Zendesk audit · Two weeks
We score your instance against a documented build standard, object by object, and hand you the list. Triggers, macros, views, fields, routing, Guide, Talk. Then we tell you whether fixing the configuration is actually the highest-value thing you could do this quarter.
Not on Zendesk? The standard is about what a help desk has to do, not about one vendor. Gorgias, Intercom and Kustomer all translate.
Four things, none of them visible from outside your Zendesk instance.
Wrong Contact Reasons
The three contact reasons in the signed Schedule B are 3.5% of your volume. They rank 11th, 12th, 15th and 19th out of your real drivers.
Wrong Zendesk Plan
Your Zendesk plan cannot run the assistant. You are on Support Professional. It needs Suite with Advanced AI.
No Channel to Build On
Zendesk AI runs on Messaging. You run Chat Classic. Zendesk switches Messaging on by itself on October 5th.
Missing Contact Reasons
Half your tickets carry no contact reason, because the field is not on 18 of your 19 ticket forms.
Finding this is what Schedule A was for.
From a real engagement. Client and figures redacted.
One week
For a Zendesk-scoped audit
From $2,500
Credited toward whatever you build
Config screens only
No export of your customer ticket data
What we score against
This is the standard our delivery team builds to. The order matters more than the list, and reporting going first is the part most instances get backwards.
01
First, not last. If nobody agreed what counts as resolved, or which field marks a contact reason, every number downstream is a negotiation. Most instances we open have reporting bolted on at the end, which is why nobody trusts it.
02
Email, messaging, Talk and Guide. We treat Talk and Guide as channels rather than as separate systems, because a phone call and a help center article are both ways a customer tries to get an answer.
03
How a ticket gets assigned, what moves it forward, and what quietly stalls it. Assignment should be manual selection from clean views or native omnichannel routing. If it needs a third-party app to work, something upstream is wrong.
04
Whether the configuration survives double the volume and twice the agents, or whether it is held together by things one person remembers.
What we actually open
Triggers and automations, including the ones quietly fighting each other
Macros, and how many of them anyone actually uses
Views, and whether an agent can tell what to work next without asking
Ticket fields, forms and whether contact reasons are captured at all
Assignment and routing, and where tickets stall waiting on someone who does not know they are waiting
Status-based notifications that remind the customer or the team to come back
Guide: whether the help center can carry self-service, and whether it could feed an AI agent
SLAs and business hours, including the holiday coverage everyone forgets until December
Groups, roles and permissions, particularly where regulated records need separating
Talk configuration, and the boundary with a phone system that sits outside Zendesk
Why it isn't Zendesk-only
Teams arrive certain the help desk is the problem, and often it does need work. But a messy Zendesk is frequently downstream of something else: no agreed contact reasons, a staffing model that does not match when volume actually arrives, or a channel nobody is watching.
Recent engagement
A client came to us to assess their Zendesk. It had been live about three months and they felt it was not working. We scored it against the standard and found real gaps, including no ticket types, no tags and no populated custom fields, which meant nobody could say what customers were contacting them about.
We also found that 57.7% of their inbound callers never reached a person, and that 56% of inbound was a repeat contact within fourteen days. Their phone platform sits outside Zendesk entirely, so a Zendesk cleanup would not have touched either number.
Anonymised at the client's stage of engagement. Figures from a 90 day contact analysis.
To be clear about what you're buying
Zendesk scope. One week, from $2,500. We read the system: the 41 checks, the object inventory, how tickets actually move. Everything on the data side comes back marked Not assessed, because configuration cannot show it.
CX operations scope. Two weeks, from $3,500. The same Zendesk work, plus interviews with the people working the queue and an analysis of your volume and handle time. We read the system, the team and the numbers.
Which one you need is a twenty minute conversation, not a checkbox. We'll tell you, and we'll tell you why. Deciding it yourself means buying your own diagnosis, and that is the thing an audit exists to check.
Scale
The configuration is the smaller problem. We rebuild staffing, quality and reporting around it.
Automate
You are ready for an AI agent. Guide gets built properly first, then the bot.
Operate
It works now and somebody has to keep it working every week.
Or you run it
The findings are yours, priced line by line. Plenty of teams execute internally.
13 consecutive 5 star Clutch reviews, across ecommerce, healthcare, nonprofit and retail. Read the case studies →
WHO'S ACTUALLY DOING THIS

I'm Ty Givens. 25 years running support operations at See's Candies, Thrive Causemetics and Intuit before starting CX Collective. I have inherited a lot of other people's Zendesk instances. I lead every engagement, and my team does the hands-on work.
25 years in support operations
Start here
Six questions, about two minutes. We come back within one business day with your range and which scope we think you need.
If we think the help desk on its own will answer your question, we will say so and quote the shorter engagement. We would rather sell you the week than the fortnight if the week is enough.
Questions we get
From $2,500 for a Zendesk-scoped audit, which runs a week. The wider CX operations scope adds interviews and a volume analysis, runs two weeks, and starts at $3,500. Within either one the price moves on how many systems are connected, whether your phone platform sits outside the help desk, and how many brands or instances you run. Not on how many agents you have.
Yes, and for some teams that is the right call. It runs a week and covers the instance properly. What it cannot do is tell you whether the instance is the real constraint, because everything on the data side (volume, reply time, reopen rate, what is happening on channels outside Zendesk) needs more than a configuration read. If the week turns up something that points outside the instance, we will say so, and the fee comes off the wider engagement if you want it.
We tell you, after about twenty minutes on the phone. It is not a checkbox on a form, and that is deliberate. Picking it yourself means buying your own diagnosis, and whether the help desk is genuinely your problem is exactly the question an audit exists to answer. Broadly: if you are describing a tool, the Zendesk scope usually fits. If you are describing an outcome (response times, escalations, a team that feels underwater), it usually does not.
It helps, but we can work from limited or borrowed access. We review configuration screens, not customer ticket content, and we do not export your ticket data. If your security team needs a narrower permission set, tell us and we will work inside it.
Yes. The build standard is about what a help desk has to do, not about one vendor's menu. The four pillars are the same and the specifics translate. We will tell you if the platform itself is the wrong fit, which is a different and more expensive conversation.
Sometimes, and the audit will say so plainly if the answer is yes. It will also say so if the answer is no, which happens more often. A migration is a real project with a real bill, and it is worth being sure the platform is the constraint before you move.
A scorecard against the build standard, an object-level list of what to change, a prioritized plan, a price against every recommendation, and a live readout walked through with you rather than a PDF in your inbox. On the wider scope that extends across all five areas with your staffing gap calculated from your own numbers. Either way the fee comes off whatever you decide to build, as long as you start within 30 days.
Two weeks, scored against the standard, with a price against every recommendation. The fee comes off whatever you decide to build.