What two years inside a precinct taught us about digital transformation

The question I get asked most about the V&A Waterfront work is not a technical one, it is some version of "how do you even start", and I have come to understand why: from the outside a precinct looks like one client with one website, and from the inside it is closer to a small city that happens to have a marketing department. Two years and two very different halves later, the things that turned out to matter were almost never the things I expected to be writing about.

So this is not the case study, which covers what we built and what it earned. This is the shorter, more useful list: what a business of that complexity, in that industry, taught a studio that thought it already knew how to run a transformation.

A precinct is not a client, it is a constituency

The numbers give you the shape of it. Around 24 million visitors a year across 123 hectares and ten districts, and behind that roughly 3,000 tenant relationships across 800 businesses spanning hotels, retail, commercial and more. What those numbers do not tell you is the part that reorders your thinking, which is that the tenants are simultaneously your customer and your operator. They are who the work is for, and they are also the people who have to keep it current on a wet Tuesday in July.

Once you see that, a whole category of obvious ideas stops being obvious. The visitor-facing feature that the brand would love and no tenant has the time to maintain is not a good idea with an implementation problem, it is a bad idea that photographs well, and the only reason it survives as long as it does is that the people who would have to run it are usually not in the room when it gets approved.

The honest diagnosis was the boring one

Before anything got sequenced we read the estate through our Digital Asset Value Model, which scores a digital asset across four dimensions and then does something deliberately awkward with the result: instead of averaging them it takes the lowest, on the argument that an asset performs at the level of its weakest dimension because the other three can only spend what that one lets through.

At the V&A the weakest dimension was data, comfortably, and the estate's real problem was not that any one system was bad. Four public websites sat on three different content systems, on-site information was static signage and a printed map, and tenant and visitor records lived in two databases that had never spoken. Every piece had been built competently. None of them had been built together, so nothing was reusable and nothing was comparable.

That finding is unglamorous, and it is the reason "put data at the heart of everything" became the commitment for the next two years rather than a slogan someone liked the sound of. The uncomfortable part, and the first genuine lesson, is that an honest assessment will often tell you to spend your first year on plumbing. Getting permission to do that (from a board, from a marketing team with a calendar, from yourself) is harder than the engineering.

Value had to be redefined before it could be measured

Property has a settled, century-old answer to what a tenant is worth, and it is lease economics: rent, term, covenant strength, income stability. It is a good answer to a different question. Nothing in it measures whether a tenant is engaged, collaborative, or an advocate, which means the thing the precinct most wanted to grow was the one thing nobody could see.

Reframing tenant value from a line on a lease into a measurable relationship was harder than anything we built, and I would now treat that as the pattern rather than the exception. In a mature industry the difficult work is rarely technical, it is persuading a business to accept a second definition of value alongside the one it has run on for decades, and no amount of engineering gets you there faster.

Three circles, and the wings where good ideas go to die

Knowing what to fix is not the same as knowing what to build, and for that we used the Product Success Blueprint, which holds that a product only succeeds where three circles overlap: what the business needs, what its customers want, and what the people who operate it can actually live with. At precinct scale those three have names. The brand carries the commercial goals, the visitors are millions of people who arrive for a day out and owe your systems nothing, and the tenants live with whatever gets built every single day.

Plotted against those circles, most of the roadmap turned out to be sitting in one of the near-misses rather than the middle, and the useful thing about the drawing is how quickly it ends an argument. Nobody has to be told their idea is bad; they can see it is in a wing.

With that many stakeholders, opinion does not scale. Scoring does.

What followed is the piece I would now insist on from the start. Every candidate on the roadmap was scored for what it gave the brand, what it gave visitors and what it gave tenants, separately, then weighed against two more numbers (what it takes to stand up, and what it takes to operate forever), then checked against the precinct's own commitments. Nineteen candidates went through that in the second year, and the effect was not that we found better ideas. It was that the roadmap conversation stopped being a contest between advocates and became an ordered list, which is a different meeting entirely.

Do not make marketing wait for the platform

This is where transformation programmes usually come apart, and it took us a while to name it properly. A platform rebuild runs on a two-year arc and a marketing team runs on a calendar, so if you ask marketing to wait for the foundation they will not wait, they will commission another standalone site, and your estate will fragment faster than you can consolidate it. The fragmentation you are being paid to fix is, in part, a symptom of exactly that dynamic playing out over a decade.

What worked was refusing the choice. One shared multi-site framework meant each campaign got a real digital home with its own art direction and its own logic while inheriting the foundations, which is why Makers Landing, Granger Bay, Our Better Nature and Mini Explorers all shipped inside the same year the platform work was still underway. Campaigns bought attention and goodwill while the unglamorous half of the programme was still being poured, and the two stopped competing for the same budget conversation.

In property, digital is the interface to a place

Most of my career has been spent on products people hold, so a precinct rearranged something. Here the digital estate is not only the thing on a phone, it is how somebody finds a shop across 123 hectares, what a screen at an entrance tells them, whether the map in their hand agrees with the sign on the wall. The physical space is the product, and digital is the layer that makes it legible.

That is why the module approach mattered more here than it usually does. Building neighbourhood, what's on and deals once as connected components (rather than three times for three screens) is what will let an entrance kiosk, a phone and a desktop tell the same story about the same place. My favourite thing we designed is also the smallest: the kiosk hand-off that carries a visitor's journey onto their own phone, which deepens the interaction and clears the queue behind them at the same time. Two problems, one idea, and it only got designed that way because somebody was thinking about the physical space rather than the screen.

The numbers did not land where anyone expected

The first year's reporting is the part I would frame on a wall, and not because it is flattering. Clicks were up 64.9% to 790,588, conversions up 286.5% to 22,327, engagement rate up 10.5% to 55.07%, and impressions up 39.1%. Total views over the same period were down 9.8%, and total events down 6.5%.

Read those together and the story is not "we got more traffic", it is that sessions held roughly flat while conversions nearly quadrupled, which is to say the estate got dramatically better at converting the audience it already had. That is the harder of the two wins and the more durable one, and we would have had no way of knowing it happened if the analytics layer had not been built in the same programme. Instrument first, or you will spend a year optimising a number that was never the point and you will not be able to prove any of it afterwards.

What I would tell someone starting one

Five things, and none of them are about technology:

• Find out who actually has to operate the thing, and get them into the room before the approval rather than after it. In a precinct that is the tenants, and they are also your customer, which is the whole puzzle in one sentence.
• Let the assessment tell you something you did not want to hear, then act on it. If your weakest dimension is data, your first year is plumbing, and dressing that up as something more exciting only costs you the second year.
• Expect the hardest work to be a definition, not a build. Mature industries have settled answers about what value is, and a transformation usually needs a second answer to sit alongside the first.
• Run the marketing calendar and the platform arc in parallel off shared foundations. Asking one to wait for the other is how estates fragment in the first place.
• Instrument before you optimise, and when the numbers arrive, publish the ones that went down as well. The declines are usually what make the rises mean anything.

Looking back at two years of it, I am struck by how little of that list is technical, which is roughly the opposite of how I would have written it at the start. The full account of what we assessed, what we ranked and what we shipped is in the case study, numbers and declines included. If you are staring at an estate that grew one good decision at a time and now shares nothing, that is the more useful document. This one is just the part I would want somebody to tell me over a coffee before I started.