Seven questions to answer before you automate anything
A pre-flight checklist for owner-operators. Most failed automation projects fail on one of these, and all seven are answerable in an afternoon.
Most automation projects that fail do not fail on technology. They fail because nobody answered a basic question before the build started. Here are the seven that matter, all answerable in an afternoon, and all cheaper to answer now than in month three.
1. What actually happens today, step by step?
Not what is supposed to happen. What happens. Take last week's inbox and call log and trace five real cases from first contact to resolution. Who touched each one, what did they look up, where did it sit waiting.
Almost everyone discovers something they did not know: an inquiry type nobody owns, a step that exists only because one person remembers to do it, a handoff that silently drops things on Fridays. You cannot automate a process you have not written down, and writing it down is usually where the first real savings appear, before any software.
2. What is the cost of the status quo, in dollars?
If you cannot put a number on the problem, you cannot evaluate a solution, and you will end up buying on how good the demo felt.
The number does not need to be precise. Missed calls times booking rate times average ticket gets you within an order of magnitude, which is enough. If the answer comes out smaller than the cost of fixing it, stop. That is a successful afternoon, not a wasted one.
3. Which part is routine, and which part is judgment?
Automation is good at the first and dangerous at the second. Go through your process and mark each step: could a careful new hire do this from a written rule, or does it need someone who knows the business?
If most steps are judgment, automation will not help much and you need a person. If most are routine, you have a real candidate. In most small service businesses the routine share is much larger than the judgment share, and the judgment share is what actually needs your morning.
4. What must never happen without a human?
Write this list before anyone writes code. Every action that would be expensive or embarrassing to reverse: refunds, price commitments, contract terms, legal notices, anything involving an angry customer.
This list is the spine of the project. We wrote about how approval gates work separately, but the short version is that a system without this list is a liability regardless of how well it performs on a good day.
5. Where does the data actually live, and can you get at it?
This is where projects stall in week two. Your property or practice software may have an API, or it may have one only on a higher plan, or it may have one that does not expose the thing you need.
Check three things before signing anything:
- Does your plan include API access, or is that an upgrade?
- Can it write, or only read? Booking requires write access.
- Who at your company can authorise the connection?
6. Who owns this in month six?
The single most predictive question. Every automated system drifts: your prices change, your policies change, a new property comes on, a staff member leaves. Something has to absorb that.
If the honest answer is "nobody, I suppose me," then a self-serve tool is the wrong shape regardless of price, because the maintenance will not happen and the system will quietly decay into something that gives out-of-date answers. Either assign a real owner with real time, or buy the managed version where it is somebody's job.
7. How will you know if it is working?
Decide the measure before go-live, while you are still honest about it. Good ones are concrete: percentage of calls answered, median first response time, number of inquiries that got no reply within an hour, jobs booked outside business hours.
Capture a baseline in the two weeks before launch. Almost nobody does this, and without it you are left arguing about impressions six months later.
If you only do one: question six. Ownership predicts outcomes better than budget, better than vendor choice, and better than how sophisticated the technology is.
What to do with the answers
Questions one through three tell you whether to automate at all. Four and five tell you whether it is buildable and safe. Six tells you what shape to buy: a tool, a platform, or a managed service. Seven tells you whether it worked.
If you want help working through them, the intro call on our booking page is essentially this list, and we will tell you if the answers point somewhere other than us. Our comparison page lays out when a cheaper option is the right call.