, and the product of those percentages is the real link between a call and a result.

1. 1Dialled call → real conversationThe first rate: how many dialled numbers end in a conversation. This depends on the quality of the list and the hour you call at.
2. 2Conversation → qualified leadThe second rate: how many conversations produce a fitting lead. The script and the targeting drive this one.
3. 3Qualified → meeting or proposalThe third rate: how many fitting leads move to the next step.
4. 4Meeting → saleThe fourth belongs to the sales team rather than to call automation — but it is part of the chain.
5. 5Sale → average order value and repeat purchaseThe last step produces the money. Leave repeat purchase out and a customer looks cheaper than they are.

Producing these five numbers once is enough. After that you can see which rate any change — a new script, a new list, automation — actually moved.

How to cost one conversation

The arithmetic is simple, provided the denominator is real conversations and not dialled calls. Dividing by dials makes the unit cost look artificially cheap.

1. 1Add up the full monthly costAll seven lines above, for one month. Include idle capacity — hours that were paid for and not spent on calls.
2. 2Count the real conversations that monthCalls whose outcome code says a conversation happened. Unanswered and cut-short calls do not belong here.
3. 3Divide the first by the secondThe result is the cost of one conversation. It is the core unit cost of telesales, and comparisons are made on it.
4. 4Repeat the same sum for a qualified leadDivide the full cost by the number of qualified leads. This number compares directly with the cost per lead of your marketing channels — and that comparison is usually what changes the conversation.

How missed calls enter this sum has to be worked out separately; the method is in how to price your missed calls.

Where automation changes the arithmetic

Call automation does not take the cost to zero. What it changes is the structure of the cost: part of what was fixed becomes variable.

• Headcount is a fixed cost — it is paid whether calls happen or not
• Call minutes are variable — they only exist when a conversation does
• Adding capacity for a peak does not require hiring
• Covering dead hours — evenings, weekends — is not measured in operator hours
• The unit cost of repeatable calls stays flat as the volume rises

The practical consequence: the return on automation shows up most clearly in two situations — when call volume is uneven, and when a large share of calls is repeatable. In a team with steady volume and individually-handled calls the gain is smaller, because most of that cost is variable already.

Three situations where the model breaks

The arithmetic fails in three cases, and it is worth knowing them in advance.

1. 1The list is poorIf numbers are stale or consent was never given, the conversation rate falls and the cost per conversation rises. The problem is the list, not the automation, and the list has to be cleaned first.
2. 2The offer is out of step with the marketIf the qualification rate is normal and the meeting rate is low, the issue is not the calling. Adding calls here adds cost, not revenue.
3. 3Outcome codes are not writtenWith no codes, none of the rates can be calculated and the return on investment stays a matter of opinion. This is the easiest of the three to fix and usually the most common.

A measurement cadence

How often you look matters as much as what you look at, because each number needs a different amount of time to settle.

• Weekly: conversation rate, qualification rate, cost per conversation
• Fortnightly: the effect of a script change
• Monthly: cost per qualified lead, meeting rate
• Quarterly: conversion to sale and customer value — these are noisy month to month

How these are collected and where the reporting is built is shown on the analytics page.

What the cost looks like on the Vexvon side

On Vexvon's call side the cost model follows usage rather than operator hours. The parts worth knowing when you do the sums:

• Balance is counted per second — a call that never becomes a conversation does not cost an operator hour
• There is a concurrent-call ceiling: how many conversations are possible at once is known, so peak capacity can be planned rather than discovered
• One call campaign holds up to 5,000 targets, and preview shows how many will be reached — and which rows have no number — before it starts
• A running campaign can be paused, resumed and cancelled, so a badly targeted campaign is not paid for to the end
• Outcome codes and contact attempts are stored in the CRM, so the rates above are not counted by hand

Which share of calls is worth automating at all is covered in scaling sales without hiring, and building the process itself in sales call automation workflow.

A checklist before the decision

Before deciding on automation, these six answers should exist in writing.

• What one real conversation costs
• How the cost of a qualified lead compares with your marketing channels
• How much call volume swings across a month
• What share of calls is repeatable and what share is individual
• Whether outcome codes are being written today — or whether that comes first
• How many hours a team lead spends listening to calls and reporting

With those six in hand the question stops being «should we automate» and becomes «which part first». To work the numbers through on your own figures, get in touch.