MGID

Background

Building campaigns in affiliate marketing is hard work. At MGID, that was costing advertisers time and causing tasks to fail. Our first tests turned up big usability problems. Users spent 16.28 minutes per task with a 65% success rate, and they rated the difficulty 3.4 / 7. We wanted more than a fresh look. The real aim was to make the flow make sense and help more people finish, with as few backend changes as possible.

A first round of testing set our baseline. It also showed where new users got stuck in the main flows.

My role

As Lead Product Designer, I led the redesign from start to finish. I worked with a product manager, a UX writer, and developers, and we kept the user first in every decision.

• Led the full redesign, from research and design through build and launch
• Ran research, workshops, user interviews, and journey maps
• Used data to find what users needed, then checked decisions with usability tests
• Designed wireframes, prototypes, and the final UI within the technical limits
• Set UI standards together with the team
• Helped plan quarterly strategy and the roadmap
• Built a design system to speed up onboarding and cut confusion

Research

First I looked at how users build campaigns, so I could see what to improve.

• Competitor audits. I asked experts and used SimilarWeb Pro to study the top 10 competitors. Then I compared how they organise things with MGID.
• Heatmaps. I studied Hotjar data and session recordings to check our assumptions and learn more.
• Workshops. I ran sessions with the product manager and experts (affiliates, media buyers, brands) to map the customer journey and find pain points.
• User interviews. I checked our ideas and got a deeper sense of how affiliates work.

Wireframes & prototyping

Once the direction was set, I built wireframes and updated the user flows. I showed the designs to users and the team for testing and feedback, then refined them from there.

Design system integration

I added a component library to speed up development. The old platform wasn't consistent. That slowed work, made teamwork harder, and caused usability problems. Research by Anja Klüver at Prospect shows design systems can make projects 30% faster and 30% cheaper.

• Made the product easier to learn, with screens you can predict
• Freed up time for research and harder problems
• Saved 50% of the time on common patterns
• Sped up development by 25%
• Sped up prototyping by 25%
• Made onboarding easier for new designers and developers

Campaign performance

The goal was to simplify the page without losing features or blocking new ones.

• Moved the Advertiser / Publisher switch into the user profile.
• Added a graph you can collapse, since many users prefer the table.
• Made the graph interactive, so you can compare your own metrics.
• Tidied up campaign actions. I hid the less-used ones on hover, and moved Statistics, Optimisation, and Ads to the campaign page.

Ad creation

Research showed that most users, and affiliates in particular, make several versions of each ad in one sitting. They mix text and images to find the pair that converts best. The old interface made you repeat the whole ad flow for every pair.

To fix this, we let users create many ads at once. The tool auto-generates a range of text and image pairs, which saves a lot of time.

Research also turned up a few common frustrations:

• Some features were confusing (Category Selection, Teaser Rich, Read Metadata).
• It was unclear how to set the image focal point and sizes.
• There was no clear preview for the different image formats.

To fix these, I made these changes:

1. Moved CPC into Limits & Schedule at the campaign level, while still letting you set it per ad.
2. Gave the focal point its own screen, with a preview of every format for the chosen teaser, so you can fine-tune where each image is centred.
3. Added time-savers like title suggestions and AI-generated images, and removed fields that weren't needed.

Campaign creation

Here the focus was simplifying the process to fix the main pain points. Fields were complex, errors were confusing, and there was no way to save a draft.

• Split creation into 3 clear steps with a progress bar.
• Added auto-save and inline checks to prevent lost work.
• Put errors on the right fields, with clearer wording.

Based on feedback that targeting was too complex:

• Simplified the layout so it's easier to scan and set up targeting.
• Replaced unclear icons with plain text labels for include / exclude.
• Removed fields that weren't needed, such as category, by collecting the data automatically.
• Moved traffic insights to a separate page, since they were rarely needed during creation.

Campaign optimisation

This part needed a lot of work with the product manager and backend developers, since the interface changes leaned heavily on the backend.

Optimising a campaign is ongoing manual work. You track stats, read data, adjust CPC, and turn off weak sources. MGID had built rule-based automation, but only 3% of advertisers used it. Interviews showed that the heavy tracking setup during campaign creation was putting people off.

We'd already rebuilt the tracking flow, so the goal here was to use design to get more people using the feature, without big backend changes:

• Made the automatic rules easier to find, with a better layout.
• Reordered columns based on what we heard in interviews.
• Moved goal settings into filters.

Results

Internal testing with 7 experts in mid-April went well. Feedback was mostly positive, and almost everyone finished the tasks on their own. Each session ran about 1.5-2 hours and combined a usability test with an interview, so we could check task completion and gather feedback afterwards.

Next steps

• Keep rolling out the design system to the rest of the product.
• Run a follow-up test with more people to confirm the gains hold over time.
• Get more people using automation, starting from the new optimisation flow.