Blueprint Registry is one of our longest standing partners. We’re proud to say we’ve been deeply integrated as the technology arm of their business since 2016, when they were a young, seed-funded start-up rapidly gaining market traction. Our partnership began when Nevin Shetty, the innovative wedding registry’s CEO, called us to confirm that his instincts for where to drive the technology were right. (They were.) He was also looking for an outsider’s perspective on how to scale the platform, enhance its feature set, and continue gaining ground on the competition. What was an aggressive but reasonable timeline? Would he need to hire a tech team in-house?

These are big business questions — and precisely where we like to ground our conversations. We know it can be lonely for a CEO to make these decisions without a true technical partner, just like it’s unsatisfying to be on a tech team that’s not invested in real business outcomes.

So, we took on Nevin’s questions as our own. After a discovery workshop to understand Blueprint’s goals and a perusal of the code base, we knew Bilberrry could have a huge impact on their success. Our partnership mentality meant Nevin wouldn’t need to build out an internal tech team; our agile work style meant we could start pushing code in a matter of weeks. We got to work — and have been getting results together ever since.

We became Blueprint’s tech team: we took ownership of every technology decision for the business, from designing the tech stack to collaborating on the feature road map to implementing upgrades, overhauls, and performance improvements. This level of partnership was not built overnight, but began on day one. Both teams set out with a goal of long-term integration and mutual trust — and it’s paid off: in 2018, Blueprint was acquired by David’s Bridal, the largest wedding dress retailer in the world, and we’ve continued to collaborate on Blueprint and David Bridal’s full suite of wedding planning tools.

Bilberrry can build anything. There hasn’t been a single thing we’ve asked of them that they said they couldn’t build, which is incredible. They’ll tell us how many developers it will take, and the cost. Nevin Shetty, CEO

An intimidating recommendation came early on: totally overhaul the platform’s infrastructure

Like a lot of start-up products, Blueprint had been built piece by piece with rapid growth in mind — when you’re getting traction, you build and build fast. It’s the right call, despite the consequences. Those quickly launched components might function fine individually, but rarely work as a cohesive system and are never very easy to maintain. This was definitely true for Blueprint, which had been designed and developed on a proprietary framework by a previous agency. 

When we looked at Blueprint’s 12-month roadmap, we predicted only about a third of the initiatives would be supported by the existing code base. The rest would be painfully slow to implement or, worse yet, might not even work. We knew we’d be spending as many resources in a year shoring up the infrastructure as scaling the platform, and only delaying the inevitable: rebuilding the whole thing from scratch. 

That’s a scary thing to hear if you’re a CEO. We’re talking about a significant investment to build the thing you’ve already built. But it was the only way for Blueprint to get to where they wanted to go. We recommended rebuilding the product on a more stable, better-supported framework like React: it’s big enough, popular enough, and funded enough (thanks, Facebook) that it isn’t going to suddenly vanish. And because it’s so robust, React could truly scale with Blueprint’s growth. 

We stomached the hard choice and rebuilt the framework. In four months. While also maintaining the legacy code. Yes, really. Dual workstreams are never ideal, but Blueprint was still firmly in a rapid-growth phase and the business couldn’t accommodate a full-blown code freeze while we did our thing. It was the right choice. Blueprint continued to deploy the new features required to keep pace with the market during that time, and since we completed the rebuild in 2017, our React infrastructure remains up-to-date and rock solid.

Migrating. So. Much. Migrating.

When you build a new house, you have to move all your furniture; when you rebuild a platform, you have to migrate your data. For Blueprint’s rebuild, that meant user accounts, financial balances — the works. The complexity surfaced some interesting challenges, and collaborating with Blueprint to determine which technical solutions were in the business’ best interest became one of our cornerstone conversations. 

In some cases, like registry products, migrating was as simple as making sure the pipes were connected to the right destination. Other scenarios were more of a scavenger hunt. One example: Blueprint users didn’t originally have accounts — their username was attached to their registry, and that was it. To support Blueprint’s big growth goals, we wanted to upgrade and allow users to have more than one registry — say, for both a wedding and a baby shower — as well as access to new tools we were launching, like a website builder. New users weren’t a problem: they could easily create a new account separate from their registry. But for the 100,000 existing users, we had to plumb the legacy database, extract the relevant information, and slot it into the right parts of the new, more robust user account database we were developing as part of the rebuild. With about two months preparation, we completed this overnight with no data lost. This is the kind of stuff we legitimately think is cool. 

The most complicated migration scenarios came up with legacy features like group gifting. Originally, a group could go in on a registry gift, but were limited to set amounts — for example, a $100 gift could only be split into four $25 increments. New version: anyone could contribute any amount. But there was no way to migrate people with partially funded group gifts in the old system perfectly into the new database. The logic of the systems weren’t compatible. This is such a good example of how technical decisions are, at their core, business decisions, and why both sides need to trust each other. We asked and answered: How much would it cost to maintain both systems? How fast could we deprecate the old version? How damaging would it be to simply cut off access to the old system? Ultimately, our solution was to maintain the old system for two years — long enough that the events for the existing registries could pass, but not indefinitely.

Getting to market fast with a low-cost proof of concept

Blueprint Registry is one of only a handful of universal registries. Rather than registering at Crate & Barrel and Target and Amazon, you could build one single registry on Blueprint and curate products from anywhere on the web. Pretty slick.

An early initiative was to reduce switching costs for potential users who’d already created those individual registries: Crate & Barrel and Target and Amazon and Blueprint? No thanks. We asked how we could get those users to join Blueprint after they’d already sunk time into other platforms. Our hypothesis: an auto-import option that would scrape any existing registry and recreate it on our platform. 

Straightforward enough, but it’s a pretty substantial development project — we were looking at two or three months of intense work. As thought partners, we didn’t launch straight into building feature sets and setting timelines. We needed a proof of concept to some quick, cheap answers: Would enough people use the functionality to pay for it? Which retailers should we prioritize? How could we beat the competition to market? 

Instead of building the full import tool, we added a simple import button: click it to import any existing registry within 24 hours. To our users, it certainly looked like we’d built the tool. But just like Mrs. Doubtfire never actually cooked a real dinner, behind the scenes we were manually importing those registries. In a week and with a tiny budget, we spun up an easy-to-use admin panel, and hired and trained a small team to rebuild the registries on Blueprint by hand. This short-term solution did double duty, both providing the data we needed to guide our future development efforts, as well as allowing us to be one of the first universal registries to offer this functionality. 

Within three months, we had the information we needed to build and launch fully automated import functionality. An impressive 75% of users had clicked the import button, confirming there was enough audience to support the investment. And, we knew that Amazon and Crate & Barrel registries were the most popular — our list of which ones to prioritize and roll-out built itself. Two months later, we were automated.