Petabytes, and no front door

The spark

ioMoVo is a digital asset management platform built to pull files scattered across machines and cloud drives into one searchable place. The engineering is strong: it indexes petabytes and answers natural-language queries.

It had also been designed and built almost entirely by developers, and it had grown into a textbook case of feature creep. Video editors and design managers, people who move hundreds of assets a day, were opening it and getting stuck on the basics.

Two paths carry the whole product. In: connect external drives and upload assets to build one source of truth. Out: find a specific file fast, using keywords, AI metadata tags, or both. Everything we tested sat on one of those two paths.

The dig

Three of us ran independent heuristic evaluations against Nielsen’s ten usability heuristics first, then compiled only the issues that overlapped. That kept the baseline from being one person’s taste.

Then a strict screener recruited six creative professionals who manage large asset volumes daily. Each sat a roughly sixty-minute session: open, non-leading interview questions to get at their mental model, followed by think-aloud usability testing across five tasks on the live platform. We tracked task success, completion time, steps taken, error rate, and observable emotional response.

The finding that mattered came out of the very first task. ioMoVo’s whole value proposition is natural-language “Relevance” search, but the platform defaults to keyword search, and the toggle that switches modes sits outside the search bar entirely. Nobody found it.

The shift

We went in framing this as a decluttering job. The home screen had a rotating high-contrast video background wrecking legibility, redundant buttons, and company blog modules burying the primary action. All true, all fixable, all cosmetic.

The real problem broke people’s model of what the product is. Everyone assumed an uploaded file would be indexed automatically. ioMoVo required you to manually select and sync files after upload before they became searchable, so a fresh upload returned zero results, and the only cue distinguishing a synced file from an unsynced one was a small green tag versus a grey one, which failed contrast standards outright.

Users were not telling us the feature was hidden. They were telling us the product was broken. That reframed the work from “simplify this dashboard” to “make the system’s state visible”, which is a different brief with a different priority order.

The failure path four of six participants walked into unprompted.

The build

We deliberately did not invent new patterns. The redesign borrows from Google Drive and Dropbox, because the people using ioMoVo already use both, and a familiar layout spends none of their attention.

• Persistent primary action. A bold upload button in the left sidebar, available on every screen, not just Home.
• Four quick actions. The dashboard reduced to connect a drive, create a folder, record video, capture screen. Everything else moved out of the first screen.
• Search that explains itself. Placeholder text reading “Search through KEYWORD or RELEVANCE” with the toggle built into the input, so the mode is visible at the moment of use.
• Feedback that lands. A floating bottom toast confirms a successful upload and offers a “Locate file” shortcut, replacing a notification-bell ring nobody noticed.
• One share modal. The confusing double-link model became a single copyable link with explicit Viewer, Commenter, or Editor permissions.

Constraints

Six participants surface roughly 85% of core usability issues, which is enough to prioritise and not enough to quantify. Nothing here is a rate.

The engagement ended at a prioritised audit. We had no route to ship the redesign, so the sync fix, the share modal, and the search toggle were never put back in front of the six people who failed on them. That second pass is the part that would have told us whether the recommendations hold.

Two of the findings carry security and privacy consequences rather than usability ones: clicking “Delete CDN” left the shared file publicly accessible, and the system accepted invalid email addresses without warning. Both were reported and neither was ours to fix.

The proof

This was a consulting engagement, not a shipped release, so the outcome is a prioritised audit rather than a post-launch metric. What makes it usable is that every recommendation is tied to a counted failure.

Five of six left the ioCloud screen entirely and navigated back to Home just to upload, because ioCloud had no visible upload button.

A powerful back end means nothing if users can’t find the front door.

Two problems appeared only on the live platform and in none of the three heuristic evaluations: the missing ioCloud upload button, and a format filter that could not be individually deselected, where the prominent blue “Clear Filter” button was routinely tapped in place of “Apply”.

The lesson

I would run one session on the live platform before writing the heuristic audit, not after. The two worst discoveries were invisible to three experienced evaluators and obvious within ninety seconds of watching somebody try. The heuristic pass is a hypothesis. I had been treating it as a finding.

The audit as delivered runs to a PDF of the full document.