Abcam is a pioneering producer of high-quality antibodies widely respected in the research community. The company has grown explosively thanks to an uncompromising attitude to producing reliable antibodies, commbined with a next-day delivery model.

After 20 years of trading online, Abcam's customers were experiencing difficulties locating the best products for their experiments due to the overwhelming number of filters, hard-to-find citations and a data model that didn’t accurately reflect the biological relationship between products.

I led a team of 3 designers, one researcher and a set of SMEs seconded from areas of the business. Our process was grounded in a culture of rigorous discovery and problem definition, ideation and validation.

Antibodies are Y shaped proteins that stick to antigens on proteins, cells, parasites, viruses, fungi and bacteria. Their utility has become ubiquitous in bioscience research. Antibodies can be used to diagnose infection, treat the infection, or measure the effectiveness of a medication or vaccine without experimenting on a human or animal. Bioscience researchers are usually interested in exploring the properties of a biological target - a collective word for whole proteins, antigens and binding areas.

There are tens of thousands of protein targets, with isoforms (different configurations of the same protein), other modifications. Each target has a name and a shorthand code. Not every scientist agrees on the name, so most proteins have a list of synonyms tied together with ID numbers by standards organisations like Uniprot.

Antibodies can be Primary (sticking directly to the antigen), secondary (sticking to the primary antibody to make them easier to observe). They can be conjugated (mainly dyes), polyclonal or monoclonal

Each target is linked by metabolic pathways - chains of events concerning many proteins and constituent parts that make biological functions - the processes that keep all living things alive. The same protein can be explored in different research fields, and scientists might use different experimental applications (experimental techniques) to test their hypotheses.

It should probably go without saying that this mind-boggling landscape of entities, relationships and properties does not easily fit into the average off-the-shelf e-commerce platform data model.

We interviewed 34 researchers from various levels in private and public organisations, and ran a diary study with scientists to discover contextual problems and unmet needs. We uncovered a lot of detail about the context of the decisions made during the antibody buying journey. we were able to expand on our original stories and put together a vision for the ideal antibody buying journey.

We described the buying journey in terms of the researcher's JTBD, rather than page-by-page, which enabled us to break out of the constraints of the existing flow. We identified four key steps of the journey: Search, Compare, Choose, Buy & Share. After organising our user needs into these four problem spaces, we undertook rigorous two-week design sprints with subject matter experts, which we repeated three times. As our confidence in the overall concept grew, we delved deeper into the details of each step, refining the design and making it more comprehensive.

Once synthesized, we re-wrote the insights as user stories, combined them with the business requirements that we could validate with the research, and plotted them on a story map. The map allowed us to scope a meaningful MVP.

• Researchers are driven people working on tight budgets. They must find significant results in the shortest possible time.
• The admin of finding reagents and supplies is squeezed into short slots around pre-booked lab time.
• The computer in a lab is often relegated to a corner. Many researchers ate lunch whilst browsing products.
• Most experiment planning was still carried out in notebooks.

We set out to reduce the time scientists spend planning, searching for and validating products so they could spend more time doing research. For each step of the process, we framed problems and focused problem statements around the obstacles to finding the most effective reagent for the experiment, in the shortest possible time.

When Abcam's website was in it's infancy, searching for a target would return a manageable list of antibodies. After 20 years of expanding their product lines, each target search now brings back hundreds of options from multiple categories.

• You could add two or more search terms in the search field to reduce options, but nothing in the UI suggested it was possible, so this behaviour was rarely observed.
• The search feature only matched on text in the product name or description, so could bring back irrelevant results - a cell line sample rather than the antibody that binds with it, for example.

The existing search results layout was well past a usable tipping point. With more filter categories and product lines added every week, filters were difficult to find in stacked vertically scrolling boxes.

• The product cards were tall and difficult to scan, making comparison difficult, and reducing the number of results visible at once.
• The left-hand column had become unmanageable for users. With difficult interactions, only 23% of users ever added a filter and opted to scroll through pages of results.
• We observed ping-pong navigation between the results view and the product page, and lots of scrolling to find relevant information.

Antibodies are supplied in small vials of clear liquid. You can’t differentiate them with an image of the product. Instead, researchers need to see citations and charts of experimental data that prove the specificity of the antibody (i.e. how accurately it binds to a specific target).

• To be trusted, the images must be accompanied by contextual information regarding the experiment in which they were produced.
• Published citations were regarded as essential to trust Abcam’s claims of reactivity and specificity. But researchers were spending a lot of time reading the detail of citations before realising they weren't relevant.

We discovered in the interviews that many researchers don’t have purchase power and had to request a purchase through a requisitions system or through their lead researcher.

• The existing e-commerce platform focused on buying. and although researchers could save a basket of items in a session cookie, they couldn’t share it, so researchers were resorting to other methods.
• Most non-purchasers shared a list of product pages by copying the URL. As a result our funnel metrics were missing a large number of journeys and our existing conversion rate was distorted.
• The shared URLs didn't contain size or delivery information, so researchers were accustomed to specifying these manually.

Validated Prototype

A user-validated prototype that demonstrated the potential of a biological data model to revolutionise the search experience.

Shift in organizational Thinking

Despite not being fully implemented, the project influenced Abcam’s appreciation for a data model that closely mirrors the biological domain.

Increased Awareness

The project highlighted the need for user-centred design in e-commerce platforms, especially in specialised fields such as scientific research.

---

Despite the robustness of our approach the team faced a significant change in direction. The technical team unilaterally decided on an off-the-shelf solution from Adobe that, while not fitting all the user requirements highlighted by the prototype, provided a perceived quicker route to deployment. This decision ended up as a false economy regarding the highly specialised features that customers were asking for. It pointed to the need for more clarity in defining strategic outcomes between the technical and product teams. The technical team expected a simple re-platforming based on current requirements, whilst the business expected a step change in the user experience to justify the investment.

This exceptionally specialist project was a pleasure to work on. Whilst I'm disappointed it hasn't yet been fully realised, I enjoyed engaging with bioscientists, who turned out to be excellent collaborators on the design. We solved problems well outside the usual domain of an e-commerce team, by involving them in every stage of design thinking. Thanks to this project, I feel my role has permanently evolved from problem solver to facilitator.

I struggled to engage the engineering organisation in the design process. Historically, the engineering team had worked in a silo. They were accustomed to scoping a full set of requirements and wanted to wait until all the design questions were answered before scoping. This disconnected the design from the build, and the implementation risk ballooned. In future I'll ask more questions about the organisational set up before embarking on projects of this size.

Despite the difficulties in the technical project, I was happy with the outcomes of the design process. By fostering an appreciation for the value of specialised data models and user-centric design, my team set the stage for future developments that could further solidify Abcam’s leadership in the market. Abcam’s new search design a is a testament to the enduring value of a design process that is deeply informed by the user's perspective, an approach that ultimately leads to a more intuitive and effective product, even if the path to implementation takes unexpected turns.

Want to work with me to create amazing products? Drop me a line