Redesigning the platform that was losing deals.
Airstride built a partner management suite, but the UX fell behind the product. Prospects were citing the interface as the reason they chose a competitor. I came in as the sole designer to redesign all six modules from scratch.
The product worked. The interface was losing the evaluation.
Airstride had a functioning partner management platform: tools for tracking deals, mapping accounts to a CRM, training partners, managing content, and measuring engagement. The underlying capability was real. What was failing was the surface.
Sales calls were surfacing a pattern. Prospects evaluated the product and then went with a rival, citing the UX as the tipping point. Six modules, each with their own flows and edge cases, all needed redesigning at once, with one designer.
The design problem was also an organisation problem.
The visual gaps were obvious. But two other problems were slowing down any solution: stakeholder decisions that kept reversing, and a build process that did not faithfully reproduce what was designed. Fixing the screens without addressing either of those would have meant designing the same things twice.
- P.01The deal-loser. Prospects were naming the interface directly as the reason they chose a competitor. UX had become a line item on the sales call.
- P.02The revision spiral. Stakeholder churn meant finished screens kept cycling back for variation requests. Design time was consumed by decisions that would not stay made.
- P.03The implementation gap. Designs were not being built faithfully. The Mantine UI library gave developers choices that diverged from intent, and there was no close review process to catch the drift.
One surface to answer the question every vendor asks on Monday morning.
Before a vendor opens a single module, they need to know whether the program is healthy. The Overview redesign started there: what does a partnership manager actually need to see the moment they log in? The answer was not a dashboard full of charts. It was a short list of decisions that needed making today, and the context to make them.
The getting-started flow is a first-class part of the design, not an afterthought. New vendors see a five-step quickstart with a progress ring, locked previews of the screens they are building toward, and a setup call option that surfaces a named partner success manager rather than a generic support link. Once the program is live, the overview becomes a configurable scorecard, an attention queue for approvals and stalled deals, and an AI briefing that surfaces the three most actionable things in the data that day. The customise panel lets each user choose which four metrics they want on their scorecard, without changing what anyone else sees.
Every deal’s status, scannable without opening it.
The deals module had to answer a practical question at a glance: which deals need attention right now? The previous design required opening each record to understand its state. The redesign introduced a clear set of deal states, default, management, and completed, so partnership managers could scan the list without drilling in.
A filter component with a warning indicator surfaces deals that need action without badge hunting. The primary and tertiary button hierarchy keeps the most common action prominent without cluttering the interface with equal-weight options. Partner detail panels open as overlays rather than routing to a separate page, so the list context stays intact while the user is working inside a record.
Name what you need before you ask for it.
Connecting a CRM is the moment a user hands over access to sensitive business data. The flow had to make the scope of that access legible before it happened. The copy that precedes the authorisation redirect names it plainly: we only access what is needed to map your accounts. That sentence sits above the button, not buried in a permissions modal.
The connection journey uses step badges to track progress through a multi-stage flow. Each state has a distinct visual treatment: Not connected, Connected but not yet selected, Selected, Pending, and Success. A verification badge confirms a successful link. Users who abandon mid-flow re-enter and see exactly where they left off, without losing progress.
Give everyone a different view of the same data.
Analytics dashboards for B2B platforms face a familiar tension: the metrics a vendor cares about are not the same ones a partner manager cares about. A single fixed layout satisfies nobody. The solution was a customise view overlay that lets users control which data surfaces they see and how they are arranged, without changing the underlying dataset.
The base dashboard ships with a sensible default. The customisation modal is accessible directly from the top bar and applies changes immediately. Features on the roadmap but not yet shipped are kept as hidden sections in the Figma file, holding the design work without surfacing it in the live product until the engineering work catches up.
The same module, three different jobs.
Training content in a partner platform serves three separate audiences at once. The Vendor builds and publishes modules. The Partner Admin deploys them to their team. The Partner User completes them. Each role needed its own view of the same underlying structure, without tripling the management overhead behind it.
Content creation offered two paths: upload existing material and have the system structure it into a module, or build from scratch manually. Both lead to the same module dashboard. Certification states, certified, not yet certified, and partner-certified, gave vendors a legible read on completion across their partner base without opening individual records. Empty states were designed for the onboarding moment, when a vendor logs in for the first time and the module list is blank.
Content access is a permissions problem first.
A content library in a partner platform is not just a file system. The wrong partner seeing draft materials, pricing documents, or unreleased product content is a real business risk. The design organised content into folders with per-folder access control: who can view it, who can edit it, set at creation and adjustable after.
Three role views, Vendor User, Partner Admin, and Partner User, gave each audience the right level of control without surfacing settings they could not act on. Three folder creation paths covered the main use cases: build manually, generate structure using AI, or mirror from an existing external source. The mirroring flow included a review step where vendors inspect what will be pulled in before confirming, reducing the risk of accidentally surfacing content from an unreviewed source.
Which partners are drifting, and which need a nudge.
Vendors managing dozens of partner relationships cannot check in manually. The engagement module answers the question a vendor asks at the start of the week: who has gone quiet, and who is active? Activity data, logins, content consumed, training completed, deal movement, rolls up into a surface that flags disengaged partners without requiring a deep dive into individual records.
The design prioritises signal over noise. Rather than surfacing every data point, the engagement view surfaces the partners who have crossed an activity threshold in either direction, so a vendor can act quickly on the ones that matter without wading through a full audit log.
What I would do with more time.
The thing I would change is where I started. I began in the interface, working from the existing product and conversations with stakeholders. What I did not do first was spend meaningful time inside the partner ecosystem industry itself: talking to partnership managers at other companies, understanding how they think about vendor relationships, where the real friction lives day-to-day.
That research would have sharpened the problem framing earlier, and likely changed some of the decisions made in the first few modules. The stakeholder revision problem would also have benefited from an explicit alignment session before the first design was shown, not after. Getting decision-makers to agree on criteria before seeing options is harder to arrange than it sounds, but it would have saved significant design cycles.
The outcome that mattered: paying customers who saw the new designs said they preferred it, and that they would recommend Airstride’s UX to other companies. The opportunity cost sits in the industry research that did not happen, the depth that might have made the six modules sharper from the start.