This guide is for founders, CTOs, security leads and procurement teams shortlisting a partner that will write code for them. Our software development outsourcing guide covers cost, team models and the wider vendor checklist; this page goes deeper on one row of it: security. It is general guidance, not a security audit.
In this guide
- Why a development partner needs different questions
- Twelve questions and the evidence to ask for
- What certificates prove, and what they do not
- How to run the security review
- Red flags
- AI coding tools: three questions to add
- Alternatives and selection criteria
- How Netbase answers these questions
- What delivery record exists, and what does not
- Limits of this guide
- Frequently asked questions
- Next step
Why a development partner needs different questions
A software vendor sells you a finished product. A development partner builds your product with you, which changes where the risk sits. Its engineers push code to your repositories, deploy to your cloud accounts, read your specifications and sometimes see production data. Their laptops, accounts and tools become part of your attack surface, and the code they write becomes your product's security.
So a partner review has two halves. The first is enterprise security: how the company protects itself, its staff and its systems. The second is product security: how the software it builds for you is designed, reviewed, tested and maintained. CISA and the FBI make the same distinction in their Secure by Demand guide for software buyers, and they stress product security, not only the vendor's own controls. Most generic vendor questionnaires cover the first half well and the second half poorly.
Twelve questions and the evidence to ask for
| Question | What a good answer includes | Evidence to ask for |
|---|---|---|
| How is security built into your development process? | Security requirements set in discovery, threat thinking in design, checks inside every sprint rather than at the end | A description of the process mapped to a public framework such as the NIST SSDF |
| How is code reviewed before it merges? | Every change reviewed by a second engineer, protected main branches, reviews visible to you | Repository settings and a sample of real review history |
| Which security tests run, and when? | Dependency and code scans on every build, penetration tests before major releases, findings tracked to closure | A recent scan report and how findings were fixed or accepted |
| Which security standard will you build against? | A named baseline, such as selected OWASP ASVS requirements, agreed as acceptance criteria | The requirement list proposed for your project |
| Who gets access to our systems, and how? | Named accounts, MFA, least privilege, access removed the day someone leaves | The onboarding and offboarding procedure, and a sample access review |
| Where will our code, secrets and data live? | Your repositories and cloud accounts, secrets in a vault, no production data in development without approval | The proposed environment and secrets set-up |
| How do you handle personal data? | Data minimisation, anonymised test data, a data processing agreement where the law requires one | The DPA template and the data flow for your project |
| Which third-party and open-source components will you use? | Components listed, licences checked, known vulnerabilities monitored | A software bill of materials (SBOM) with each release |
| Who else works on our project? | Employees by default; any subcontractor disclosed and bound by the same terms | The list of contributors' employers and the flow-down terms |
| What happens when you find a vulnerability or an incident? | A named contact, a notification time agreed in the contract, a written review afterwards | The incident process and a sanitised example of a post-incident review |
| How do your engineers use AI coding tools? | Approved tools only, no training on your code, human review of every AI-suggested change | The approved-tools list and the data settings of those tools |
| Which certifications or reports cover your company? | Named credentials, what they cover, and willingness to share details under NDA | Certificate or report details, requested during due diligence |
Two items on that list are newer than most questionnaires. On components, CISA, the NSA, the FBI and international partners published the 2026 Minimum Elements for a Software Bill of Materials on 29 July 2026, replacing the NTIA elements of 2021; asking for an SBOM is now a mainstream request, not an exotic one. On AI tools, see the section below.
What certificates prove, and what they do not
ISO 27001 certification and a SOC 2 report are useful signals, but they answer narrower questions than buyers often assume.
- They cover the company, not your product. An information security management system or a set of audited controls says how the vendor runs itself. It does not certify the application the vendor builds for you, or your hosting.
- Scope matters. Ask which entities, locations and services the certificate or report covers, and whether your delivery team works inside that scope.
- Type matters for SOC 2. AICPA's illustrative SOC 2 Type 2 report shows that a Type 2 report includes the auditor's tests of controls and their results over a period, not only a description of the controls.
- Dates matter. A certificate expires and a report covers a past period; ask for the current one.
A partner without a certificate can still practise security well, and a certified partner can still ship insecure code. Use certificates to shorten the review, never to skip the product questions.
How to run the security review
-
Tier the partner by risk
A team that will touch production data or payment flows needs the full review; a design-only engagement needs less.
-
Send the questions before the meeting
Written answers give you something to verify and compare across vendors.
-
Ask to see, not to hear
Screen-share the repository settings, a pull request, a scan report and an access review.
-
Talk to an engineer, not only to sales
Ask the person who will lead your project to walk through how a change reaches production.
-
Put the answers in the contract
Security requirements, the SBOM, incident notification, the DPA and subcontractor terms belong in the agreement, not only in the proposal. CISA's guide makes the same point: build product security into contract language.
-
Start small and check
A short first engagement shows whether the practices described are the practices used.
-
Review again after launch
Access lists, dependencies and people change; repeat the review at least yearly.
Red flags
- Answers that name tools but cannot show their output.
- Shared logins, or access that is never removed.
- Your code held in the vendor's repositories and accounts, not yours.
- No written process for incidents, or no one who owns it.
- Real customer data copied into test environments by default.
- Security testing offered only as an optional extra at the end.
- A certificate presented as if it covered your product.
AI coding tools: three questions to add
AI-assisted engineering is now normal, including at Netbase, and it adds three questions. First, which tools may see your code, and are they configured so your inputs are not used to train shared models? Second, is every AI-suggested change reviewed by a named engineer and covered by the same tests and scans as human code? Third, may you switch AI assistance off for sensitive repositories? A partner that cannot answer these clearly has not thought about where your code goes. Netbase works with the major commercial and open-source AI tools and models, chosen per project, and every AI-assisted change stays under human review.
Alternatives and selection criteria
| Approach | Strength | Weakness | Use it when |
|---|---|---|---|
| Written questionnaire only | Cheap and comparable across vendors | Answers are claims, not evidence | Low-risk work, or as the first filter |
| Questionnaire plus evidence review | Tests claims against real artefacts | Takes a few hours per vendor | Most development partnerships |
| Independent security assessment of the vendor | Expert, unbiased view | Costly and slower | High-risk data, regulated sectors or large contracts |
| Paid pilot project | Shows practice, not description | Commits time before the main contract | You are choosing between two strong candidates |
Choose by four criteria: the sensitivity of the data the team will touch, whether the vendor will access production, the contract's size and length, and whether your sector sets its own rules.
How Netbase answers these questions
Netbase's security practices are secure code review and version control, TLS in transit and AES at rest, role-based access control, MFA for admin dashboards, vulnerability scanning and penetration testing, and disaster recovery. Contributors work under NDA, and NDAs, data processing agreements and SLAs are available to clients on request. Netbase follows GDPR alignment for data privacy in Europe, HIPAA-aligned methodologies for healthcare data handling and CCPA compliance practices for clients with U.S. customer bases.
Netbase holds ISO 27001 certification and a SOC 2 Type II attestation. Both cover Netbase's own operations, not a client's product or hosting, and certificate details are shared during due diligence rather than published. In Netbase custom development the client owns the IP created for it, while Netbase productized modules and Business Division products are licensed, not transferred; our guide to protecting IP when outsourcing covers those clauses. Most Netbase projects are delivered on fixed-price contracts agreed after discovery, and the same security terms apply to dedicated development teams; see dedicated team vs fixed price for the trade-offs. The full statement is on security and compliance.
What delivery record exists, and what does not
- What exists. Since 2020 Netbase has worked as offshore development and managing partner on a multi-tenant cloud ERP for a US client, sold as SaaS to small and mid-sized businesses; the public record withholds the client's name. Netbase also builds and runs its own SaaS, Printcart.
- What does not. No published Netbase record states penetration-test results, incident history, audit findings or response times, and neither record above is evidence of any particular control. Netbase does not publish certificate numbers, dates, auditors or scope.
Limits of this guide
- It helps you choose a partner; it does not test your application. For that, see application security assurance.
- Regulated sectors such as payments or healthcare add their own requirements; take specialist advice.
- Standards change: check the current ASVS, SSDF and SBOM guidance on the day you review.
- For governing security release by release once work starts, see software quality governance for outsourced delivery.
Plan the next step with a Netbase consultant
Frequently asked questions
Ask about its secure development process, code review, security testing, access to your systems, handling of personal data, third-party components, subcontractors, incident response, AI tool use and certifications, and ask for evidence behind each answer.
No. Both describe how the vendor runs its own security. They do not certify the software it builds for you, so you still need the product questions and evidence.
Yes, for any software you will run in production. A software bill of materials lists the components in each release, so you can track licences and newly found vulnerabilities.
You should. Invite the partner's engineers with named accounts and MFA, so access can be reviewed and removed without the vendor's help.
Next step
Send us your security questionnaire or the questions above, and we will book a solution review to walk through our answers with the engineer who would lead your project. You can also browse more Netbase insights.
Related services and solutions
An AI-augmented dedicated development team in Vietnam that answers for results
Netbase provides dedicated development teams from Vietnam for product companies and enterprises to add engineering capacity without losing control. Teams of 3 to 30 people work only on your product, use AI-assisted engineering under review, report to a named account manager and project manager, typically start within one to two weeks after discovery, and the IP they create is yours.
Learn More
Application security assurance for web, SaaS and AI features
Netbase provides application security testing for product and engineering leaders to find and fix vulnerabilities before release, including in AI features. We review code with AI-assisted triage, scan dependencies and test running applications, then rank findings and retest fixes. Netbase's own ISO 27001 certification and SOC 2 Type II attestation cover Netbase's operations, never your application.
Learn More
SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules
A SaaS product accelerator is a set of reusable Netbase modules for accounts, billing, roles and integrations that helps founders and product teams launch subscription software faster, with room for in-product AI from the first release. Reusing these modules can cut development time by up to 60%, and the approach is proven on Printcart, Netbase's own web-to-print SaaS.
Learn More
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.