What is CES? Customer effort score formula, questions, and examples
When customers say support is a hassle, customer effort score is the metric that shows where. It points to the steps where people work too hard to get something done, which a satisfaction score can miss.
What is customer effort score (CES)?
Customer effort score (CES) measures how easy it was for a customer to get something done, such as resolving a support issue, completing a return, or updating an account. It's usually a single survey question sent right after the task.
The idea was popularized by a 2010 Harvard Business Review article, Stop Trying to Delight Your Customers, which argued that companies earn more loyalty by making service easy than by going above and beyond.
CES survey question examples
| Format | Example question | Typical scale |
|---|---|---|
| Agreement statement | “[Company] made it easy for me to handle my issue.” | 1 (strongly disagree) to 7 (strongly agree) |
| Ease question | “How easy was it to complete your return?” | 1 (very difficult) to 5 or 7 (very easy) |
| Effort question | “How much effort did it take to handle your request?” | 1 (very low effort) to 5 (very high effort) |
Check which way the scale runs. On an ease or agreement scale, a higher score is better. On an effort scale, a lower score is better. Share the exact wording and direction along with the score.
How to calculate CES
There are two common ways to calculate CES. Both examples below use 10 responses on a 1 to 7 scale, where 7 means the task was easy.
Average CES
Average CES = sum of all ratings ÷ number of responses
If the 10 ratings add up to 57, the average CES is 57 ÷ 10 = 5.7 out of 7. The average picks up movement anywhere on the scale, so a customer going from 5 to 7 still moves the number.
Percentage of easy responses
CES = easy responses ÷ all responses × 100
If 7 of the 10 customers chose a rating you count as easy (for example, 5, 6, or 7), the CES is 7 ÷ 10 × 100 = 70%. This is easier to explain, but a customer who moves from 5 to 7 doesn't change the result.
Pick one method as your main CES and label it clearly, such as “Average CES, 1 to 7” or “% of customers rating the task easy.” Don't compare a score from one method with a score from the other.
What is a good CES score?
There's no universal benchmark, because companies use different questions, scales, and calculations. On a 1 to 7 ease or agreement scale, an average above the midpoint of 4 means customers leaned toward easy. Past that, the most useful comparison is your own trend for the same journey, measured the same way each time.
When to send a CES survey
Send the survey right after the task is finished, so the customer knows exactly what they're rating. Good triggers include:
- A return is submitted and confirmed
- An appointment is changed
- A case is closed after the customer confirms it's resolved
- A conversation ends
- A self-service flow finishes, whether it succeeded or not
Timing changes the answer. A survey sent right after a transfer may measure the transfer, not the whole issue. One sent days later can be colored by other experiences. Limit how often the same customer is surveyed, and track response rates by channel.
How to run a CES survey
- Name the task. Ask about the thing the customer just did, not “customer service” in general.
- Ask one thing. Don't combine ease, speed, and resolution in one question. A task can be easy but slow, or quick but unresolved.
- Label both ends of the scale. Make it obvious which end means easy.
- Keep the scale the same everywhere. If 7 means easy in email but difficult in chat, a combined score is meaningless.
- Ask why. Add an optional follow-up such as “What made this difficult?”
If you change the question, scale, trigger, or calculation, note it on the trend so a change in method isn't mistaken for a change in the experience.
How to find what's causing effort
A company-wide average is fine for tracking, but it rarely shows what to fix. Break CES down by:
- Journey or reason for contact
- Channel
- Outcome: resolved, unresolved, abandoned, or transferred
- Number of handoffs
- Identity verification path
- Whether an AI agent, a person, or both handled the request
- Version of the policy or workflow
Then compare CES with whether the task actually got done:
| CES | Resolution | What it usually means | What to check |
|---|---|---|---|
| Easy | High | The journey is working | Keep an eye on response rates |
| Hard | High | Customers get there, but it takes too much work | Repeated questions, wait times, handoffs, and policy steps |
| Easy | Low | The conversation felt simple but the work isn't done | How resolution is defined and when the survey is sent |
| Hard | Low | The journey has both an outcome problem and an effort problem | Start here: find the root cause and redesign the journey |
Other measures worth tracking next to CES include resolution rate, first-contact resolution, repeat contacts, abandonment, and total time to resolve.
Reduce effort, then measure again
- Find the friction. Narrow it to the journey, channel, and customer group where effort rose.
- Look at real conversations. Read customer comments, transcripts, transfers, and system logs.
- Name the cause. Common ones are repeated identity checks, unclear policy, a missing system action, or an unnecessary handoff.
- Change one thing. Simplify the step, fix the knowledge or integration, or rework the handoff.
- Check the full result. Track CES along with resolution and repeat contacts to confirm the fix worked.
Avoid making “raise CES” a goal on its own. Teams can hit it by changing who gets surveyed or when, without making anything easier.
CES vs. CSAT and NPS
CES asks how easy a task was. CSAT (customer satisfaction score) asks how satisfied the customer was with it, and Net Promoter Score (NPS) asks how likely they are to recommend you overall. A task can be easy and still leave the customer unhappy, or hard but resolved, so many teams track CES and CSAT for individual interactions and NPS for the overall relationship.
Measuring customer effort with AI agents
An AI agent can answer a question, take an action like processing a refund, or hand the conversation to a person along with what it has learned. Measure effort across the whole journey, from the customer's first message to the finished result, including identity checks, waiting, and any follow-up. A quick conversation that leaves work unfinished isn't a low-effort experience.
When comparing an AI agent with your team, compare the same types of requests. If the two handle a different mix of requests, a straight comparison reflects the mix as well as the experience.
How Sierra customers use CES
Sonos built its own Customer Effort Score from how often, how many, and what kinds of questions customers ask, so it looks beyond a single interaction to the broader customer relationship. Improving that score was one of Sonos's goals for its AI agent built on Sierra, which follows the context of a whole conversation.
Sierra customers can use Ghostwriter, Sierra's agent for building agents, to analyze how customer experience metrics are trending and suggest changes to the agent to improve them.
See how Sierra customers use AI agents to get better results.
