Most automation projects that fail were dead before any software was installed. The system got pointed at a price book with three living versions, or a shared drive nobody has opened since the last office manager left, and it did exactly what it was told — faster, and to more people, than the mess ever spread on its own.
A system inherits everything, including the mess
An automated quote follow-up pulls its numbers from your price book. A document assistant answers from whatever files you handed it. Neither has judgment about which version is current, because judgment is exactly the thing you were hoping to stop paying for.
An HVAC shop makes the point well. Equipment costs moved hard through the refrigerant transition — the sheet in the office binder still shows last year's condenser pricing for R-410A units that the distributor no longer stocks. A human estimator knows to ignore the binder. Software does not. It will quote the old number with total confidence, in writing, to a customer who will screenshot it.
Construction has its own version. The bid goes out, the sub replies with a number scrawled on a photo of a spec sheet, texted to the estimator's personal phone. That number never enters any system anywhere. Automate the bid tracker without fixing that intake and you have automated a record everyone knows is incomplete. The software isn't wrong. The seam it sits on is.
The readiness checklist
Three checks, none of them technical. A business that passes all three is ready to automate something. A business that fails all three should not spend a dollar yet.
One current source of truth per fact
For every fact a system will repeat — a price, a warranty term, a service area, a lead time — there should be exactly one document that is current, and everyone should agree which one it is. Here is the test: ask the person at the front desk and the person in the truck what a service call costs after 5pm. If you get two answers, an automated system will pick one of them, and about half the time it will pick the wrong one, at scale, in writing.
A named human who owns the reply
When a missed-call text goes out and the customer replies, the question that decides whether the system works is where that reply lands and who is responsible for it. Not "the office." A name. And the name has to be someone who can actually look at a phone during working hours — roofing crews are on a roof, not near a desk, between 7am and 4pm, so if the named person is the crew lead, the design is already broken. The most common collapse of these systems is not technical. The reply lands in a dashboard nobody opens, and the customer quietly concludes nobody is home.
A document inventory with dates
Write down what you would want a system to know: warranty terms, installation specs, the price book, financing terms, the service-area map, manufacturer bulletins. Next to each item, the date it was last revised and the person who would revise it. If half the list reads "not sure," that is the project — before any software. A warranty document from 2019 indexed alongside the bulletin that superseded it last spring produces confidently outdated answers, and no retrieval system can tell them apart on its own. A revision date in the file can.
The walk-away case
Sometimes the right recommendation is to buy nothing. If the price book has not been reconciled in two years, if no one can be named to own replies, if the documents worth indexing do not exist yet, then automation adds a fast, tireless amplifier to a process that is not ready for one. We would rather say "fix these three things first, then call" and mean it.
The fix usually costs nothing but attention: a week of reconciling the price book, an hour deciding who owns the inbox, an afternoon dating the documents. Some businesses do that work and discover the process runs fine without any software at all. That is a good outcome, and it should not embarrass anyone that it happens.
Start with the smallest workflow that touches money
When the checklist passes, resist the urge to automate the whole intake at once. Pick the smallest workflow with a clear before and after — often the missed call on the main line. One number, one trigger, one message, replies routed to a phone someone already looks at. Then measure for a month: how many calls were missed, how many texts drew a reply, how many replies became jobs. Those three counts, from your own line, are worth more than any industry statistic, and they tell you whether the next piece deserves to exist.
The sequencing rule underneath this: every workflow must prove itself before the next one is built, because every workflow you add is also a new thing that can break silently. Two proven systems beat six installed ones.
What readiness will not fix
A clean price book and a named inbox owner solve the capture problem, not the demand problem. Automation recovers leads you are already generating; it does not create them. If the phone is not ringing, the constraint is advertising or reputation, and this is the wrong tool — spending on capture before there is anything to capture just produces very well-organized silence.
Readiness also decays. The price book that is current in March drifts by August, especially in trades where material costs move with tariffs and supply. Ownership has to include a revision habit — a standing fifteen minutes a month — or the checklist you passed becomes the checklist you would now fail.
Where to start
Spend thirty minutes building the document inventory: one sheet, three columns — document, last revised, who owns it. Bring it to any vendor you talk to, including us. Whether they ask to see it tells you most of what you need to know about them.