Most agency process pages are decoration — five vague verbs in a curved diagram. This one is the actual playbook. If you hire us, this is the sequence the work runs through, the artefacts we hand over at each step, and the moments where you have to decide things. We are publishing it openly because the process is most of why projects succeed or fail, and because keeping it private has not been protecting any secret worth protecting.
The Five Phases at a Glance
- Discover — understand the business, audience, constraints and current digital surfaces.
- Define — translate the evidence into a written brief and explicit priorities.
- Design — move from structure to system through architecture, wireframes, visual design and components.
- Deliver — hand off or build, test, document and launch with appropriate controls.
- Iterate — measure against the outcomes agreed in Define and improve. Ongoing.
Phase 1 — Discover
The first phase is the one most agencies skim, which is why most projects end up redesigning the wrong thing. Discovery is not workshop theatre. It is a small number of structured conversations and a serious audit of what is actually live.
Stakeholder interviews
Focused conversations with the people who own the outcomes, maintain the system and handle customer enquiries. The number and format depend on scope.
Audit of every existing surface
Marketing site, product UI, brand assets, email templates, paid landing pages, anything customer-facing. We document what is working, what is broken, and where the inconsistencies are.
Competitive and category read
What the closest comparable businesses are doing well. Not so we can copy them — so we can spot what is unowned and where the brand can credibly stand.
A written discovery record covering audit findings, interview themes, category context and agreed priorities. The format should be only as long as needed to support a sound scope decision.
Phase 2 — Define
Define is the shortest phase and the highest leverage. This is where the discovery findings turn into a project brief that says, in plain language, what the work will achieve, what surfaces it will touch, and how we will know it worked.
The reason this phase exists separately from Discover is that the same audit can support several different briefs. Define is where you choose which one. A weak product UI plus a tired marketing site plus a stretched brand could lead to a brief that focuses on product, on marketing, or on brand. We force the choice in writing, with you, before any design work starts.
Outcome statement
One sentence that names the commercial result the project has to produce. Concrete enough to argue about. Not "improve the website".
Surface list, in priority order
The specific pages, screens or assets that will be touched, sequenced by impact on the outcome statement.
Success metrics
A small set of measurable signals, a documented baseline and an agreed review window. The metric must map to the intended outcome and available data.
Scope boundaries
What is explicitly out of scope. This protects the project from feature creep more than any change-request process ever will.
Phase 3 — Design
The design phase has its own internal sequence, and the sequence matters. Visual design before structure is one of the most common reasons digital design projects fail. Structure first, polish later.
Information architecture
Site maps, navigation, page hierarchy, content models. We sign this off before any pixel work happens, because rearranging the IA later is the most expensive change you can make.
Low-fidelity wireframes
Greyscale layouts that prove the page can carry the message without leaning on visual polish. If a page works in wireframe, it will work in colour. The reverse is not true.
Visual design system
Type scale, colour palette, spacing tokens, component states. Built as a system, not as page mockups, so the result extends cleanly into future surfaces.
High-fidelity screens
The actual pages or screens, assembled from the system. This is the moment that looks like "design" to most people. By the time we get here the hard decisions are already made.
Component library handover
A documented set of reusable pieces — buttons, cards, forms, navigation patterns — that the development team builds against. This is what makes the system survive past launch.
Phase 4 — Deliver
Delivery is where the design becomes something a customer can use. We do not always build it ourselves — when an in-house development team owns the build, our job in this phase is to support them, not replace them.
Build, in priority order
The highest-impact surface ships first, on its own, so the team and the business get a working asset before the rest of the work is done. No big-bang launches.
Real-condition testing
The site or product is tested on the devices and connection speeds your actual users have, not on the studio's M1 Mac with fibre. This is the difference between a launch that works and one that does not.
Phased launch
Where the surface allows it, the new design goes live to a subset of users first. We watch behaviour, fix the obvious, and then go broad.
Search and tracking continuity
For the marketing site, this is where redirects, sitemaps, structured data and analytics are double-checked so the SEO position is preserved through the launch.
Phase 5 — Iterate
After launch, review the metrics and evidence window agreed in Define. Where a signal has not moved, inspect the likely contributing surfaces before proposing another change. Do not treat correlation after launch as proof that the design caused an outcome.
This is not "phase 2 of the redesign". It is a steady cadence of small improvements to the surfaces that are not pulling their weight. Most of the long-term return on a digital design investment comes from this phase, not the launch itself.
What Determines the Timeline?
There is no honest universal project duration. Timing depends on the number of surfaces, content readiness, stakeholder availability, technical dependencies, migration risk and the review process. A proposal should show milestones, client decisions and dependencies rather than promising a generic launch week.
What a client should prepare
- A decision-maker and named day-to-day contact.
- Access to the current site, analytics and relevant brand assets.
- A list of priority audiences, offers and known constraints.
- Existing research, enquiry evidence and content that can be verified.
- Realistic review windows and a route for resolving conflicting feedback.
What We Do Not Do
- We do not do six-week "discovery" workshops with no audit underneath. If you have heard the term "design sprint", we do not sell those.
- We do not do open-ended retainers without a defined outcome statement.
- We do not redesign things visually when the real problem is structural.
- We do not hand off Figma files and walk away — the component library and the build relationship are part of the deliverable.
Frequently Asked Questions
Can we skip Discover if we already have a brief?
Sometimes. If the brief is recent, evidence-backed and has stakeholder buy-in, Discover compresses to a one-week sanity check. Most briefs do not survive that check intact.
How involved do we need to be?
Heavy in Discover and Define — at least one decision-maker available for 2 hours per week. Lighter in Design (an hour a week for reviews). Available in Deliver for content sign-offs and launch decisions.
What if the brief changes mid-project?
We re-open Define rather than absorbing changes silently. It costs a few days and saves weeks of drift later.
Do you work with our existing dev team or build it yourselves?
Both, depending on the project. Where the dev team is in-house, we deliver against their stack. Where there is no team, we build through trusted partners.
Want to see this process applied to your current setup?
The free audit is a one-day version of the Discover phase. You get an honest read on the strongest and weakest surfaces and a sequenced plan for what to fix first.
Request the free audit