Security checklist before putting a system into production
A pre-production security checklist with an owner and expected evidence for each item, for managers to walk through with the tech team before go-live.
Quick answer
A pre-production security checklist is the list of checks a system must pass before it receives real users and real data. It covers login and roles, passwords and keys, environment separation, tested backups and event logging. Each item has an owner and a piece of evidence someone can show in the go-live meeting.
What you will find
- Why the security review tends to slip until after go-live
- What a go-live checklist covers and what belongs to a later stage
- A checklist with an owner and expected evidence for every item
- How to run the go-live meeting without mastering every technical detail
- What to do when the date is fixed or the system comes from a vendor
On this page
- What is a pre-production security checklist?
- Why does the security review slip until later?
- What goes into the checklist and what stays out?
- Which checklist should you use in the go-live meeting?
- How do you run the go-live meeting?
- What if the date cannot move or the system comes from a vendor?
- When is this not enough?
- Frequently asked questions
The go-live date was set by the board two months ago. The new work order system will replace the spreadsheet on Monday, and the field crew has already been trained. On Friday, someone in the meeting asks whether security has been reviewed. The tech team says yes, broadly, and that a few points will go into the next version. Nobody in the room can say which points, or who will handle them.
This article gives you a security checklist for that meeting. Each item comes with an owner and the evidence a manager can ask for on the spot, without having to master the technical detail. If the system in question was built with an AI tool by someone in the business area, start with the guide to reviewing AI-built systems, which covers that specific case.
What is a pre-production security checklist?
A pre-production security checklist is the list of checks a system must pass before it receives real users and real data. It turns "security has been looked at" into questions that ask for proof: show where the database password is stored, open this page without logging in, restore this backup.
The most widely used reference for building this kind of list is the ASVS, the OWASP Application Security Verification Standard. OWASP is a nonprofit foundation that publishes open material on software security. Version 5.0 came out in May 2025 and brings together hundreds of requirements, organized in chapters such as authentication, authorization, configuration and logging. The project itself suggests using the standard as a yardstick for system owners to gauge how much trust they can place in an application, and as a basis for requirements in contracts.
The full ASVS is work for a specialist. A go-live checklist takes the part that most often brings down internal systems and writes it in language a manager can follow up on. The technical detail stays with the team, and the decision to release stays with whoever answers for the operation.
Why does the security review slip until later?
The review slips because the go-live date is set by the business and security has no visible deliverable on screen. A late feature shows up in the demo. A weak configuration shows up nowhere until the day someone finds it.
Public data shows this gap is common. In the 2025 edition of the OWASP Top 10, the list of the most common flaws in web applications, security misconfiguration moved up from fifth to second place. According to OWASP, 100% of the applications tested had some form of misconfiguration: default accounts with the original password, test features left in production, error messages that reveal server details, cloud storage open to the internet.
The other side is how fast attackers move. Verizon's 2026 Data Breach Investigations Report found that 31% of the breaches analyzed started with the exploitation of a software flaw, ahead of stolen credentials for the first time. The report also notes that AI tools are shortening the time between finding a flaw and attacking it. A system that goes live with an outdated component or an exposed panel has no months of slack.
There is also a less visible cost. When something goes wrong and there is no record of what was checked before go-live, the company cannot show a client, an auditor or its legal team that it took care. A checklist with evidence solves that too, because it becomes the record of the decision.
What goes into the checklist and what stays out?
The go-live checklist covers what has to be right on the first day of real use. Anything that needs a specialist, time or its own budget goes into a second stage, with a date set.
| Part of go-live | Later stage |
|---|---|
| Mandatory login and roles that limit what each person sees | Penetration test by an external team |
| Passwords and keys out of the code, separate production credentials | Certifications such as ISO 27001 or SOC 2 |
| Database and admin panels out of reach from the internet | Line-by-line source code review |
| Automatic backups and one test restore already done | Full business continuity plan |
| Logging of logins and key changes | 24-hour security monitoring center |
| Agreement on who acts in an incident | Formal incident response plan and regulator notification |
The right column still matters. It just should not block the release of an internal system once the left column is done, and it should not be an excuse to skip the left column. For personal data, the article on LGPD in internal systems explains what to review in screens, reports and logs. LGPD is Brazil's general data protection law, similar in spirit to the GDPR, and its regulator is the ANPD.
Which checklist should you use in the go-live meeting?
Use a short checklist, split by owner, where each item asks for concrete evidence. The manager reads the item, the owner shows the evidence and the item is marked done or open with a deadline.
Security checklist for releasing a system to production
Walk through the groups in a meeting, with the business owner and the technical lead at the same table. For each item, ask for the evidence listed: a screen, a setting, a test done on the spot. Any item without evidence stays open, with an owner and a deadline. The checklist covers the minimum for go-live and does not replace a penetration test.
Access and roles (owner: tech team with the business owner)
Secrets and environments (owner: tech team)
Data and continuity (owner: tech team with IT)
Logging and response (owner: business owner with the tech team)
Some items look too simple to be on a security list. Those are exactly the ones that show up most when we review systems already in use: the report page that opens without login because nobody tested it outside a session, the "admin" account created on the first day of development and never deleted, the backup that exists but has never been restored.
Adapt the list to your operation. A system that only runs inside the company network may handle the internet exposure item differently. A system with health data or data about minors needs extra items about who sees each record. The item-plus-evidence format holds in any case.
How do you run the go-live meeting?
Run the meeting as a walkthrough, with the list on screen and the system open next to it. The goal is to leave with every item marked done or open with an owner and a deadline.
- Send the checklist to the tech team a week ahead, so the evidence is ready and the meeting does not turn into a discovery session.
- Book an hour with the business owner, the technical lead and someone from IT. If a vendor built the system, the vendor joins.
- Go through the groups in order. For tests that can be done on the spot, such as opening a page in a private window or triggering an error, do them in front of everyone.
- Classify each open item as blocking, which holds the release, or acceptable for a short period, which goes live with a date to fix.
- Record the decision in simple minutes: what was checked, what is open, who fixes it and by when.
- Schedule a review 30 days after go-live to close open items and check that logs and alerts work under real use.
Step 6 is the one people forget. It is where the problems that only appear with real users come up, such as roles created in a hurry for people who missed the training.
In the systems we build, we treat change history and role-based limits as requirements from the start. In the access control system for an industrial complex, for example, each tenant company sees only its own records and the history keeps who approved or denied each access.
What if the date cannot move or the system comes from a vendor?
If the date cannot move, release with a smaller scope instead of releasing with a blocking item open. You can start with one site, with a smaller group of users or without the modules that handle sensitive data, while the remaining items are fixed.
When a vendor supplies the system, the checklist becomes a list of questions for the vendor. Ask for the same evidence, in writing or in a meeting. A serious vendor can show where credentials are kept, how backups are made and what the system logs. A vague answer is already information for the decision, and the article on custom software projects helps you put these points into the scope of your next contract.
If the system is old and has never been reviewed, the checklist works as a first assessment. Items such as "no password written in the code" can take more work in a legacy system, and sometimes the fix belongs to a staged modernization rather than a one-off patch.
Then there is the system used in the field, by crews on platforms, in ports or on construction sites, with weak signal. In those cases, also check what stays stored on the phone or tablet when it is offline and what happens if the device is lost.
When is this not enough?
The checklist is not enough when the system holds data that would cause serious harm if exposed: health data, financial data, client information under a confidentiality agreement, or data shared among several companies. In those cases, a penetration test by people who did not build the system is the next step.
It is also not enough when the company has to prove compliance with a specific standard to a client or auditor. The list helps put the house in order, but the evidence a formal audit requires is usually more detailed. Bring in whoever is responsible for information security and, when personal data is involved, the data protection officer and the legal team.
And there is the limit of logging. OWASP ranks logging and alerting failures among the top ten because, without logging and alerting, an attack can go on for months unnoticed. Logging logins and changes is the minimum. Knowing who reads those logs and what triggers an alert is a process decision that goes beyond the system.
If you would rather have an outside team walk through this list with yours, boosting.tech, a custom software company, offers an internal systems security review with a fixed scope. The assessment points out what to fix first and what can wait. The first conversation is free and can be booked through the contact page. For systems that need ongoing fixes after that, see our support and modernization service.
Frequently asked questions about pre-production security checklists
Who should approve a system going into production?
The owner of the business area that will use the system approves it, with input from the technical lead and IT. The tech team shows the evidence and points out risks, but accepting an open item for a short period is up to whoever answers for the operation. Writing that decision in the minutes protects everyone involved and keeps the open item from being forgotten after go-live.
Do I need a penetration test before every release?
Not necessarily. For an internal system with low-risk data, a well-run checklist covers the essentials on the first release. A penetration test makes sense when the system holds sensitive data, is reachable from the internet by clients or third parties, or when a contract requires it. In those cases, plan the test before the go-live date, because fixes take time.
Does the checklist work for systems already in production?
Yes, and it usually reveals more open items than on a new system. Go through the same items with the system in use, prioritize access and secrets, and treat the result as a remediation plan with deadlines. For systems built with an AI tool by someone in the business area, the guide in the article on AI-built systems complements this list.
How long does it take to go through the checklist?
The meeting itself fits in about an hour when the tech team prepares the evidence beforehand. What takes longer is fixing what turns up. That is why it pays to send the list in advance and, on new projects, to use the checklist from the start of development, so the items are handled during the build and not the day before go-live.
What if the vendor cannot show the evidence?
Treat missing evidence as an open item, with a deadline. Ask in writing how each item is handled and set a new meeting. If the vendor cannot explain where credentials are kept or how a backup is restored, the risk is real and should weigh on the decision to release, to reduce the initial scope or to seek an independent review of the system.
References
System support and modernization
We take over existing systems and modernize them in parts, without stopping the operation.