Reinventing Search At Abcam
Executive summary
Abcam sells reagents to scientists running experiments that can take months and cost five figures. As the catalogue grew, search became the bottleneck — 60% of users dropped off before viewing a product, and only 4% engaged with filters.
I led the foundational research that reframed the problem: this wasn't a search UX issue, it was a workflow mismatch. Scientists search the way they design experiments — target, application, species, in that order — but the platform asked them to search like shoppers.
Forty digital ethnographic interviews and a global diary study produced a single insight: make the platform mirror the experiment, not the catalogue.
The redesigned search ships a guided three-step funnel built around scientific intent.
Outcome (measured 90 days post-launch):
+2.5% revenue vs. YoY forecast - approximately [$7.9M annualized]
+10% NPS among repeat researchers
−2% support call volume related to product discovery
−15% search result list size with no drop in conversion
The work became the methodological template for three subsequent foundational studies at Abcam.
context & objectives
context
I had noticed in some previous research that with the catalogue increasing very fast, users seem less likely to find the right products for their experiments.
From a business perspective, this is likely to drive conversion as well as the user satisfaction and the customer live time value down.
Problem
Problem
Catalogue expanding rapidly
Lower Filter Engagement (<4% of visitors)
Drop Off at SRP (Search Result Page)
Impact
Harder product discovery
Inefficient search behavior
Loss in conversion
How might we redesign the search functionality so that all users find the product they need for their experiment?
Objective
Research Approach
Pre-existing Data
Data analysis suggested that search was likely to be a problem:
GA data suggested that sessions were dropping off at the SRP level
Low interaction with the filters (less than 4% of users) GA + Hotjar heatmaps
A UX survey realized in early 2020 indicated that overall search was the main concern from our users (chart on the right)
“Abcam’s platform receives millions of searches from scientists globally — but 60% of users dropped off before viewing a product. Finding the right reagent wasn’t just frustrating — it could cost researchers time, grant money, and experiment success.”
WHAT METHODOLOGY
The methodology chosen is always defined by the type of insights that needs to be uncovered or discovered.
Digital Ethnographic Study:
Allowed to gather both, attitudinal and behavioural insights.
40 interviews split across all the digital personas - users and non users
Diary Study:
Allowed to gather some contextual insights.
12 participants 2 of each main user personas - worldwide
Methodology
What type of insights
In order to define which type of insight I needed for this research I decided to conduct some hypothesis workshop. The objective was to define what kind of insights we needed to gather in order to answer our design problem.
2 major hypothesis emerged from these sessions:
Users might feel overwhelmed with the number of search results
Users must have a search pattern since the first search terms are only what we scientifically call “targets”
Which led to 3 type of insights:
Attitudinal:
We wanted to understand our users' frustrations and concerns regarding their search journeys
We wanted to get an understanding of what the conscious issues were
Behavioral:
We wanted to observe users pain points during their overall search journeys
Identify any unconscious or nonverbal issues our users were facing (focusing on the findable discoverable understandable)
Contextual:
The objective was to uncover the pain points faced by users offline
We needed to understand the context of their usage
What we learned
findings & Insights
INSIGHT 1:
Scientists search by experiment, not by product. Every researcher in our sample followed the same three-step mental model: target → application → species. The platform forced them to invert that. Every time we surfaced experiment-context filters first, latency to "right product" dropped by [4.5 seconds] in usability sessions.
Implication for the business: the SRP wasn't broken. The entry to it was.
INSIGHT 2:
Scientists plan months ahead, then need products in hours. The job-to-be-done is bimodal. Long planning cycles followed by emergency reordering between failed experiments. A search experience optimized for browsing fails the emergency case.
Implication for the business: a saved-experiment shortlist is worth more than a recommendation engine for this audience.
INSIGHT 3:
Trust signals differ by seniority. Senior researchers trust peer citations. Junior researchers trust visual confirmation. Both distrust marketing language. We dropped one and replaced it with structured peer evidence (citations, validation data, lab-of-origin).
Implication for the business: the product page is a trust transaction, not a marketing surface.
What changed
The search architecture. A three-step guided funnel replaced the open-text default. Scientists narrow by intent before they see results.
The product taxonomy. The team rebuilt filter logic around the target → application → species pattern. The change required taxonomy work across SKUs.
The roadmap. Two planned features were killed (an AI-style "smart search" experiment, and a marketing-driven recommendation widget) because the research showed they'd add friction in the bimodal use case.
A new search functionality:
A 3 steps search to narrow down the results from the beginning.
Users have the opportunity to narrow down their search from the start by choosing options among 3 of the main research criteria identified during the exploratory research.
The filters displayed will vary depending on the product type the users are searching for.
It is an adapted intuitive search.
Modifying the user journey:
Introducing an additional step in the process, a (purposeful, useful friction)that allows users to select a specific type, enhancing precision compared to the previous experience.
Catering for overwhelming search results:
Results are refined to a manageable selection, preventing users from feeling overwhelmed and enabling them to quickly identify the right product for their experiment.
three step funnel
Conclusion
This project transformed Abcam’s search experience from an information-dump into a guided, researcher-centric journey. The redesigned model aligns with real scientific workflows, balancing precision and efficiency. Next, we plan usability testing on the prototype and iterative refinements based on real lab contexts.