Plate Repository | Ginkgo Internship
Developing a plate repository from 0 -> 1 to enable 500+ scientists to identify, learn about and select labware
ROLE
Product Design Intern
User Research, Usability Testing, Wireframing, Card Sorts, Visual Design, Prototyping
DURATION
Jun 2023 - Aug 2023
TEAM
Product Designers
Product Managers
Automation Engineers
Software Engineers
The TLDR;
CONTEXT
Being the first product design intern at Ginkgo Bioworks— a synthetic biology company.
I worked closely with the Automation Integration, Product and Engineering teams to address scientist's pain points surrounding labware selection and use.
PROBLEM
Scientists don't have an accurate source of truth for labware details, resulting in labware selection errors that damage experiments.
The result? Costly experiment replication, frustrated scientists, and unnecessary troubleshooting.
SOLUTION
I designed the first plate repository enterprise software at Ginkgo as well as a new naming schema. Both simplify the lab selection process by providing in-depth labware details that highlight key differences and similarities.
Visible filtering
I made all the tags visible upon entering the search to allow for scientists to easily narrow down their search.
Comparison
A comparison page helps scientists see exactly what the differences between plate types are, no matter how small.
Search results (Grid view)
Since some scientists said they prefer visuals, I created a grid view that would show the number of wells each plate has as well as its top geomtery.
Search results (List view)
For scientists who want a compact view, I created a list view that still showcases important plate details.
Plate details
Knowing that scientists constantly looked for manufacturer site information and images, I included them on this page. Additionally, for users who needed similar plates, I created a section to help reduce the time it'd take to find it.
Admin page to onboard plate
For scientists who want a compact view, I created a list view that still showcases important plate details.
Examples of Our Final Naming Schema!
SO WHAT?
Easier labware selection means less time and money wasted on replicating costly experiments! Our project was also honored as the first feature in a 3-part series about impactful 2023 Digital Tech intern projects.
Scientists and automation engineers across Ginkgo felt that the repository and new nomenclature was the best solution to the labware selection problem because it would mitigate assumptions and provide clarity! The impact this would deliver to the company led to our project being highlighted on Ginkgo's site from August 2023- November 2024!
Discovery
Uncovering the world of plates (no, not dinner plates)
Plates are used to hold samples and are arranged in grids that contain wells.
PLATE CHARACTERISTICS
Plate characteristics can affect things like microbial growth, liquid transfers, and other processes.
This means they can vary in a lot of different ways, such as having different:
Well bottom geometry
Well # / well density
Well top geometry
PLATE PROBLEMS
The current plate infrastructure at Ginkgo has been a cumbersome problem that's plagued scientists and the automation team for years.
In 2022, the company found 3 main problems with the lab infrastructure:
PROBLEMS
Inconsistent naming schema
Confusion about what plate types exist
Digital plate selection doesn't match to a physical plate in the lab
Ginkgo has a large physical and digital labware inventory to accommodate various types of experiments. Labware, such as plates, can have very miniscule differences that can drastically affect the results of an experiment.
When experiments go awry, they can take days to replicate. This means we need to minimize as much scientist error as possible! Addressing these pressing paint points can boost efficiency and maximize the output of successful experiments. But to make sure we're aligning as best we can with scientist's needs, let's talk to them!
I conducted 8 discovery interviews with the following user personas to understand the plate selection process across different roles.
I focused on centering the needs of our primary users (method developers and operators) as they would be using the plate repository the most:
Our secondary users (organism engineers and automation engineers) have important perspectives that are crucial to accommodate for. By also interviewing them, I can make sure our repository addresses as many use cases as possible.
Define
How did we make sense of the data?
I created an affinity map to understand key characteristics and common pain points.
KEY FINDINGS
I unraveled 3 main findings regarding scientist's proccesses.
FINDING #1
Some plate type names don’t relay important information to scientists.
"Every now and then… we find out that we thought a plate in LIMS was one plate and it turns out…it was not the right plate and there was a better plate type we could have been using.”
- Participant 2 | Operator
FINDING #2
People commonly seek out external data about plate types to figure out what they need.
“There's a G sheet or Excel sheet now that was kind of put together….But like I said… you have to find that sheet, find your plate, open the data sheet and then like look at it there versus it's just being in a table for you.”
- Participant 1 | Method developer
FINDING #3
Plate type characteristics are important to know since they can be influential on a method/automation.
“If we want real quantitative OD data, then we'd have to use either a transparent plate for the culturing…or we would have to transfer our cultures from a 3 84 echo PP to a clear bottom plate.”
- Participant 8 | Method Developer
After my initial discovery interviews, I created current customer journeys to map out scientist's processes from beginning to end.
To ensure accuracy, I conducted follow-up interviews with method developers, operators, and organism engineers to review my map.
I brought my findings to the team to deliberate on our defined problem statement!
DEFINED PROBLEM STATEMENT
The current plate selection process requires a lot of mental and physical labor from scientists because their decisions are not thoughtfully guided.
What does thoughtful guidance look like?
A plate repository!
Develop
Opportunities to address pain points
Exploring a new naming schema to ensure consistency and findability
WHY?
Our current naming schema is confusing scientists due to inconsistencies that lead to experimental errors.
This makes it hard for scientists to figure out what plate they're actually selecting. For example, a basic plate name like "96 Deep Well" could map to multiple different plate in the lab.
How do we make sure scientists are choosing the right plate and having that reflected in their workflows?
DECIDING ON OUR NEW SCHEMA
Since our affinity map told us which characteristics scientists focus on, we can conduct a closed card sort to determine which ones will be useful in a name.
I conducted a closed card sort with 8 scientists across different departments. The categories were:
Need to know in a name
Nice to know in a name
Don’t need to know in a name
Card Sort Findings
I did a spreadsheet matrix to see the most prioritized characteristics to the least prioritized characteristics.
A quick summary of our findings
A/B Testing With Different Schemas
Including just the "Need to Have" category would cause duplicates because there are hundreds of plates with the same manufacturer, well #, and brand. I decided to pull the traits from the "nice to have" category and test out which ones would be the most useful.
VS.
Our two naming schemas that we tested with!
Findings showed that both well depth / volume and part # were important for two different reasons:
REASONS FOR IMPORTANCE
Well depth/volume tells us if it'll fit our experiment
Part # helps us locate it in the lab
We settled on combining the two schemas because users felt that it was descriptive enough to both visualize and locate it in the lab.
Group feasibility mapping to prioritize user needs
ALIGNING WITH STAKEHOLDERS
I hosted a workshop with other designers, developers, and product managers to map opportunities for the plate repository UI.
This helped us align on product needs and consider the impact/effort of each feature.
Based on technical constraints and product goals, we decided to move forward with the following features:
Ability to filter beyond # of wells and type of automation
Comparison feature to find differences between similar plate types
Extensive plate type details page
Admin page for automation engineers to input new plate types and their informaton
Sketching key concepts
Taking what I gathered from our group feasibility session, I created the following sketches to get an idea of the functionality of our tool.
I later created mid-fidelity wireframes to generate feedback on user flows.
Test and Iterate
Gathering user feedback to improve happy paths
USER FLOW
As a method developer/operator/organism engineer, I want to filter my labware search to find the most relevant plate for my experiment.
Original Experience
Scientists would click on the filters button and open a side panel that listed all the filter capabilities.
USABILITY CONCERN
Filters should be more prominent because this gets us our desired plate faster.
It's important that scientists use the filtering system to navigate Ginkgo's large labware inventory. 13% of people used the filter for their search in the mid fidelity mocks.
New Experience
Scientists would open the repository and immediately see the filter capabilities available.
USABILITY TESTING RESULTS
100% of users used the new filtering experience to search for plate types.
In the new experience, users didn't feel overwhelmed and were satisfied because they felt like their search was more efficient and accurate.
A quote from our users regarding the new experience
"It's like specific where I know 'yeah, it'll definitely have this characteristic, this characteristic and that characteristic'. So I liked that."
USER FLOW
As a method developer/operator/organism engineer, I want to select and add a plate to my experiment.
Original Experience
Scientists would select the checkbox next to their desired plate and then confirm their selection in the top right corner.
USABILITY CONCERN
Calls to action were weak and confused users.
Some users thought clicking the check box meant the plate would be added to their workflow. Also, the placement of the 2nd button was not clear. 25% of scientists needed assistance for tasks.
Taking this feedback into account, I opted to make the confirmation process 1-step only.
New Experience
Scientists would only click "Use Plate" to confirm their selection.
USABILITY TESTING RESULTS
100% of users completed tasks without assistance
In the new experience, I placed call to actions next to the plate types so users would see the direct relationship and know what to do next.
A quote from our users regarding the new experience
“Just click use plate, right? I found that very easy.”
Deliver
Introducing the Plate Repository!
Finalized Site Map
I created the information architecture so that pages can be added later on for different labware types. The screens I prioritized were interactions done on the plates and admin pages.
Final Screens
Using the Ginkgo design system, I was able to create an interactive prototype that integrated the repository seamlessly into scientist's existing workflows and tools. Below is the final prototype!
Takeaways
Reflecting on the learnings of my first internship
Solidifed my love for product design
This was my first internship in Product Design and working alongside such passionate people was a fulfilling experience that I'll cherish forever. To connect on such a personal level with users and help make their processes easier-- it's something I can see myself doing for a long time.
Learned new research methods + design thinking skills
I was able to gain a solid understanding of how information architecture and aligning with user's mental models is the key to having a successful product.
Understood the power of balance and collaboration
With 4 user personas to account for, it was hard to ensure that each persona's needs were being accounted for. By circling back to the user, I was able to step into their shoes and design with empathy.
I'm always hungry for feedback and ways to improve, so I enjoyed working alongside a team where we could bounce ideas off each other, collaborate and learn together.
Rapid prototyping and iteration can foster better solutions
One thing I'd do differently would be to step away from being pixel perfect and instead prioritize user testing earlier on in the low fidelity stage. It would've saved time and allow me to come up with more ideas that I didn't initially anticipate having.
As a spoiler, if anyone else is ever looking for a great addition to their team and needs an original thinker and a great problem solver, Roshida brings all that and more. Having her with us this summer was a pleasure, and really moved the needle on the "plate selection" problem!
- Dave Philbin, Prev. Head of UX at Ginkgo Bioworks