The customer service audit checklist
Forty questions across six areas — accessibility, speed, resolution, quality, agent experience, and measurement — with guidance on scoring what you find.
Most support teams have never been audited — they've been surveyed, dashboarded, and reviewed, but nobody has systematically walked the customer's path and checked each link. That's what an audit is, and a good checklist is most of the method.
Use this one in two passes: first as a customer (actually contact your own support, on every channel, including at 9pm on a Saturday), then as an operator (pull the data each question names). Score each item works / partially works / broken, and the priority list writes itself.
Area 1: Accessibility — can customers reach you?
The audit starts before any conversation does.
- Is every advertised channel actually staffed — phone, email, chat, social, SMS — or do some exist only on the contact page?
- How many clicks from your homepage to a live human? Count them; more than three is friction by design.
- What happens outside business hours — a queue, a promise with a real SLA, or a dead end?
- Can a customer choose their channel, or does the IVR/bot funnel everyone through one?
- Do you offer support in the languages your customer base actually speaks?
- Test it: contact your own support anonymously on each channel. Note everything that annoyed you — your customers noticed it first.
Area 2: Speed — how long does everything take?
Pull real numbers, not targets. The audit question is always what actually happens.
- First response time by channel — and is anything acknowledged instantly, even when resolution takes longer?
- Average and 90th-percentile wait to reach a human by phone. The percentile matters more than the average; that's the experience your unluckiest fifth gets.
- How often are callbacks and follow-ups delivered when promised? Sample twenty promised follow-ups and count.
- Time-to-resolution for the median issue — and for the worst decile. What's the longest currently-open ticket, and why?
- Abandonment rate by hour: where in the day or week do customers give up waiting?
Area 3: Resolution — does contact one fix it?
Speed without resolution is a fast treadmill. This area is where CSAT actually lives.
- First-contact resolution rate, honestly measured — does a repeat contact within a week count against it, or does closing the ticket count as resolving it?
- What share of contacts require a transfer, and how many transfers does the worst path involve?
- Does context follow the customer across transfers and channels, or does the story get retold?
- What can frontline agents NOT do — refunds, exceptions, account changes — and how many escalations does each restriction generate weekly?
- Sample twenty resolved tickets: was the fix confirmed with the customer, or assumed?
Area 4: Quality and consistency — is it good every time?
One great agent isn't a quality program. The audit looks for the floor, not the ceiling.
- Does a QA program exist with defined, observable criteria — and does it sample every agent, every channel, every period?
- How wide is the gap between your best and worst agent on FCR and CSAT? That gap is your training opportunity, quantified.
- Are tone and information consistent across channels, or does chat contradict what phone said?
- Read your ten worst-rated interactions from last quarter: is there a pattern, and did anything change because of them?
- Do required disclosures and compliance language actually appear in sampled calls, or only in the script document?
Area 5: The agent side — can your people succeed?
Every customer-facing failure has an agent-facing cause. Audit the working conditions of the service.
- How many systems does an agent touch to resolve the median contact — and how much handle time is tool-juggling?
- Is the knowledge base current, searchable, and trusted — or do agents keep private notes because the official answers are wrong?
- What does an agent's first month look like? Time-to-independent-handling is a direct quality predictor.
- Attrition rate over the trailing year, and the stated reasons in exit conversations.
- Do agents get coached from real interactions at least monthly — or only scored?
- Ask five agents what they'd fix first. They know where the audit should look; they've been filing this report verbally for years.
Area 6: Measurement and follow-through — does anything change?
The final area audits the audit-ers. Which metrics are reviewed on a schedule, and — the harder question — what was the last operational decision a metric changed? Is customer feedback from surveys and complaints routed anywhere with an owner, or collected into a dashboard nobody is accountable for? Are the top ten contact drivers known, ranked, and being actively reduced, or is every month's volume treated as weather?
Score all six areas and the shape of your problem emerges. Accessibility and speed failures are usually staffing and coverage math; resolution failures are usually authority and context-plumbing; quality variance is coaching; agent-side failures are tooling and management; measurement failures are ownership. Fix in the order the customer encounters them — being reachable beats being measured — and re-run the audit in six months. The second pass is the one that tells you whether you have a support operation or a support situation.
“Nobody has walked the customer's path and checked each link. That's what an audit is — and a good checklist is most of the method.”
The bottom line
Forty questions, six areas, two passes — one as your own customer, one through your own data. Score works / partially / broken, fix in the order customers encounter the failures, and re-audit in six months. Teams that run this annually catch their degradation before customers do; teams that don't find out from churn numbers, which are the same audit with a nine-month delay and a bigger invoice.
