The biggest red flag when hiring an automation or AI consultant is a tool recommended before anyone has looked at your process. The others follow from it: no look at your data, a design that locks you into one platform, no plan for exceptions and approvals, no support after go-live, an explanation you can't follow, and a quote too vague to hold anyone to. A good consultant asks how the work is done now, looks at real examples and real data, tells you when a cheaper option is enough, and gives you a clear scope and price. Below are seven warning signs, what a better answer sounds like, and the questions to ask any consultant before you sign, including us.
1. They pitch a tool before they look at the process
If the first meeting ends with "you need an AI agent" or "we'll put it all in platform X" before anyone has asked how an order, a job or an invoice actually moves through your business, the recommendation isn't about your business. It might be the tool they know, the one they resell, or the one that's in the news.
A better sign: they ask to see the process with real examples, including the messy ones, and talk to the people who do the work, not just the owner. The tool comes after the problem is clear. That's the same idea as our post on fixing the process before you buy the software.
2. Nobody looks at your data
Automation and AI run on your data. If customer names, product codes or job numbers don't match across your systems, an automation will copy the mess faster, and an AI tool will give confident answers built on it. A consultant who quotes without asking what state your data is in has either assumed it's clean or doesn't plan to deal with it.
A better sign: they ask for a sample of the real records early, point out duplicates and gaps, and tell you whether clean-up is part of the job. Our data cleansing services page explains what that involves.
3. The design locks you into one platform
Some lock-in is normal: every tool has its own way of doing things. The red flag is a design where your processes, your data or your rules only exist inside a platform you can't easily leave, or inside the consultant's own account, so that moving later means starting again.
A better sign: they explain what you'll own at the end (the configuration, any code, the documentation, the logins), where your data is stored, and what it would take to move it later. Your accounts should stay in your accounting software, and your operating data in systems you control.
4. No plan for exceptions, approvals or a person in charge
Demos show the happy path. Real work is full of exceptions: the invoice with no purchase order, the customer on credit hold, the timesheet with no job number, the manager on leave. If the proposal doesn't mention what happens to those, they'll land on your team as manual workarounds after go-live.
For AI, this matters more. Someone needs to check the output, at least at first, and approve anything that touches money, customers or staff. The Australian Government's Guidance for AI Adoption sets out six essential practices for using AI responsibly, including keeping human oversight, with a simpler foundations version for businesses earlier in their AI use.
A better sign: the proposal lists the main exceptions, who handles each one, and where a person approves before anything goes out. Our post on running a small AI pilot safely shows what that looks like in practice.
5. No support after go-live
Systems change. Vendors update their screens and connections, staff leave, the business adds a product line. An automation nobody looks after tends to fail quietly, and you find out when a customer or supplier complains.
A better sign: they tell you who will notice if it stops, who fixes it, how fast, and what that costs. If they expect your team to look after it, they should show you how and leave written instructions.
6. They can't explain it in plain English
You don't need to understand the technical detail, but you should understand what the system will do, what it won't do, and what your team does differently afterwards. If every answer comes back in acronyms, either the consultant doesn't understand your business well enough to translate, or they'd rather you didn't ask.
A better sign: they can describe the result in your words ("supplier bills arrive in the accounts matched to the purchase order, and anything that doesn't match goes to Sam to check"), and put the technical detail in a separate section for whoever looks after your systems.
7. The quote is vague
Watch for quotes that are only a day rate with no end point, a list of phases with no deliverables, promises of precise savings before anyone has seen your business, or a very low price for "phase one" with everything that matters in "phase two".
A better sign: a clear scope, what's included and what isn't, what happens if the scope changes, and what support costs afterwards. When a fixed quote isn't possible yet because the problem isn't clear, a fixed-price discovery step, such as our Process & Systems Audit, is a fairer start than an open-ended day rate. Our guide to what an automation or AI project costs explains what drives the price.
Questions to ask any automation or AI consultant
These work for anyone you're talking to, including us.
- What would you need to see before recommending a tool? Listen for "your process, real examples and your data".
- When did you last tell a client they didn't need you, or that a cheaper tool would do? A good consultant has an answer.
- What happens to the exceptions? Ask them to walk through the three most common ones in your process.
- Where does a person stay in charge? Especially for AI: who checks the output, and what can it do without approval?
- What will we own at the end, and could we move to someone else? Configuration, code, documentation, logins.
- Do you sell, resell or earn commission on any software you'd recommend? That's fine if they do, but you should know.
- Who actually does the work? The person in the meeting may not be the person building it.
- What's fixed in the price, and what isn't? Get the scope and the change process in writing.
- What does support look like after go-live, and what does it cost?
- Can we start with one small piece and prove it works? The answer should be yes.
For more on choosing someone for the process side of the work, see what a business process consultant actually does, and how to choose one.
How we try to avoid these red flags
We start with the process and the data, not a tool. We look at the steps that sit between your systems, where information gets typed in again or stops before it reaches the people who need it. We'll tell you when a setting, a connector or a low-cost tool is enough. AI does the first pass on specific tasks and a person approves. Work is scoped and given a fixed quote after a free initial chat, and when the problem isn't clear yet, the $5,000 Process & Systems Audit maps your processes and systems first, with a written report that's yours to keep.
If you're looking at AI specifically, see our AI consultant for small and mid-sized business page. For repetitive work across your systems, see business process automation.
Frequently asked questions
What should I look for in an automation consultant?
Someone who looks at your process and data before recommending a tool, plans for exceptions and approvals, explains the result in plain English, and gives you a clear scope, a clear price and a support plan.
Is it a red flag if a consultant recommends a specific platform?
Not on its own. It's a red flag if they recommend it before understanding your process, or if they earn commission on it and don't tell you. Ask why that platform suits your situation, and what you'd own if you moved later.
How do I know if an AI consultant knows what they're doing?
Ask them to explain where AI fits in your process, where it doesn't, and who checks its output. Ask what they'd need from your data. Vague answers, or promises of precise savings before they've seen your business, are warning signs.
Should I get more than one quote?
Yes, if you can. Give each consultant the same examples and the same description of "done", so the quotes are comparable. Compare the scope and support, not just the price.
If you'd like a straight conversation about your process before anyone mentions a tool, book a free initial chat.