Sarah Downs joined the Synaptica team as Director of Client Solutions earlier this year. For this Insights interview, we found out more about Sarah’s experiences, being a client of Synaptica and a hint of this new role. Sarah has two decades of corporate experience, including, most recently, leading a Taxonomy team at Facebook/Meta. In addition to her commanding project management skills, Sarah brings a deep understanding of how taxonomies are built and how to leverage them to enrich content and make it discoverable.
Tell us about you? Your experience your early life and interests?
SD: I’m an American living in London. My husband and I moved here in November 2016, and we have just been granted dual British citizenship. I grew up in Connecticut, the east coast of the US. I’m one of four kids in a very blended family. I moved to New York City for college and graduate school. After this I lived in California for nine years before the relocation to London.
My mother would tell you that I have been a taxonomist from very early on. On the first page of my first-grade journal, I wrote: “Things I like to do: 1. Making lists.” In kindergarten, my teacher reported that I would sit myself in a corner to sort small stones by shape and color. Clearly the innate urge to organize and classify was strong! Equally, I had an early drive to make things run efficiently. My mother also loves to recount this story: Me sitting in my highchair at 2 years old, watching her make toast. She moved around the kitchen from toaster to bread box, and back again, repeatedly crossing the kitchen. My response to watching this was “Mommy, it would really be much more efficient if you move the toaster over there (closer to the bread box)”.
How did you shift from healthcare to the world of taxonomy?
SD: The path was certainly winding and unexpected. After college, I began work in a specialty radiology practice while also pursuing a Masters of Public Health (MPH) in Healthy Policy & Management. After working in various roles across clinical research and direct patient care in an academic medical center, I moved to PwC’s healthcare advisory practice. For five years I worked as a management consultant with US and global clients. In 2013, a former PwC colleague contacted me about a job at Twitter. I had no background in tech, but when I looked at the specifications the skills matched those I used as a consultant. My role at Twitter was in strategy, operations and program management, and it was within this role I had my first introduction to taxonomy and data management. My arriving to a career in Taxonomy was really the result of luck and a supportive professional network. This is a theme that has continued across my Taxonomy career.
With my move to the UK in 2016, I had an opportunity to think about what I wanted to do next and what that would look like. I decided to train in Library & Information Studies (LIS) and began attending the LIS course at University College London (UCL). When I started that course, I thought one of three things could happen: I might become a librarian; or return to a tech company with more technical information skills; or maybe nothing would come of it – I would just have given myself a gift of studying something for a year.
I learned quickly a traditional librarian role wasn’t a good professional fit for me. Fortuitously, a former Twitter colleague encouraged me to explore the role of taxonomist, which was beginning to really expand rapidly across tech companies at the time. Through a referral from this colleague, I started working as a taxonomist at Facebook/Meta.
It was at Meta that I worked with Dave Clarke and the Synaptica team. After leaving Meta, Dave and I re-connected around the same time Graphite Knowledge Studio was launching. This led to this opportunity to form my current role at Synaptica.
Tell us about your new role with Synaptica.
SD: My role is Director of Client Solutions. Enterprise auto-classification and knowledge graph projects generally require some level of supporting professional services, and in my role of leading Client Solutions I will be helping clients turn their business requirements into practical working solutions.
I really like that the job title includes the world “Solutions.” It’s a role that goes beyond services, beyond a specific product. It anchors in understanding challenges and opportunities on the client side and thinking through how Synaptica can deliver a whole client-focused solution. This could be through our selection of products, or through a combination of products and services.
The role includes support, training, and all documentation that a client needs to succeed with the software. For example, Synaptica recently launched Graphite Knowledge Studio with support materials including documentation, videos and webinars. A sample of these are available from our website and YouTube channel.
It’s never enough when you’re adapting or implementing a new system to simply throw it over the fence and say, “here you go”. From my time at Meta, I know that’s not the Synaptica approach. I think there’s a true partnership and understanding between Synaptica and each of their clients. I’m thrilled to be a part of this going forward. Another element of my work will be public facing: talking at events, presenting webinars, creating videos, writing blog posts, and otherwise sharing use cases.
You have used several Synaptica tools? What do you look for in a taxonomy management tool?
SD: I believe that taxonomies are rooted in a specific need or use case – they need to be tuned to a specific business problem or opportunity. I think there’s sometimes an urge to push toward some platonic ideal of a pure taxonomy, but often that results in a structure that doesn’t truly address the business need. That’s not to say that a taxonomy should be hyper-focused on the needs of today – strong taxonomies solve not only the problem in front of you but anticipate problems in the future and are designed for extensibility and flexibility.
”I think taxonomists and taxonomies need to be flexible, adaptive responders to the problem they have been asked to solve. And consequently, taxonomy software needs to be the same: it must be flexible and adaptable to the needs of specific teams and products.
As a result, I think taxonomists and taxonomies need to be flexible, adaptive responders to the problem they have been asked to solve. And consequently, taxonomy software needs to be the same: it must be flexible and adaptable to the needs of specific teams and products. In addition, the team supporting the software needs to be willing to sit with their customers, listen and understand use cases, workflows, business needs, and data outputs – all the things that are required to ultimately solve a business problem.
As a Synaptica client, I enjoyed using the Synaptica tools and working with the Synaptica team to accomplish exactly this adaptation and customization. We spent a lot of time discussing the data, our needs, and what we needed the tool to do. We had an opportunity to work with the Synaptica team to build out custom reports or custom features that we needed for our team, which were then made available to other Synaptica users as well. I liked that experience – working at the interface of system solutions and business requirements.
What are you most looking forward to about joining the Synaptica team?
SD: I’m excited about the opportunity to be on the other side of the table, in conversations with clients listening, learning about their data, understanding their needs, helping to think about how to leverage enterprise tools for what they’re trying to achieve. That sort of diversity of experience and engagement in the taxonomy world is exciting to me.
With joining the team, I feel part of a group who are committed to their customers’ success. My experience with the Synaptica team is that they genuinely want to understand customers’ needs, data and use cases in order to provide smart, flexible solutions. The team engages empathetically and with curiosity and then think creatively about how to address the specific business requirements.
What makes a good taxonomist?
SD: I believe a good taxonomist is a good listener, a deep thinker, and an iterative worker. A good listener needs to understand the business needs the taxonomy is serving. This requires taking on different stakeholder positions and asking fundamental questions:
What business or customer problem are we trying to solve?
How will a taxonomy solution be used to address this problem?
What are the technical specs and how might that impact taxonomy design?
How will end users consume taxonomy data?
This involves active listening and asking thoughtful, probing questions. Then when you have all the data, you have to go away and really investigate deeply. I think this is something that taxonomists do very well: analyzing, interrogating, and distilling from complexity.
”Taxonomists need to not only define a strong solution but also need to be able to take their internal customers on a path toward understanding and owning the solution.
Taxonomists need to not only define a strong solution but also need to be able to take their internal customers on a path toward understanding and owning the solution. This is where iterative work comes in – sharing initial solutions, gathering feedback, incorporating that feedback in rounds of revision.
What general advice would you give to people developing taxonomies?
SD: I would definitely recommend any taxonomist have a consistent and ever-improving system for intake – A system that allows them to speak with internal customers to understand exactly what’s required, who will be using the ultimate information architecture, how and when. I’ve found the word “taxonomy” is often used as a catchall for any range of information architectures from a simple flat-list controlled vocabulary all the way through to a complex ontology. A standard intake questionnaire can help to quickly resolve any misunderstandings, reveal the ways in which you need to educate your stakeholders on what taxonomies can support as well as which alternatives to consider.
By standardizing and evolving the way in which you engage at the beginning of a project, you establish yourself as a true business partner for the challenge at hand. The intake questionnaire is only the beginning, though. You may need to revisit key questions with internally customers repeatedly. Continually anchoring in and refining scope and requirements is key to ultimately delivering a powerful information solution for the organization.
Why are taxonomies important to organisations?
SD: Taxonomies are now widely acknowledged as valuable in the world of tech and big data. These companies have learned the hard way what can go wrong without the light but powerful touch of human curation.
The union of human curation with algorithms has been shown repeatedly to yield better solutions than either method used in isolation. This combination allows organizations to surface and use complex data in actionable, productive, safe, and responsible ways. Taxonomy plays a powerful role is bringing together human curation with machine learning algorithms.
”Taxonomy plays a powerful role is bringing together human curation with machine learning algorithms.
Can you tell us about developing taxonomies for auto-categorization?
SD: The auto-categorization capabilities of Graphite Knowledge Studio unite the capabilities of information science with those of data science. It marries the power of human-curated, human readable, definitions of knowledge with computational, machine-readable inferred knowledge.
Successful auto-categorization, however, is not as simple as simply marrying your existing taxonomy concept schemes to text extraction algorithms. Even well-curated sets of preflabels and altlabels are unlikely to match the rich complexity of an organization’s content as parsed by a machine.
To use a simple example, a human can read “fruit” and understands intuitively that this also includes “apple”, “banana”, “strawberry” etc. Often building out this level of taxonomic detail is inefficient and not required for an organization’s purposes. But during autocategorization, you want the algorithm to recognize that documents about “apple”, “banana”, and “strawberry” are actually documents about “fruit.”
To increasingly improve the accuracy and power of an autocategorization, GKS enables iterative tuning of your taxonomy to your actual content – through the use of custom properties to define a set of evolving autocategorization rules. For example, expanding the definition of “fruit” to explicitly include “apple,” “banana,” and “strawberry.”
And going even deeper, if you find the algorithm has now tagged a family memoir blog post with “fruit” because of the phrase “the apple doesn’t fall far from the tree!”, then you can address that context as well with additional fine-tuning of the autocategorization rules.
This iterative work of running existing taxonomies against specific content, evaluating performance against human benchmarks, and refining rules is wonderful fun for a taxonomist brain.
What do you think are the biggest challenges for the sector (industry) in the future?
SD: We have all begun to collect and maintain huge amounts of data. Data can be powerful but can be used in opaque ways with the best of intentions. Intentionally or unintentionally, the data might be misused. We see increasingly privacy development emerging in many countries around the world. This is going to affect every company that holds troves of data.
The challenge of effective privacy and data use controls is a problem that is yet to be solved definitively. There is a huge opportunity through the use of taxonomy and ontology to move beyond reactive implementation of single regulations to privacy-proofing and safety-proofing your data from the beginning.
Synaptica Insights is our popular series of use cases sharing stories, news, and learning from our customers, partners, influencers, and colleagues. You can review the full list of Insight interviews online.