AI for Small Business

Customer Service Audit Checklist: Test the Route an Enquiry Takes

By Dante AIOctober 9, 20266 min read

TL;DR: A contact page can show several ways to reach you while none of them reliably reaches a person who can help. This customer service audit checklist gives you a five-check test for one contact route at a time: can a customer use it, who owns it, what expectation it sets, what happens when it cannot help, and whether it survives closing time. Tick only what you observed, fix or test the first check you did not observe, and repeat the test on your next route.

A contact page can look busy while nothing behind it works. An email address, a chat button and a phone number sit side by side, and each one is a way for an enquiry to become nobody's problem.

Many small businesses collect enquiries through more than one route. The useful question is not how many routes you offer. It is whether each one actually reaches a person who can finish the job. The simplest way to find out is to follow one enquiry from the customer's screen to that person.

That is the customer service audit in this guide. It is deliberately narrow: it tests the route an enquiry takes, not your whole support operation or contact centre. Pick one route, send a harmless test enquiry, and check five things you can observe directly. It is a check of your own routes, not a comparison of channels or vendors.

Run the five-check route test

Choose one route your business actually offers: your contact form, support email, website chat, social inbox or another channel you use. Tick only what you observed during the test. The tool below gives each check equal weight and lists the checks you have not observed, to fix or test first. It is an editorial worksheet, not a validated customer experience score. Five out of five means you saw these five parts of this route working today, nothing more.

1. Can a customer find and use it?

Open your site on a phone and look for help from a product page, rather than starting from the contact page you already know exists. If a button is hidden behind another widget, or the form asks for something the customer cannot reasonably know, write down the specific obstacle. "Our contact page is live" describes a URL. "A customer can get through" describes an experience.

2. Who owns the enquiry after it arrives?

Find the actual inbox or queue the route feeds. Name the role responsible for it and the backup when that person is away. "The team" is too vague if everyone assumes someone else is watching. You do not need a support department for this. A small business can assign a role, a check frequency and a backup.

3. What expectation did you set?

Read the message shown after you sent the test. Does it confirm receipt? Does it say when a useful reply can be expected, and is that realistic for the hours you actually staff this route? An automatic receipt can help, but it is still a receipt. Do not count it as an answer.

4. What happens when this route cannot help?

Send a question outside the route's purpose: a support question to the sales form, or a question your chat content does not cover. Check what the customer is told and where the enquiry goes next. Do not assume a fallback works because someone wrote an instruction for it. Watch what actually happens, and check the person who receives it gets enough context to understand the problem.

5. Does the route survive closing time?

Send a test outside your staffed hours and look for it during the next working period. The question is not whether your website stays online. It is whether the business receives and handles the enquiry. A hypothetical repair business might find its form works but nobody checks weekend submissions. Three observed checks and two gaps is enough to know where to start: a Monday morning inbox check and an honest reply expectation. There is no need to add a new channel first.

Fix the weakest link, then repeat

Record the route, the test date, what you observed, the owner and one fix. Repeat the test after the fix. Once one route works reliably, run the same five checks on the next one. Routes drift as people change roles and hours change, so the test is worth repeating whenever your team or opening hours change.

An AI agent can be part of a contact route. An AI agent trained on your own content can be available to answer common questions around the clock on the routes it is connected to, which can help with the out of hours check. Being available is not the same as resolving every question, so test it on real enquiries. It does not decide who owns the route or what happens when it cannot help, so the other checks still apply. For a comparison of chat and forms, see AI agent vs contact form. For questions that need a person, the human handover buying guide covers that decision separately.

Start with a route you already have. Make it work from both ends. Then decide whether to add another.

Customer service audit: test one route end to end

Without the interactive tool, count the checks you observed on this route. Each one is worth 20 points out of 100. An unticked check means not observed: it may have failed or not been tested yet, so fix or test it first.

Frequently asked questions

Is this a full customer service audit?

No. It audits one thing well: whether an enquiry sent through a route you offer reaches someone who can help. A full audit of a support team would also look at answer quality, workload and customer feedback.

What counts as a contact route?

Any way a customer can reach you: a contact form, a support email address, website chat, a phone line, a social media inbox or a messaging app. The test applies to one route at a time.

Why test one route instead of listing every channel?

A list tells you which channels exist. A test tells you whether an enquiry actually reaches someone who can help. One observed test usually reveals more than a long list of channels you assume are watched.

What if I do not offer a channel at all?

Then there is nothing to test. The checklist only scores routes you actually offer. Leave out routes you do not have, rather than counting them as gaps.

Does a perfect score mean my support is good?

No. It means you observed five basic parts of one route working on the day you tested. It is a rough heuristic, not a measure of customer satisfaction or answer quality.

Start free