Skip to main content
Blog Business Automation 11 min read

Buying a Business AI Systems: What to Check Before You Sign

The seller will hand you three years of financials, a plant list, a lease and a customer count, but buying a business AI systems review can reveal what actually runs the place: whatever is sitting inside the current owner’s head. I get called in after settlement more often than before it. The new owner has […]

Two men in suits reviewing buying a business AI systems opportunities at a desk, looking at a laptop displaying a digital network diagram, with stacks of documents and a pen in a modern office.

The seller will hand you three years of financials, a plant list, a lease and a customer count, but buying a business AI systems review can reveal what actually runs the place: whatever is sitting inside the current owner’s head.

I get called in after settlement more often than before it. The new owner has bought a business that looked organised on paper and discovered in week three that the quoting logic, the supplier relationships and the reason certain customers get certain prices were never written down anywhere. They have also discovered that the systems holding the business together may be undocumented, disconnected or entirely dependent on the seller. That is the difference between buying an asset and buying yourself a job with debt attached.

Here is what to check before you sign, and what to fix first if you already did.

Key Takeaways

  • Buying a business means buying its systems, and if those systems are one person’s memory, you are buying a job rather than an asset.
  • AI systems due diligence asks one question: can this business answer, decide and follow up without the current owner in the room?
  • Ask for the process documentation, then ask a staff member to run a real job from it. If they cannot, the knowledge leaves at settlement.
  • A CRM full of untouched contacts is not dead weight. One reactivation campaign recovered $49,000 from 319 leads a team had written off.
  • Check inbound response times before you sign. Most businesses take hours to reply, and 78% of deals go to whoever gets there first.
  • Key-person dependency shows up in the price. A business that runs on documented systems sells higher and settles faster.
  • Budget the first 90 days after settlement to capture what the outgoing owner knows, while they are still contractually around.

The Financials Tell You What Happened. The Systems Tell You What Happens Next

Standard due diligence is backwards-looking by design. Profit and loss, debtors, stock, contracts, staff agreements, the lease. All of it describes a business that already happened, under an owner who is about to leave.

None of it answers the question you actually care about, which is whether the revenue survives the handover. In New Zealand, business.govt.nz covers the legal and financial checks well. The operational side gets far less attention, and it is where most acquisitions quietly lose value in year one.

The usual pattern looks like this. Revenue holds for a quarter because momentum carries it. Then the enquiries that used to get answered in ten minutes start taking two days, because the owner who answered his mobile at 7 pm is gone. The regulars who rang and asked for Dave stop ringing. Six months in, the new owner is working 60 hours a week to hold the number where it was, and calling that a successful acquisition.

The systems question is simple. Does the business hold its own knowledge, watch its own numbers and handle its own repetitive work? Or did a person do all three?

If it is the person, the price you agreed was for a machine that is walking out the door.

A prospective buyer and a retiring owner discussing buying a business AI systems considerations across a cluttered workshop office desk, with a thick folder of printed invoices open between them and unfiled job dockets stacked behind.

What AI Systems Due Diligence Looks For When Buying a Business

You do not need a technical audit. You need four honest answers, and you can get them in a single site visit if you ask properly.

Can the business explain itself without the owner?

Ask for the process documentation. Then do the part most buyers skip: pick a staff member and ask them to price a job, onboard a customer or handle a complaint using only what is written down. Watch whether they reach for the document or reach for the owner.

Tribal knowledge is invisible until you test it. If three separate people tell you “you’d have to ask Dave,” you have found the real asset in the deal, and it is not for sale.

Can the business see its own numbers?

Ask how the owner knows how the week went. If the answer involves opening the accounting software, then the job management tool, then a spreadsheet someone updates on Fridays, the business has data but no picture. Every number in that business is a manual export, which means once the owner stops doing the exports, nobody knows anything until the accountant says so at month-end.

How fast does an enquiry get answered?

Send one. Use a friend’s phone number and a real question, at 4:50 pm on a Friday. Time the reply. This one test tells you more about the operation than the last three years of accounts, because it measures the gap between demand arriving and the business responding. I have written more on this in after-hours call answering, but the short version is that most small businesses lose a meaningful share of their leads in the first hour and have no idea it is happening.

How much of the work is repetitive?

Ask the owner to describe a normal Tuesday. Count how many of those tasks are the same every week: quotes typed by hand, appointment reminders sent manually, invoices chased one at a time, the same five questions answered over and over. That count is your automation opportunity and your risk. Repetitive work done by a person is a cost you inherit. Repetitive work handled by a system is margin.

The Owner’s Head Is the Biggest Item Not on the Balance Sheet

Valuation multiples reflect key-person risk whether or not anyone names it out loud. A business that runs on documented, automated systems trades higher than one of identical revenue that depends on the founder being reachable. Buyers know it instinctively. Very few price it deliberately.

Use it in the negotiation. If the diligence shows that the operating knowledge lives in one person, that is a real, quantifiable transition cost, and it belongs in the offer. It also belongs in the handover terms. A three-month consulting period where the outgoing owner is available by phone is worth far less than three months where they sit down weekly and answer structured questions about how the business actually makes decisions.

The other side of that coin matters if you are the one selling. If you are two or three years out from an exit, the work of getting the business out of your head and into a system that thinks is the highest-return project on your list. It lifts the multiple, shortens diligence and removes the earn-out clause that ties you to the business for another two years after you thought you had left.

A single empty office chair pushed back from a desk covered in handwritten sticky notes, representing the operational knowledge involved in buying a business AI systems, with a landline phone off the hook and the room otherwise deserted.

What Is Already Sitting in the Business You Just Bought

Here is the upside nobody puts in the information memorandum. Most established businesses come with assets the previous owner never worked, and you get them for free.

Start with the database. Every business trading for five or more years has a CRM, a booking system or at minimum an inbox full of people who once bought and were never contacted again. The seller wrote them off. The team wrote them off. They are still customers who know the name and had a good experience.

One of the first things I automate for a client is exactly this. For a finance broker named James, 319 dormant contacts his team had given up on produced $49,000 through a conversational, multi-touch reactivation across text and email. No ad spend. No new leads. Just a systematic conversation with people who were already there.

Then look at the missed calls. Ask the seller for the phone records, not the answered calls, the unanswered ones. In most service businesses, the number is uncomfortable, and it is the cheapest revenue in the deal. Every one of those is someone who wanted to buy and could not get through.

Then look at the quotes that never got a follow-up. Most trades and service businesses quote far more work than they win, and the difference is rarely price. It is that nobody rang back.

None of this requires new marketing. It requires a system that does the following up, which is precisely what the previous owner did not have time to build. More on that in AI agents for small business.

The First 90 Days After Settlement

If you have already signed, the sequence matters. Do these in order.

Weeks one to four: capture the brain while it is still in the building. The outgoing owner is your most perishable asset. Sit them down and record structured sessions covering how pricing decisions get made, which customers have arrangements and why, what goes wrong most often and how they fix it, which suppliers matter and what the relationship history is. Turn those recordings into written context the business can hold permanently. This is not a wiki nobody opens. It is the material an AI brain reads before it answers anything, so a new staff member can be briefed by the system instead of by you.

Weeks four to eight: get the numbers into one place. Connect the accounting, the CRM and the job or booking system so the current state of the business is visible in one summary rather than four logins. You cannot make good decisions in a business you just bought while looking at last month’s data.

Weeks eight to twelve: automate the two loudest repetitive tasks. Not ten. Two. Usually it is lead response and follow-up, because those move revenue immediately and the infrastructure already exists. Pick the tasks that eat the most hours and require the least judgment, cross them off permanently, then pick the next two. The 90-day AI automation checkpoint covers how to measure whether it worked.

The reason this order works is that automation without context produces confident nonsense. The system has to know the business before it can act on the business.

A man in a suit reviewing buying a business AI systems strategies at a conference table, with an open book, a cup of coffee, and a laptop displaying a neural network diagram in a modern office with city views.

The Real Question in Any Acquisition

Buying a business is a bet that the thing keeps running after the person who ran it leaves. Financial diligence tells you what the business earned. Systems diligence tells you whether it can keep earning. That is why buying a business AI systems due diligence matters before you commit.

Ask the four questions before you sign: can it explain itself, can it see its own numbers, how fast does it respond, and how much of the work is repetition. Price the answers into your offer. Then spend the first three months giving the business a brain of its own, so the knowledge you paid for stops living in one head and starts living in something that does not resign, get sick or forget.

Most acquisitions do not fail on the numbers. They fail because the new owner inherited a business that only worked because someone was working it, and then became that someone.

If you are looking at a business right now, or you have just taken one over and can feel the handover gaps opening up, book a 30-minute discovery call at octavius.ai/book. Bring the situation, not a brief. Half an hour is usually enough to work out whether you have bought an asset or a job, and what to fix first.

Frequently Asked Questions

What should I check about a business’s systems before buying it?

Four things. Whether the operating knowledge is documented or lives in the owner’s head, whether the business can see its own numbers in one place, how quickly inbound enquiries get answered, and how much of the weekly work is pure repetition. You can test all four in a single site visit. The response-time test alone tells you more than the last three years of accounts.

How do I value a business that depends entirely on the owner?

Key-person dependency is a discount, and it should be an explicit one. Quantify the transition cost: how long it takes to rebuild the knowledge, what revenue is at risk during handover, and what you will need to spend on systems. Then negotiate on that basis and structure the handover so the outgoing owner answers structured questions rather than just being contactable.

Can I ask a seller for their CRM data during due diligence?

You can ask for anonymised counts and activity summaries, and a reasonable seller will provide them under the confidentiality agreement. Ask how many contacts exist, how many have been contacted in the last 90 days, and what the follow-up process looks like. Full personal data usually transfers at settlement, subject to privacy obligations and the original consent under which it was collected.

What are the red flags that a business has no real systems?

Staff who answer operational questions with “you’d have to ask the owner.” Pricing that gets worked out fresh each time. A phone that goes to voicemail during business hours. Reporting that only exists after the accountant produces it. Customer history stored in one person’s memory or personal inbox. Any of these on their own is fixable. Three or more together means you are buying a person, not a business.

Should I install AI systems before or after settlement?

After, but plan before. You cannot make changes to a business you do not own yet, and the seller will not thank you for trying. What you can do during diligence is scope the work so it starts in week one. The most valuable window is the first month, while the outgoing owner is still available to explain how everything actually works.

How long does it take to get systems running in a business I just bought?

Capturing the operating knowledge and getting the numbers into one place is a matter of weeks, not quarters. Automating the first couple of high-volume tasks follows straight after. The sequence matters more than the speed: context first, data second, automation third. Automating before the system understands the business produces fast, confident, wrong answers.

Does this apply if I am buying a very small business?

More so. The smaller the business, the higher the proportion of it that sits in one head. A three-person operation where the founder does quoting, scheduling and customer relationships is almost entirely key-person risk. It is also the easiest to fix, because there are fewer processes to capture and the wins show up within weeks.

About Octavius

Titus Mulquiney is the founder of Octavius AI, where he builds AI brains and AI workforces for founder-led businesses stuck running everything out of their own head. Twenty years in marketing, ex-Sony product manager, ex-GM Zeal NZ. Based in Auckland, working with operators across NZ, Australia, and the US. Connect on LinkedIn.

Stuck running your business out of your head?

Thirty minutes. No pitch deck. Just a read on what's clogging your week.

Book a discovery call →
Keep reading All articles
Business Automation· Oct 14, 2026

Operator Trap Signs: 10 Ways You’re Running a Job

Business Automation· Oct 13, 2026

AI for Cafe Owners: Catering, Suppliers and Rosters Sorted

Business Automation· Oct 13, 2026

AI for Real Estate Open Home Follow-Up: Done by Monday