Thirty-five years turning design and emerging technology into experiences customers come back to.

The same interface resolved through the same semantic tokens. Switch it and watch what stays fixed — hierarchy, state color, contrast, density. That is a dual-mode system rather than two separate skins.

Every capability the portal did, was asked to do, or had been promised — put through one model, then cut by a threshold the executive team set itself.

Four hundred and fifty spreadsheet rows hide the shape of a portfolio. Here it is plotted — value against ease of delivery, sized by reach, colored by persona. Move the weight and watch the funding line rotate: the argument worth having is about the weight, not the list.

The portal serves three people who want different things from it. Each got its own weight set, its own ranking of the same 450 items, and its own roadmap view. Select a persona to isolate it.

What cleared the line, ordered by dependency into four horizons and four tracks — the view a board can fund, on top of the sequence a delivery team can run.

One workbook, eight sheets. Three are the scoring lenses, one is the answer, and four are what make the answer defensible six months later — the register, the criterion definitions, the KPI per initiative, and the capability gap each one has to close.

Each initiative was scored independently in three sheets before anything was averaged. Keeping them apart shows which lens is carrying an initiative and which is sinking it.

All fourteen initiatives as the workbook scored them. Re-sort and watch it re-order — the composite is what got funded, but the three lenses underneath disagree, and the disagreement is the finding.

Ten chapters building one argument in order: why experience is now a competitive position, what best-in-class looks like and where it falls short, what the investment is worth, the architecture and organization to deliver it, and how the return gets proven.

The chapter the engagement turned on. Experience initiatives rarely fail on design — they fail on the org chart, where function, line of business, and channel each own a piece and measure something different. Select a model and watch the wiring change.

Companies design experiences continuously and measure them annually. The assessment replaces that with a seven-step method — and the step most skip is normalization, which stops four disagreeing sources being averaged into a number nobody trusts.

Four categories of source feed it. Select one.

And a named owner for the reporting, because reports lapse when nobody is accountable for them.

Three workstreams in sequence, each feeding the next: understand the market and the audience, decide what to do with what already exists, then define the architecture that carries it.

Six weeks stayed six weeks because both lists were written down before the work started: what was defined precisely enough to analyze, and what the business wanted but held for a later phase — a not-this list agreed in advance.

Ten inputs requested before any outside data was layered on top — analytics, CRM, research, call center, NPS. The market view was built around what the company already knew about itself.

Twenty-eight pages across six areas of the statement experience. Not six features — six parts of one control system, each drawn in every state it moves through.

The core of the redesign, rebuilt here so you can use it. Switch the card context and watch controls appear and disappear — a corporate card gets no points or receipts, and a card with no supplementaries has nothing to break spend down by. Change the time control and the period follows.

The heaviest chapter in the set. Tags are the cardmember’s own vocabulary laid over the bank’s — select one and watch the total recompute. Tag a second charge from the same merchant and the advisor offers to do the rest, once, reversibly.

What the proof of concept had to demonstrate, end to end. Each stage shows what happens and a schematic of the screen that carries it.

One object, several products, everything each product needs hanging underneath. This is what the portal had to let a client build unaided — deliberately the same structure the back office already used.

The deck closes on questions rather than conclusions — the six the proof of concept left on the table, named so they got answered in design rather than discovered in build.

One page: seven phases across the top, five swim lanes down the side, every cell filled in before day one. Select a phase for its intent, what happens in the room, and what it produces.

Session one walks the method end to end, grounded in the d.school model and carried one stage further than most versions of it. Select any stage.

The blueprint names the facilitation method explicitly: a non-linear process. The stages are drawn in order and deliberately not run in order — a room marching through six boxes has stopped doing what the boxes are for.

The second and third sessions exist because design thinking installed beside an existing delivery model gets ignored by it. Three practices, brought into one operating model.

Each tile is a component with its own design, iteration, and sign-off, sized by how much of the archive it accounts for. The four largest are not member-facing screens — they are the shared foundations that let the other twenty-six stay coherent.

The composition of the archive says more than any summary would. Editable wireframe source rather than exported pictures; layered originals beside the flattened comps; a type system packaged so everyone rendered the same faces.

Iteration is not overhead here — it is most of the archive.

And it ran for years before it peaked, which is what a genuine revamp of an entire online experience looks like from the outside.

The pattern the engagement was built around, reconstructed so you can use it. Turn the spread, tap a point to shop the look, then switch to the grid — the same assortment read two ways, because browsing and hunting are different jobs done by the same person minutes apart.

Why this was a priority rather than a project. Three channels against what a customer actually needs from them. The store keeps one thing outright and no amount of digital replaces it; everything else is where the shift happens.

A shoppable catalog looks like a photograph with dots on it. Underneath it is six problems solved together — and the one that decides whether it survives its second season is not the one customers see.

What the organization needed to know, in its own framing — and a deliberately narrow scope to answer it properly rather than broadly.

Every registration touchpoint mapped and scored against the same nine criteria, so the ranking was arithmetic anyone could re-derive rather than a matter of taste.

Then each touchpoint was valued twice: what it is worth to the person going through it and what it is worth to the institution are not the same number.

Behind the pilot sits one method applied the same way to every kind of work — Discover, Design, Develop, Delivery. Select an offering for what each phase contains.

Four objectives set the boundary of the work.

Each study answers a question the next one depends on. Select a phase to see what happens inside it and what comes out.

The enabling study ran fourteen workstreams in parallel, each producing a piece of the reference architecture and a line in the business case. Select any of them.

Three scenarios were built out to show what re-imagining actually produces, rather than describing it. Select a scenario.

Payer spending in that period concentrated on five themes. Select one.

Five working sections, in an order that put the organization’s own position first. Select any of them.

The framing underneath the whole conversation: five digital technologies had matured far enough to change a business, and their value shows up where they land on the business model, not in the technology.

The closing mechanism, named rather than left to a follow-up meeting.

The architecture put against it: a member view assembled from what the plan holds, what can be derived from it, and what happens outside it.

Six use cases, each one a decision somebody in the plan has to make anyway — currently without the whole picture.

The engagement framework spans the whole member lifecycle rather than the acquisition end of it. Select a stage.

The home screen was reorganized around what an employee wants to know when they open it, and each column was given one of those questions to answer.

Twelve modules make up the screen. Select any of them to see what it does and the usability decision behind it.

The experience had to work across devices, so the design was delivered at two sizes. Switch between them and watch the modules move.

Carried on both layouts Only in this layout Steps back at this size

Three properties, and the argument rests on all three holding at once.

Foundations and accelerators recombine into disruptive scenarios. Select any point on the curve.

Each one starts from a single premise and follows it into every industry it touches. Select a scenario.

The ecosystem of stakeholders that scenario reaches — and the point is how few of them think of themselves as being in that industry.

What follows if it happens.

And how it gets re-imagined by the people who move first.

What an organization actually has to change to be on the useful side of any of this. Select any enabler.

Sense and respond assumes an analytic foundation, and the foundation has three stages that have to be built in order.

Seven elements hold their position on every screen, which is what makes an unfamiliar surface learnable in one session. Select any of them.

The whole purchase, from picking up an existing quotation to a policy in force. Step through it — the schematic rebuilds for each screen.

Nineteen decisions that make a regulated financial product buyable on an unattended table. Filter by where each one sits in the experience, and select any card.

Not making it prettier. Optimizing an experience means reducing the work a person has to do — measured in steps, in clicks, in what they have to hold in their head, and in how much they have to re-learn on each screen.

Each token carries a meaning rather than a color. Dark left, light right — the role is identical, only the value resolves differently.