Technology rarely decides whether a deal happens. It regularly decides how much is held back at closing.
Buyers in the lower middle market are not evaluating your software estate the way a venture investor evaluates a product. They are asking a narrower question: what breaks when the current owner leaves, and what liabilities are we inheriting that nobody has looked at?
Here is what they actually check, and what a good answer looks like.
Ownership and transferability
The first thing a diligence list asks for is your contracts. The concern is not cost.
Whose name is it in? Subscriptions on a founder's personal card or personal email are extremely common and always flagged. They imply the business does not own its own tooling.
Does it assign on a change of control? Some software licenses do not transfer automatically. Enterprise agreements, industry-specific systems and anything negotiated at a discount frequently require consent. Finding this in week eight of diligence is a delay; finding it earlier is admin.
Who holds the domain, and the DNS? More deals are held up by an unreachable former web developer than anyone expects. Confirm you control the registrar, the DNS and the hosting account, in the company's name.
Is the source code yours, if custom work exists? A developer agreement without an IP assignment clause means the business may not own the thing it depends on.
A good answer: a one-page register — system, vendor, account owner, renewal date, assignability, annual cost.
Data
The area where findings become expensive rather than administrative.
What customer data do you hold, where does it live, and under what terms? Particularly relevant if you use tools that process customer records, transcripts, images or payment information.
What have you told customers? If your privacy policy makes commitments your systems do not honor, that is a live exposure the buyer inherits.
Who has access? Shared logins, credentials held by departed staff and admin rights nobody has reviewed are common and easy to correct in advance.
If AI systems touch client data, what does the vendor's agreement permit? Whether inputs are used for training, where processing happens, and what retention applies are all reasonable diligence questions and awkward ones to answer for the first time under a deadline.
We cover the practical version of this in AI Agents and Client Data.
A good answer: you can say what data exists, where it sits, who can reach it, and which agreements govern it.
Key-person risk in systems
The same discount that applies to owner dependence applies to technical dependence, and it is often overlooked because it sits with someone other than the owner.
Buyers probe for:
- The one person who understands the system. If the spreadsheet that runs scheduling is maintained by a single employee with no documentation, that is key-person risk in a different costume.
- Undocumented custom work. Automations, integrations and reports built by someone who has left.
- Whether anyone else can administer the core systems. Two administrators is a materially different answer from one.
A good answer: documented procedures, more than one person with administrative access, and a written record of what custom work exists and why.
What buyers make of your AI specifically
The bar has moved. Presence is no longer credited — roughly 89% of US small businesses report using AI, while under 9% use it in producing goods or services. Saying "we use AI" places you in the first group, and the first group is not paid for.
What earns credit:
- It survives your departure. A system running on written procedures transfers. One running on the founder's undocumented prompting does not.
- It changed the cost structure measurably. Not "we use it for drafting" but a task that no longer consumes the hours it did.
- Someone checks the output. An automation with no verification step is a liability. Buyers ask who reviews it and what happens when it is wrong, and "it has always been fine" is a poor answer.
- The data position is clean. See above.
What attracts a discount: automations nobody can explain, tools with access to customer data under unreviewed terms, and anything that stops working when one person is unavailable.
Security, proportionate to your size
Nobody expects an enterprise security program from a $20M business. They do expect basic hygiene, and its absence reads as a proxy for how the rest is run.
Multi-factor authentication on email and financial systems. Backups that have actually been restored at least once. A password manager rather than a shared document. Offboarding that removes access. Some cyber liability cover.
These are cheap and they are noticed.
Preparing, in about a day
If a transition is 12–24 months out, most of this is one focused session:
- Build the systems register. Every tool, its owner, cost, renewal date and whether it assigns.
- Move everything into company accounts and out of personal names.
- Confirm you control domain, DNS and hosting.
- Review access — remove departed staff, eliminate shared logins, add a second administrator to anything critical.
- Read your own privacy policy against what your systems actually do.
- Write down the undocumented things — the custom automations, the spreadsheet that matters, the process only one person knows.
Step six is the one that pays twice. The document a buyer wants here is the same document an AI system needs in order to do the work at all, which is the argument in The Process Documentation Dividend.
The realistic framing
Technology diligence is unlikely to win you a higher multiple. Almost nobody pays more because the software register was tidy.
What it does is prevent value leaking at the end — through holdbacks, extended escrow, specific indemnities, or a re-trade when something surfaces late. Those are the mechanisms by which an agreed price becomes a smaller number, and they are all easier to avoid than to renegotiate.
Related
AI Tool Sprawl covers auditing the software estate. AI Agents and Client Data covers the data position. The Sale Process, Start to Finish covers diligence overall. Does AI Adoption Increase Enterprise Value? covers what acquirers pay for.
This guide is for informational purposes only and does not constitute investment, financial, legal, tax or technology advice.