Company · Process
How we work: our software development process from first call to support
Every project runs through the same four stages: discover, design, build and run. This page shows what happens in each, what you receive at the end of it, who is involved and how you stay in control of scope and budget.
Short answer
Innovation Insight runs every build in four stages. Discovery takes one to three weeks and costs $3k to $8k, ending in a written scope, a clickable prototype and a fixed estimate. Design follows, then the build runs in two-week sprints with a demo of working software at the end of each, milestones you sign off against written acceptance criteria, and automated tests on every pull request. After launch we support the product on a retainer of $1.5k to $8k per month.
Reviewed by Zohaib Khalid, CEO & Co-founder · Updated
The four stages at a glance
| Stage | Length | What you receive | Who you work with |
|---|---|---|---|
| 1. Discover | 1 to 3 weeks | Written scope, user flows, a clickable prototype of the main screens, an architecture note and a fixed estimate for the build milestones | A delivery lead and a senior engineer, plus a designer |
| 2. Design | 1 to 3 weeks, overlapping the first sprint | Design system and production screens for the first milestones, reviewed with you and, where possible, with real users | Designer and delivery lead |
| 3. Build | Two-week sprints, usually 8 to 20 weeks to launch | Working software on a staging server after every sprint, milestone sign-off against acceptance criteria, release notes | The squad: engineers, QA and the delivery lead |
| 4. Run | Ongoing | Production release, monitoring and alerts, a support retainer, and a roadmap for the next releases | A smaller support team or the same squad |
Discovery is paid and standalone ($3k to $8k). You own everything it produces, so you can take the scope and estimate to another vendor if you prefer. Most clients continue with us because the estimate is specific enough to sign.
Discovery: deciding what to build first
Discovery exists to remove the expensive surprises before the build starts. We read what you already have (briefs, spreadsheets, an existing codebase), interview the people who will use the product, and agree what the first release must do and what can wait.
- Scope written as user stories with acceptance criteria, grouped into milestones of two to six weeks each.
- A clickable prototype of the main flows, so you can test the product with users before paying for the build.
- An architecture note: the stack, hosting, integrations, data model and the risks we see, with the reasons for each choice.
- A fixed estimate per milestone, or a monthly team plan if the roadmap is still moving. See engagement models.
- For an existing codebase, an audit of security, tests and deployment first. The code audit checklist shows what we look at.
Build: two-week sprints with something you can use
Each sprint ends with working software on a staging server, not a status report. You see the demo, try it yourself, and reprioritise the next sprint if something has changed. Milestones group several sprints and end with a sign-off against the acceptance criteria agreed in discovery, which is also when milestone payments fall due on fixed-price projects.
| Every sprint | Every milestone | Every release |
|---|---|---|
| Planning call and a written sprint goal | Acceptance testing on staging by you, with a written list of what passed | Release notes and a rollback plan |
| Daily async updates in your Slack or Teams | Sign-off or a written list of fixes, handled before the next milestone starts | Database migrations tested on a copy of production data |
| Demo of working software | Updated estimate for the remaining milestones | Monitoring and alerts checked after deploy |
The structure of milestones and acceptance criteria is described in detail in software project milestones and acceptance criteria, and our statement of work template shows how it is written into a contract.
Engineering standards on every project
- Code lives in your Git organisation from the first commit, and your cloud accounts hold production. IP is assigned to you on payment.
- Every change goes through a pull request with review by a second engineer and automated tests in CI. On the Paint Nite rebuild that meant 754 automated test files running on every pull request.
- Separate staging and production environments, infrastructure as code where the project warrants it, and deploys that can be rolled back.
- Security basics from the start: least-privilege access, secrets in a vault, dependency scanning and the OWASP ASVS checks relevant to the app. Our security page has the detail.
- Documentation kept in the repository: setup, architecture decisions, runbooks and the API schema.
- AI coding tools are used under written rules agreed with each client; see how we use Claude Code on client projects.
Communication and time zones
Our team is in Pakistan (UTC+5). We schedule shifts around the client rather than the other way round, so stand-ups and demos are booked in your working hours, decisions are written down for the hours we do not share, and you have a named delivery lead who answers for the whole team.
| Your location | Live overlap with our team | Where to read more |
|---|---|---|
| UK and Western Europe | Most of your morning and early afternoon | For European companies, for UK companies |
| US and Canada | 9:00 to 14:00 Eastern, with async cover until your evening | For US companies, for Canadian companies |
| Australia | Your afternoon: our 9:00 start is 14:00 in Sydney, 15:00 during daylight saving | For Australian companies |
People and continuity
You meet and approve the senior people before they start. If an engineer is not the right fit, we replace them free of charge within the first two weeks, and after that with a 30-day notice and a documented handover. On longer engagements we keep a second engineer familiar with each critical area, so a holiday or a resignation does not stop the work.
Run: launch, support and what comes next
Launch is a planned release with monitoring in place, not the end of the project. After launch most clients keep a support retainer of $1.5k to $8k per month covering security updates, dependency upgrades, bug fixes and small improvements, or keep the squad on a monthly plan to work through the roadmap. See software maintenance and support for what each retainer covers.
What this looks like on real projects
- Paint Nite: a live events platform rebuilt beside the legacy system, with sync services so features moved across in slices while events kept selling.
- GovDoc AI: an ongoing engagement from MVP to production for a US GovTech startup, with a free screening tier and a paid deep-analysis tier.
- CareFlair: an Australian NDIS marketplace that went from user research to store release in 16 weeks, then into ongoing support.
Ready to start? Book a 30-minute call or see our published pricing first.
Sources and further reading
Next step
Tell us what you're building and get a written estimate.
A senior engineer replies within one business day. NDA on request.
Products we've shipped, and what happened next.
Case studies written from the technical documentation of each project: the stack, the scale and the outcome.
Questions we get asked a lot.
How much does discovery cost?
Discovery costs $3k to $8k and takes one to three weeks. You receive a written scope, a clickable prototype, an architecture note and a fixed estimate per milestone, and you own all of it.
How often will we see progress?
At the end of every two-week sprint you get a demo of working software on a staging server that you can try yourself, plus daily written updates in your Slack or Teams.
Do we own the code?
Yes. Code lives in your Git organisation from the first commit, production runs in your cloud accounts, and IP is assigned to you on payment.
What if we want to change scope mid-project?
On time and materials or a dedicated team you reprioritise at any sprint. On fixed price, changes are written up as change requests with a price before any work starts.
What happens after launch?
Most clients keep a support retainer of $1.5k to $8k per month for updates, fixes and small improvements, or keep the squad on a monthly plan for the roadmap.
Can we work in our own tools?
Yes. We work in your Jira or Linear, your Slack or Teams, and your GitHub, GitLab or Bitbucket. If you have none, we set them up in your name.