Meet Kavya, an invented customer of an invented insurance company. Her two-wheeler policy renewal failed on the payment page on a Sunday evening. She did what most people do. She retried twice, opened the chat widget, got a bot that asked her to rephrase, then found a phone number. On the call she gave her policy number, her date of birth and the registered mobile number. The agent was polite, apologised sincerely, and transferred her to the payments team, where she gave the same three details again. The payment was finally processed on Monday afternoon, after a callback.
Two days later a survey arrives: “How satisfied were you with the agent who helped you?” Kavya clicks 5 out of 5. The agent was lovely. Then a second question: “The company made it easy for me to handle my issue.” She picks 2 on a scale of 1 to 7. That second question is Customer Effort Score, and the gap between those two answers is the reason it exists.
What CES actually asks
Customer Effort Score measures how much work a customer had to do to get something resolved. The idea was popularised by research from the Corporate Executive Board, published in the Harvard Business Review in 2010 under the title “Stop Trying to Delight Your Customers”. Their argument: in service interactions, reducing customer effort does more to build loyalty than exceeding expectations does. Versions of the question vary. An early form asked how much effort the customer had to put in; a later one, often called CES 2.0, asks for agreement with “The company made it easy for me to handle my issue”, on a seven-point scale from strongly disagree to strongly agree. Different companies use different scales, so always check which one a number refers to.
Returning to Kavya. Notice that her CSAT was high because satisfaction measures a feeling about an interaction, and the interaction with the agent was good. Effort measures the whole journey: the failed payment, the bot, the repeated details, the transfer, the wait. The agent controlled only a slice of that.
Doing the arithmetic
Say 20 customers answer the seven-point question after a service interaction. Six give 7, five give 6, three give 5, two give 4, two give 3, one gives 2 and one gives 1. The mean is 5.2. The share who agree (5, 6 or 7) is 14 out of 20, which is 70 per cent. Both are valid ways to report CES; some organisations use the average, others the percentage agreeing. They tell slightly different stories, so pick one, define it in writing, and keep it fixed. A shift from average to percentage in mid-year will look like a miracle or a disaster for no real reason.
With 20 responses, one customer is 5 percentage points. That is a very small sample, and a team leader reading a daily CES for one agent is reading noise. Use CES on volumes large enough to be stable, and slice it by journey or issue type, not by individual agent.
Where effort hides
Back to Kavya’s Sunday. Each step added effort, and almost none of it shows on the usual contact-centre dashboard:
- The failed payment itself, which generated a contact that should never have been needed.
- The bot dead end, which cost her time and goodwill before she reached a person.
- Repeating identity details to a second agent after a cold transfer.
- Waiting overnight with no status update.
Several of these relate to metrics covered elsewhere. A repeat contact is effort. A transfer is effort. A customer who switches from chat to phone is paying a tax. Average handle time might even look good for Kavya’s calls, because the agent was quick. Efficiency of the agent and ease for the customer are different things, and sometimes they pull opposite ways. An agent who rushes through a call to protect handle time may leave out the one sentence that would have saved a second call.
What a team can do with the score
A bare score does nothing. The useful part is the follow-up: ask low scorers an open question, “what made this difficult?”, and tag the answers. After a month you may find that “had to repeat information” accounts for a third of the low scores. That is an actionable finding: pass authentication details along with the transfer, or let the receiving agent see the verification status. The fix lives in the process or the tool, not in agent coaching.
Typical effort reducers include fewer hand-offs, consistent answers across channels, proactive status messages, a single owner for the issue, and self-service that actually works for the common cases. Notice that many of these sit with the client, not the BPO. A BPO that tracks CES and reports effort drivers with evidence becomes more useful to the client than one that only reports its own service level.
Where CES falls short
It is not a universal metric. CES fits transactional, issue-resolution situations. It fits poorly for a sales call or a purely relationship moment, where the customer’s goal was not to fix something. A low-effort experience is also not automatically a pleasant one; some customers want a human to listen, and the lowest-effort path can feel cold. And like any survey, it hears only from people who respond. Customers who gave up silently never answer.
Be careful about tying agent incentives to CES. The score depends on things an agent cannot control: product failures, policies, system speed. Linking pay to it invites arguments or pressure on customers to give high ratings. Use it as a diagnostic for the process, not a stick for individuals.
As for Kavya, the company in this story would learn the most from her second answer, not the first. Her agent did nothing wrong. Her journey did. If the dashboard only measures the first, the problem stays invisible, and she renews with someone else next year.
Figures and names in the examples are illustrative.
Leave a Reply