Better answers
Customers keep asking things that are on our website. What should we fix?
Put your website to the test with real customer questions, a five-part diagnosis and a downloadable answer-finding scorecard.
In this guide
THE SHORT ANSWER
Choose a small set of questions customers actually ask, then watch likely customers try to find and use the answers on your website. Check whether each answer is missing, buried, confusing, contradictory or needs a person. Fix the observed obstacle, retest it and keep human help easy to reach. An answer existing somewhere on the site does not mean it works where customers need it.
“It is on the website.” You know exactly where. You wrote it, approved it or have pointed people towards it a hundred times.
Your customer is trying to work out whether you deliver to their street, whether a booking is confirmed, or what they need to bring. They may be using a phone, following an old link or reading a page that uses different words from theirs. They may also want reassurance before spending money.
Put the website on trial. Let customer questions be the witnesses. The useful verdict is a specific change: a clearer label, a missing condition, a corrected promise or a working route to help.
The exercise below is a small diagnostic you can run yourself. The examples and scorecard are original fictional material, not results from a customer study.
First, check whether callers used the website
Ask neutrally: “Where did you look before contacting us?” If they used the site, follow with “What were you trying to find?” Avoid “Did you read the delivery page?” which points towards your preferred explanation.
Some callers never visited the site. Others found it through a search result, social post, map listing or a link in a confirmation email. Record that starting point. A caller who saw an outdated listing presents a different problem from someone who could not understand your current page.
Do not assume every call represents a website failure. A person may need an exception, an accessible alternative or an answer about their own order. Keep that distinction throughout the exercise.
Collect ten questions, then choose a few tasks
Use recent notes or transcripts your business is authorised to review. Remove identifying details. Write the question in the customer's language and the decision they needed to make. Ten is a convenient starting pool, not a required sample size or proof that you have covered every need.
Choose three or four distinct tasks for an initial session. Include a routine question and one with an important condition. The recurring-question guide can help you choose the material; this exercise starts once you have the questions.
Turn each question into a realistic goal, without telling the participant where to click:
- Leading: “Open our delivery FAQ and find the postcode list.”
- Useful: “You want this item delivered to the fictional postcode AB12 3CD. Work out whether delivery is available and what you would do next.”
For your own test, use an appropriate real service area and a made-up customer scenario. Stop before a purchase, booking submission or message is sent.
GOV.UK's usability-testing guidance recommends tasks with a clear, believable goal that do not reveal how to complete them, and participants who are actual or likely users. Our smaller worksheet adapts those principles; it does not replace a full research programme.
Watch someone try, without becoming their tour guide
Invite a willing person who resembles a likely customer and has not helped build the page. Someone outside the business can expose obvious problems, but one friend cannot represent all your customers. Over subsequent sessions, include relevant differences in language, familiarity and accessibility needs.
Let participants use a familiar device and their usual accessibility settings. A phone is useful if that is how your customers visit. Begin at the page or entry point relevant to the task rather than automatically at your homepage.
Say: “We are checking whether the website helps you. Please tell me what you are looking for and what you think the page means.” Then observe.
Record the first place they look, the words they hesitate over, the answer they find and the action they would take. Ask “What would you do next?” before explaining anything. If you help them, note the help; do not count an assisted answer as independently found.
A person can locate a sentence and still misunderstand it. If they find “request received” and conclude that their appointment is booked, the page has not supplied the certainty they need.
Give each obstacle the right name
Use these five diagnoses. More than one may apply. Also allow “no obstacle observed” and “not assessed” so the worksheet does not force a fault where you have no evidence.
On a small screen, swipe the table to see all columns.
| Diagnosis | What you observe | A useful next step |
|---|---|---|
| Missing | The needed condition or next step is absent | Add the approved answer at the relevant decision point |
| Buried | The answer is accurate but the participant cannot find it | Improve the route, heading or placement |
| Confusing | They find it but draw the wrong conclusion | Rewrite the wording and test its meaning |
| Contradictory | Two pages or messages give different answers | Ask the policy owner to resolve the conflict and update affected copies |
| Needs a person | The answer depends on a case, live status or exception | Explain what help can do and give a clear contact route |
The downloadable website answer scorecard has a repeatable record for each task, space for observed evidence and a retest. It is a plain-text worksheet you can copy or print, with no account or software purchase required.
Three fictional verdicts
“Do you deliver to my area?” — buried
The delivery page contains an accurate postcode list. The service page links to it using the label “Our commitments”. The participant opens the contact page and says they would call.
The proposed fix is a descriptive “Check delivery areas” link beside the delivery option, with a short explanation of what to check. Adding another FAQ entry would leave the weak route in place. Retest from the service page and observe whether the person can make the delivery decision.
“Am I booked for Saturday?” — confusing
The form confirmation says “All set!” but the small print says a member of staff must confirm availability. The participant plans their trip immediately.
The policy owner confirms that submitting the form only requests a slot. A clearer message might be: “We have received your request for Saturday. Your booking is not confirmed yet. We will contact you by [a time the team can meet].” The business must also have a process to send that confirmation. Better wording cannot supply a missing follow-up.
“Can you tell me when my repair will be ready?” — needs a person
The website gives a typical turnaround range. This repair is waiting for a part. The participant reasonably wants a case-specific answer before travelling.
Do not make the general estimate look like a confirmed date. Explain how to request a repair update, what reference to provide and when the team can respond. The appropriate result may be a clearer, better-prepared call.
Fix the cause, then try again
Start with an obstacle that could lead to a wrong purchase, wasted journey or unconfirmed commitment. Agree the correct answer with the person responsible for the service, then choose the smallest change that addresses what you observed.
Keep the original question and entry point for the retest. Where practical, use a fresh participant: someone who has already been shown the answer may remember the route. Record whether the answer was found independently, understood correctly and enough to choose a next step. A passed retest supports that specific task; it does not prove the whole site is easy to use.
If the answer itself needs writing, use the FAQ writing guide. If your team disagrees about the answer, run the same-question team exercise before publishing a new promise.
For a later business review, count the same question across comparable periods and account for relevant activity such as orders or bookings. Keep an eye on messages as well as calls and check that customers can still get help. A fall in phone calls alone cannot tell you whether the page worked or people gave up.
Where Call Nerd helps
Call Nerd can surface recurring customer questions and the transcripts behind them. Those questions can supply tasks for this exercise, after a person checks their meaning and removes identifying details.
The call record does not show where someone clicked unless they described it, and it does not explain every caller's reason for contacting you. Watching likely customers use the page supplies different evidence. Your team still approves the answer, changes the site and checks the result.
The next time someone says “but it is on the website”, you have a practical follow-up: “Can someone who does not already know the answer find it, understand it and act on it?”