Blog/What to Ask a Software Company About Data Security Before You Sign

What to Ask a Software Company About Data Security Before You Sign

The security questions every UK small business should ask before a software supplier handles its customer data, and what good answers sound like.

TH
Taylor Henderson
Agentic Operator
30 October 2026
5 min read

A custom system soon ends up holding your customer records, payment details, documents and call recordings. Most owners never ask a software supplier a single security question before the build starts. These are the questions we would ask, with our own honest answers, including the parts we are still working on.

When you buy a CRM or an AI agent, you are handing over more than a project. You are giving another company access to information your customers trusted you with. If that information leaks or gets corrupted, your name is on the apology letter, not theirs.

You do not need to be technical to ask good questions. A handful of plain ones, asked before you sign, will protect the business and tell you a great deal about how the supplier actually works day to day. The answers matter, but so does the way they are given.

Are you registered with the ICO, and are you certified?

Most UK organisations that process personal data need to pay the data protection fee and register with the Information Commissioner's Office. A software company building systems that hold customer records is processing personal data. You can search the ICO's public register yourself in a couple of minutes, so there is no reason to take anyone's word for it. NTSF is registered with the ICO.

Certification is the next question. Cyber Essentials is a government-backed scheme that checks the basics are in place: firewalls, secure settings, controlled access, malware protection and keeping software up to date. IASME Cyber Assurance goes further, looking at how a business manages risk, handles incidents and runs security as an organisation rather than as a set of settings.

Here is our honest answer. We are working towards Cyber Essentials and IASME Cyber Assurance. We are not certified yet, and we will not tell you otherwise.

Insight
Straight answers are a good sign

A supplier who tells you plainly what they have and what they are still working towards is giving you a useful signal. If a company will blur the truth about a certificate to win your work, think about what else it might blur once it has your data.

Where do you test changes before they go live?

Ask whether fixes are made directly on the live system. If the answer is yes, your real customer records are the test bed. A bad change can corrupt data or expose it, and you will find out when a customer does.

How we fix a bug in a client's system

01
Work on a copy

The fix is made on a separate copy of the system that holds fake data, never real customer records.

02
AI coding agents do the work

Our AI coding agents find the problem and make the change inside that copy.

03
Recorded on video

The work is recorded so there is a clear record of what was changed and how it behaves.

04
A person approves

Nathan reviews the change. Nothing goes live until he has approved it.

You do not have to copy this exact process. What you want to hear is that testing happens somewhere other than your live data, and that a named person signs off before anything changes.

Who is watching the system at night?

Problems do not keep office hours. Ask who would notice an attack, a crash or a security weakness at 3am on a Sunday, and how. A small supplier saying "we check it in the morning" is an honest answer, but it is worth knowing before you sign.

We run NTSF on our own AI agent, Jarvis. Alongside handling WhatsApp, drafting emails, managing the calendar and reconciling our finances, it watches security around the clock.

What Jarvis keeps an eye on

01
Cloudflare

Traffic reaching our systems, so unusual activity is spotted rather than missed overnight.

02
Error monitoring

Errors as they happen, so a failing feature is picked up before a client has to report it.

03
Code scanning

Our code, checked for known weaknesses on an ongoing basis rather than once a year.

How do you handle sensitive documents and AI agents?

Plenty of businesses still collect ID, insurance certificates and contracts as email attachments. Those files then sit in inboxes indefinitely, get forwarded on, and are hard to account for later. Ask your supplier how documents get into the system.

Two ways to collect a sensitive document

Where the file ends up
Email attachment: In inboxes, sent folders and wherever it is forwarded
Secure time-limited link: Uploaded straight into the system, against the right record
How long it is exposed
Email attachment: Indefinitely, unless someone remembers to delete it
Secure time-limited link: The link stops working after a set period
Who can reach it
Email attachment: Anyone with access to any mailbox it passed through
Secure time-limited link: Only people with access to that record in the system
Tracking it later
Email attachment: Searching through old emails
Secure time-limited link: Stored in one known place

For PM+, a surveying business we work with, compliance documents such as DBS checks and insurance are collected through secure links that expire after 48 hours. Subcontractors upload what is needed, and nothing is left floating around in email.

AI agents raise a newer question. If a supplier is building you an agent that can read your CRM, ask whether it can change records on its own. Our view is that it should propose changes and a person should confirm them first. That is how the agents we build work.

How a change made through one of our AI agents goes through

Request made
Agent proposes change
Person confirms
Record updated

What a good answer sounds like

A good answer is short and specific. It names a process, a person or a status, and it admits gaps. A vague answer, or one that changes the subject to features and price, is the warning sign. You do not need to judge the technical detail. You only need to notice whether the supplier can explain it plainly.

Questions to ask before you sign

  • Are you registered with the ICO? (Then check the public register yourself.)
  • Do you hold Cyber Essentials or IASME Cyber Assurance, and if not, where are you with it?
  • Are fixes made on the live system, or on a copy with fake data?
  • Who approves a change before it goes live?
  • Who watches the system out of hours, and what are they watching?
  • How are sensitive documents collected, and do upload links expire?
  • Can an AI agent change records on its own, or does a person confirm first?

We are happy to be asked every one of these. If you would like to put them to us in person, that is exactly what our free initial consultation is for.

Ask us the hard questions before you sign

Book a free consultation and ask us anything about how we would look after your data.

[email protected]