Why We Stopped Leading With the Product Clients Loved Most

So we just took the product our clients loved most and moved it to the back burner.

Crazy, right? Interactive video worked. People clicked through it live in front of their own clients, and we still support it. It's just not the first thing a new client sees anymore, and the reason has almost nothing to do with the product being wrong.

For about a year it was the core of what we built: video a presenter could click through and adapt live, branching and reordering as a conversation moved. Explainer video is the front door now. This is the honest version of why.

Quick note on names, because I use both a lot. Internally these are IVs and EVs. Not the drip, not the Tesla. And "explainer video" is just what people already call this stuff, the shelf a buyer reaches for, not a claim that every one is explaining anything.

Where Hyperframe actually came from

The first version was a Figma prototype, and honestly a bit of a hack. Illustrated animated scenes with labels and branching paths, built for an open brief from an industry-leading accounting firm that wanted a more engaging pitch experience. It landed, so we built more variants for other parts of their business, until Figma's limits started fighting us.

A run of good luck left us some spare profit, and we spent it building proper software. The client gave us the problem, we built the solution for ourselves.

Version one assembled and played back the scenes, internal only. Version two added share links and analytics. Version three, the current product, is AI-assisted.

Where interactive video genuinely worked

In the right hands it was excellent. One of the best examples I saw was a partner presenting on tax disputes at the firm's client experience centre, working the format live on a big touchscreen in front of a room of senior clients, clicking through as the discussion turned. It held the room in a way a static deck or a fixed video couldn't. It also read as visibly techy, which that firm valued more than anyone quite said out loud.

The economics worked once we optimised for reuse. Build one video for a client, then reuse its elements, in that client's own brand style, in the next. Across a single year we produced nearly 40 videos for that firm, more than one a fortnight, at a fraction of what rebuilding each from scratch through an agency would have cost.

So this was never a story about a product that didn't work.

The problem was the sale, not the usage

The trouble showed up before any of that value could land. Most people carry a clear mental model for two things in this space: a PowerPoint and a video. They know roughly when each is the right tool.

Interactive video sat in the gap. Often the better option in the exact moments our Big Four client reached for it, but with no existing shelf for a buyer to put it on. Explaining why it was worth the extra effort, why software instead of a video file, why a click-through instead of a straight watch, meant winning an argument about the format itself before you'd said a word about what it did for anyone.

That's a category problem, and it's a trap we walked straight into. The category a buyer files you under sets the terms they judge you on, so inventing a brand-new one means teaching the market from scratch before you can sell anything. We were doing exactly that, one meeting at a time. The people already using it loved it. Getting a new buyer to that point was the hard part.

The roadmap we're glad we shelved

We had a much bigger interactive feature set designed and close to built: deeper AI editing, and dynamic branching where the content itself changed based on choices a viewer made earlier. We launched smaller instead, and tested what we had against real buyers first.

In hindsight the relief is considerable. Those features would have eaten months and moved nothing, because the constraint was never capability. A buyer who can't place the format doesn't change their mind because it gained branching logic. If anything, the extra sophistication makes the explanation longer.

Why explainer video skips that step

Explainer video (more on the format here) doesn't ask anyone to accept a new category first. People already know where a video fits inside a client engagement, a pitch, an onboarding flow, a piece of internal comms. And a good chunk of that territory either can't get a video made today or is limping along with something worse.

So the job shrank. From "convince someone this format deserves to exist" to "show them a better way to do something they already do." A much smaller argument to win.

It's also the same ground where we make the risk case. Not a fully bespoke agency build that takes weeks and a heavy budget, and not an open prompt-to-video tool with no brand governance in the loop, but something in between that a marketing or brand team can actually trust with client-facing work. Once we put explainer video in front of people, they invented their own use cases, unprompted. You don't get to fake that signal.

What carried over, and what's genuinely new

Most of the engineering came straight across. The video pipeline work from interactive video taught us how to assemble a production process efficiently, and our approach to scene matching and script prompting gave us a template to adapt rather than a blank page.

What's new is more specific. We're putting real work into deeper brand control for explainer video. Not a logo on the end and a font swap, but tuning fairly granular parameters so an output matches how a brand team actually wants to sound and look. And we've kept building the pipeline that takes a client's own stock or DAM footage and filters, tags and prepares it, work that started for interactive video and now does most of the heavy lifting here.

Search at scale was the one thing we had to build from nothing

One piece of infrastructure genuinely had to be built almost from scratch: search at scale. Script generation runs on general-purpose language models with a lot of our own prompting on top, so an output reflects a brand's voice instead of reading like a stock prompt handed to an off-the-shelf model. But to pick the right scene for a given line, that model has to search a footage library across several dimensions at once: how many people are in shot, the setting, the tags a clip carries. It can't just skim a flat list.

Interactive video never forced this on us. It ran on a small set of hand-made animated clips, small enough that search was never the bottleneck. The libraries explainer video works against run to one, two, sometimes three thousand assets. Finding the right clip in a set that size needed a real search system, so we built an MCP server the script-writing model could query directly. That was the one place a bigger version of the old product wouldn't have done.

Interactive video isn't being retired

It's still part of the platform (here's how interactive video works), and the clients who get real value from it, mostly in live, presenter-led settings, keep using it. It's just not where new feature investment goes, and not what a new client meets first. A deliberate call, not a quiet sunset.

The lesson underneath all this took a year of real client use to see clearly, and it's fairly ordinary. A product doesn't have to be worse to be a harder sell. Sometimes it's just asking a buyer to make two decisions instead of one: accept an unfamiliar category, then accept your value inside it. The market was only ever going to make time for one.

If you'd rather watch explainer video do this than read me argue about it, that's what a demo is for. Book one here. No sales theatre, just the product in front of you.