What does "selling business data to AI companies" actually mean?
AI companies have trained their models on the public internet: websites, books and forums. What they largely lack is a record of how real businesses run day to day. How a job gets booked. How a technician is dispatched. How a fault is diagnosed and fixed. How a supplier order is chased.
That knowledge sits inside established businesses, in job management systems, spreadsheets, ticket queues and maintenance logs. An AI data buyer pays the business for an anonymised copy of those records. The buyer uses them to train and test AI systems that need to understand real work.
You are not selling your business. You are selling, or licensing, a copy of records your business already creates.
What kind of data do AI data buyers want?
The short answer is operational data. That means records of process, not records of people.
| Usually wanted (operational) | Not what is being bought (personal) |
|---|---|
| Job and service history | Customer names and addresses |
| Scheduling and dispatch records | Phone numbers and email addresses |
| Maintenance and repair logs | Payment and card details |
| Workflows and standard operating procedures | Health or financial records about individuals |
| Quotes, work orders and process notes | Anything that identifies a person |
Everything is anonymised before it leaves you. Anonymised means identifying details are removed so a record cannot reasonably be traced back to an individual. If you want to understand which records carry the most weight, read what operational data is worth.
Who can sell their data?
Our partner is looking for established businesses: firms that have run real jobs, real processes and kept real records for a number of years. Start-ups with a few months of history are usually not a fit.
The business does not need to be trading. If you closed or sold a business but kept its records, those records may still be worth something. One caveat: if a company went through liquidation, the records may belong to the liquidator rather than the former owner.
How does the process work?
- Claim a valuation. You give a few details about the business and the records you hold. It takes about a minute on our data page.
- The buyer reviews what you have. Our partner looks at the type, depth and span of your records and how they are stored.
- You get a number. The buyer puts a value on the data. Deals typically land between $200,000 and $2 million.
- You decide. If the number works, you proceed to a contract. If it does not, you walk away.
- Anonymise and transfer. Records are anonymised and handed over under the agreed terms. Your business carries on exactly as before.
Is selling your business data safe?
It can be, if it is done properly. Our practical view is that three checks matter more than anything else:
- Privacy obligations. Check what your privacy policy and local privacy law allow you to do with records you collected. Check the rules in your state or country.
- Contracts. Some customer, franchise, software or supplier agreements limit how data can be used or who owns it. Read them, or have your lawyer read them, before you sign anything.
- Anonymisation. Make sure the deal spells out what is removed and who does it.
If any of those three is unclear, stop and get advice. A buyer with serious money committed will expect you to have done this.
What can you do with the money?
Because there is no equity sold and no loan taken, the cash goes onto your balance sheet without strings. Owners commonly use it to:
- Recapitalise the business without bringing in an investor or diluting shareholders.
- Pay down debt, such as an overdraft or equipment loans, which takes the repayments off monthly cash flow.
- Fund a new product, location or hire that has been on hold.
Illustrative example. A field-services business has eight years of job records in its scheduling software. The owner claims a valuation, the buyer reviews a sample, and the offer lands at, say, $450,000. The owner uses it to clear an equipment loan and a working-capital overdraft. Nothing changes for staff or customers on Monday morning. This is illustrative only, not a quote or a typical result.
Why now?
Until recently, nobody was paying for this kind of data. Now AI companies need it, and our partner has around $1 billion committed to buying it. Budgets like this are deployed over time, so there is no guarantee prices or demand stay where they are.
Disclosure
G&L M&A Advisory may receive a referral fee from our partner. It does not reduce what you are paid. This guide is general information, not financial, legal or tax advice.
Find out what your data is worth
If you have run an established business, trading or closed, the first step is a number. Claim my data valuation and we will send you on to our partner to get it.