Nicole Buckenwolf: Making Sense of Complex Product Graphs with Taxonomies and Ontologies – Episode 92

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Google Podcasts | Stitcher | TuneIn | RSS

Nicole Buckenwolf organizes the product information that you see at businesses like Amazon and Etsy.

Big retailers use product graphs, ontologies, and taxonomies to classify and categorize the millions of products they carry.

Building these massive catalogs involves a variety of users and collaborators. Both the vendors who provide the products and the shoppers who buy them – as well as the in-house teams that build these complex shopping systems – benefit from the valuable metadata that Nicole and her colleagues create.

We talked about:

• her path from medieval studies to her current role as a product ontologist at Amazon
• the valuable training in taxonomy that she got in her first job
• her on-the-spot take on the classic “is a hot dog a sandwich” taxonomy question
• how data helps her taxonomy and ontology work, especially around the lack of standards in creating product ontologies
• the surprising commonalities between the very standardized topics she worked with in the industrial supply world and the wide-open world of custom goods at Etsy
• how much she enjoys working at the “hinge point” of the metadata landscape
• the challenges of categorizing goods at Etsy, where vendors and buyers often disagree about how to classify products
• how the nature of taxonomy and ontology work can affect work culture
• how structured data practices can help with taxonomy work
• the need for fluid taxonomies, since the body of things being classified changes so often
• how Etsy balances the need to address sellers needs, buyer needs, and the needs of the team building the system
• how UX researchers at Etsy help with her work
• some advice for aspiring taxonomists and ontologists

Nicole’s bio

• Follow Nicole on Twitter

Video

Here’s the video version of our conversation:

Podcast intro transcript

This is the Content Strategy Insights podcast, episode number 92. Nicole Buckenwolf helps digital teams classify and organize product information. She currently works on product graphs at Amazon, but we talk in this episode mostly about her earlier work at Etsy. In both jobs, she has dealt with complex product lines and a variety of users and collaborators. In situations like these, it’s crucial to give all of the folks involved a clear picture of the metadata landscape that they’re navigating.

Interview transcript

Larry:
Hi, everyone. Welcome to episode number 92 of the Content Strategy Insights podcast. I’m really happy today to have with us Nicole Buckenwolf. Nicole is currently an ontologist at Amazon working on the product knowledge graph there. But I met her about a year ago at World Information Architecture Day, and she was at that time working with Etsy. Anyhow, welcome to the show, Nicole. Tell the folks a little bit more about your work, how you ended up at Etsy and then Amazon and all the interesting stuff you did before that.

Nicole::
Sure thing. Thank you. So right now, I, as you said, work at Amazon. I’m on the product graph team. And the main customer I work with is the Amazon Music team. So I’m building data models for music. Yeah. My bypath here is fairly circuitous as I think a lot of taxonomists and ontologists are. I majored in medieval studies. I was fairly a typical undergrad that I had no idea what I wanted to study. So I played around with some things. Ended up in medieval studies. Graduated with a history degree. Ended up going back to school a few years later for a graduate degree in library science. I wanted to focus on rare books and be able to use my medieval history degree. So I did that, kind of alongside of that. And after I graduated from library school, I was working with a lot of nonprofits just as a personal interest of mine. Interested in museums and art galleries and things like that.

Nicole::
And I found that the work I always ended up doing with those institutions was sort of data management, data organization work. Some of that involved donor databases. Some of it involved survey research, taking survey responses and sort of categorizing and coding the responses to derive meaning from them. And I got really, really interested in that work and it was something I hadn’t really considered as a manifestation of my library science degree when I was there. So when I was looking for jobs a few years later, when I was looking to leave the small town I was living in and going to Chicago, I was doing a job search. And actually, I was very frustrated.

Nicole::
And so I did like a joke search for logic puzzles in indeed.com. And this job came up and it was a taxonomist job. And it was like, do you like logic puzzles? You’ll love taxonomy. And they were offering to train people that had a background in library science. So I interviewed and I got the job and I moved to Chicago and I loved it pretty much immediately. It probably helped that they had a great trainer there, a great manager that was training taxonomists, but I absolutely fell in love with it. And so that was the point when I was like, this is really going to be my career now. And yeah, it’s history from there. I worked there for a while at the consulting firm for a couple of years. From there, I went to Amazon. I then left and went to Etsy for a year. And now I’m back at Amazon again.

Larry:
Nice. To harken back to that year, I don’t expect the whole curriculum, but I’d love it if you could just give our… Because a lot of our listeners, I’ve had a couple of other taxonomists on, and everybody knows the notion of organizing and categorizing things. But could you give us your take on taxonomy and maybe some of the key lessons that you learned in that training that you got in your first job?

Nicole::
Yeah. Sure. Obviously, I think one of the things that for me, was really interesting about it and really critical was that categorization, it’s really subjective. I’d always thought of it as being like an objective thing that makes logical sense. And it does to some extent. But you do need to build a model for the use case that you’re serving. And having come from the library science world, there’s a lot of bibliographic standards that they use and reuse again and again.

Nicole::
And that makes sense for that space, but it had never occurred to me that there could be this kind of frontier of classification in the world of e-commerce where you’re not using standards for the most part, you’re kind of making it up for whatever client you’re working with. And so yeah. I guess that was one of the big takeaways for me is it was interesting mix of building structure, but sort of being forced to rethink it every time for a different use case, which is really fascinating to me. I don’t know if I answered your question.

Larry:
That’s interesting to me. And I’ve been geeking out on taxonomy and ontology and all this stuff for the last year or so I think. And I’ve been practicing, I’ve been doing taxonomy work, not like the level you do it, but as an information architect, doing it as part of that practice. But I think one of the things you just talked about there reminded me, and taxonomists and ontologists constantly, there’s a lot of, not pedantry, but really interesting to what you were just saying about the imprecision of it. Like the notion of, is it a hotdog or a sandwich? This just comes up all the time in the taxonomy world. Actually, let me get your take on that. And have you walked through with your taxonomist hat on, is it a hotdog or a sandwich?

Nicole::
Man, that’s a tricky question. God, honestly, I haven’t thought too much about it. I would say yes. And again, this is less thinking about its physical characteristics and more about the fact of the context of when people would eat a sandwich or when they would eat a hotdog. I feel like there’s a similar space which those things are consumed. But I’m sure somebody is going to say, “Oh, but a hotdog is in a bun and the bun is connected.” And yes, that’s true. But yeah. For me, and I think this is a big part because I’m so informed by my work at Amazon where everything is very customer-focused, but it’s all about how is the end user going to use this? Because that’s the tie breaker. So I’ll say it’s a sandwich.

Larry:
Yeah. Okay. And I’m with you on that, because when you go back to user intent, which is a big thing in our world and satisfying that user’s needs, it was the Earl of Sandwich had it invented so he could hold it in one hand and eat while he still wrote and worked and all that other stuff. And that gets the meaning around the sandwich and the meaning associated with how you’re classifying and organizing that thing. I think when people think of taxonomy, they think of the Linnean biological taxonomy of kingdoms and phyla and all that, the ways of organizing all the biological entities in the world. But you were just saying that you came from academia where everything is pretty tucked in and it’s all been worked out. There’s a place for everything, everything in its place to this wild new world of e-commerce where it’s like you’re kind of making stuff up. And it’s probably not a black or white thing. It’s probably a continuum. How do you navigate that continuum from precision to the Wild West and how to make sense of the Wild West?

Nicole::
Yeah. That’s a good question. For me, a lot of it comes down to the data, I guess, you have available first of all. You you can only do so much if you don’t have very detailed data. But yeah. Also again, I think it depends on your use case. The first company I was working for, they did ontologies for industrial supply companies, which now these are companies that they’re selling materials, build airplanes and stuff. You don’t want to get that wrong. You don’t want someone that wants aerospace glue and they get the wrong glue for the… So in that case, precision was fairly important and it meant a lot of research into military standards for building materials and things like that. I got to use all those college research skills.

Nicole::
So in that case, there wasn’t really a classification standard, but there were these industry standards. And it was a matter of researching those and finding a way to organize them in a structure that someone could navigate on a website. But on the other hand, that level of precision isn’t always needed. And sometimes you want flexibility. And I think this is where my current job, working for music is a good example because music’s not really a hierarchical thing. Sometimes you’ll have a navigation on a site which is by genre and you might have rock and then under that is alternative rock, but on the whole music metadata, isn’t really hierarchical. It’s a lot of relations between members of bands and songwriters and instruments and it really lends itself well to a web. And I think that’s a case where it’s less important to be precise, and more important to think about how is someone going to use this? What information do I want to get out of it? I don’t know. I guess precision is a factor in both places, but it kind of factors in in a different way or with a different need.

Larry:
Yeah. And that’s interesting because that’s another distinction I’ve heard taxonomists talk about. Is the difference between classification and taxonomy. Taxonomy there, I think, and again I don’t know if there’s agreement across the board on this, but generally taxonomy, it’s like there’s one place to put everything. Whereas in a classification scheme, like in music, you could say, well, this is an example of both R&B and some other genre at the same time. It can fit in both places. And that would be more of a classification thing. I think food is often brought up like Thai food versus…

Larry:
A taxonomy would have classified the ingredients and the cooking regions or something, whereas recipe classification might be all over the map. But I want to harken back to what you were saying about the industrial supply work versus your work at Etsy. Which strikes me as being the exact opposite from very precise, orderly industry standards around how you classify and organize things to this wide open marketplace of handmade stuff that people are just making up. Tell me about the difference in your practice at those two places, the industrial supply versus Etsy?

Nicole::
Yeah. Actually it’s funny, I feel like there are actually a lot of similarities and not only because some of the people I’ve worked with at Etsy had also worked for industrial supply taxonomies. But it’s a small world . . .really. Yeah. I think one of the interesting things that they shared actually is that most of the people that I worked with anyway, I don’t know if I can say this about the industry as a whole, but most of the people I worked with in the industrial supply space, they were not very trustful of the internet.

Nicole::
This was earlier in the 2000s. And these were a lot of people at big old companies that had paper catalogs for years. And they were just getting their catalogs online. And they had a lot of distrust for what we were doing and these people coming in and trying to organize our products. And I feel like there’s a similar sentiment among Etsy sellers, where there’s this feeling of like, well, you don’t understand what I do. Can I trust you with my information? Which is interesting, because they’re such different worlds.

Nicole::
Yeah. In terms of the differences, like I mentioned earlier, with industrial supply, there is a lot of structured metadata available, because they have to provide that stuff. Because like I said, there are these very literally life or death use cases that we need to know what standards this product meets and what’s the strength of this metal. Whereas with Etsy, sellers know a lot about their products, but how they want to present it, the use case is just totally different. They’re looking to sell something for aesthetic value. And so they’re going to maybe highlight different things. So I think that’s yeah.

Larry:
And it reminds me of, you’re in one of roles in the middle of this whole thing between the underlying information architecture structure that drives these whole, whether it’s a commerce thing or a catalog thing or whatever it is up to the end user and matching up their needs. It’s occurring to me as you talk about that dynamic. In Etsy, you also have two users. You have the end user, the consumer that’s actually buying these handmade things from them.

Larry:
And then you have this vendor that’s supplying it. There’s just a lot going on there. And tell me how you interact with the front end design. There’s so many people you could be interacting with it. It occurs to me that the people who are managing the relationship with the vendors, as well as the actual relationships with vendors themselves, the end user customers. How you present things to them. How they talk about stuff and then tidying it up for whatever system you’re operating in. Seems like you’re really in the middle of a lot of that stuff. How do you navigate that?

Nicole::
Yeah. That’s interesting. That’s actually one of the things I like best about this job is you are sort of like in this hinge point place, I feel like, in the metadata landscape yeah. Etsy, it was particularly interesting because, like I said, sellers, they are very attached to their products. They have very strong feelings about them and how they want them to be presented. And in most cases, the sellers have complete control over that. Because they’re usually selling on their own at craft fairs or whatever. So Etsy for them is a bit of an anomaly. Yeah. I guess at Etsy, we had to meet the needs of putting categories like in the left-hand nav and in the top bar at website so that people could navigate.

Nicole::
And then that same taxonomy needed to feed into the backend taxonomy that sellers were classified into. And most of the time that was more or less one-to-one. I think both sellers and customers expect to find things like dishes under the kitchen section. It’s not a contentious categorization. Whereas we had something, and I think I may have mentioned this at World IA day, there are a number of fiber art categories. And some sellers like to put their wall hangings into fiber art and others like to put those very similar hangings into home decor wall art. And so one is sort of more like a wayfair.com type category rights with a home decor lens you’re looking at it through. And the other is more of like an art fair fine art lens, but they look exactly the same product. They’re both fiber wall hangings.

Nicole::
And a lot of artists would like to be perceived as fine artists. So they really want their thing to be in the fine art category. Whereas customers by and large are mostly looking for something to hang on their wall. They’re not necessarily thinking about it as fine art. They want it to be art, that’s why they’re on Etsy. But they’re thinking about it as, I need something for my wall. I’m going to go to Etsy and look for it. So that was one of those interesting cases where it’s like, how do we make both of these people happy? How do we get it so that their things are in the right place? And some customers might truly want to look for fine art wall hangings, but the majority, I think do not.

Larry:
And that’s interesting, because I’ve resolved that as an information architect, figuring out that, what you call it versus what they’re looking for. It was kind of a big part of that. And one way you handle that is by labeling things in the interface the user way and labeling things on the backend, the supplier way or however you call that. But that can get messy in terms of you need some kind of concordance chart or something to tie it all together. So what’s the practice in taxonomy to resolve that? Is there an established professional, an old standard for how you stitch together things that have different names for the same thing?

Nicole::
Yeah. I don’t know if there’s a standard. Personally, I don’t. There may be. But I don’t know if there is. I think the most common thing I’ve seen is the use of polyhierarchy, where you maybe have the same node listed under two different parents and maybe you name it something else, but all of the items assigned to that node are coming from the same place on the backend. But yeah. Not every system can handle polyhierarchy. And so that’s, I think, when you enter this place where maybe you need to make a decision about how you present it. And it’s been really more of like a technical limitation that you have to figure out how to work around.

Larry:
Yeah. And that gets back to that you are at that hinge point between the technicians and the technologists and the end users and the customers and all that. Actually, I wanted to ask specifically, and maybe the question I was just about to ask is in this, about your time at Etsy, when you were talking about the Wild West earlier, that seems as Wild West-ey as it can be. In terms of that, were there differences, you mentioned that there was even some overlap between maybe the people who worked at the first place you worked in at Etsy. But how did the staffing and how people worked together in that sort of like the industrial standardized world versus the Etsy world, did that reflect in the culture of the folks who you were working with?

Nicole::
Oh yeah. Definitely. Absolutely. I don’t know if it can be totally attributed to the culture necessarily, but there was certainly a very different atmosphere between the two places. The staff at Etsy is generally much younger as a whole. It’s longer a B Corp, but it was, and they still have a history and a vested interest in being very green and very liberal in advocating for sellers and things like that. Advocating for artists. Yeah. Industrial supply industry is not… Yeah. I don’t know. It’s just different thing. It’s not a bunch of young artists. It was just a totally different set of people. And so Etsy tends attract the kind of person that would shop on Etsy, or that would sell on Etsy, I think by and large on the staff. So yeah. It was definitely a much, much, much different culture.

Larry:
Yep. And I was just curious, because there’s so much of what we do, and I’m projecting the world of content strategy onto your work. We’re all like ostensibly word people and maybe a little bit of technology people, but we’re really like psychiatrists and therapists trying to get there, and so very curious about culture and that kind of thing. Another thing that you said in there, I can’t remember what you said a couple minutes ago, but it got me thinking about the connection between categorizing and classifying things.

Larry:
It seems like some of those obstacles you come up against like the need for polyhierarchies and things like that, is like going from just what a thing is to what it means. Or sort of assigning other things to it, besides just its physical characteristics. Like assigning some meaning to it. I know your job title now is ontologist and there’s a lot of different ways how that’s practiced, but I’m curious about, how meaning manifests in your work and how you, and it doesn’t even have to be an ontology thing, it can just be within taxonomy. You have to deal with that somehow. Can you talk a little bit about that?

Nicole::
Yeah. That’s a good question. I’m trying to think if I can come up with an example quickly.

Larry:
We were talking before we went on the air and you mentioned the use of the word silver at Etsy.

Nicole::
Yeah. Yeah. Yeah.

Larry:
That whether it’s a color or is that a mineral? That’s actually what I was thinking of when I mentioned that. Does that make sense? Yeah.

Nicole::
Yeah, yeah, yeah. Etsy does have some structured data and so sellers, they will classify items either into silver, the material or silver, the color. But there are a lot of cases where they’re not putting that information into the structured data. They’re just putting it into the free text description. And so that’s where there’s a lot of ambiguity because a person reading the description might be able to tell if it’s silver, the color or silver, the metal. But it can be hard to understand the meaning of silver without someone going through and reading every single description to get the full context. So yeah. There was a lot of that where there were just buckets and buckets of free texts, that it was very hard to pull out the meaning from.

Larry:
Yeah. You just do your best, I guess, to organize things and to give… And I think I’ve talked to people who see taxonomies as kind of a fixed thing, but they’re not, right? This must be growing and evolving and these kinds of questions that I’m asking you will come up and you’ll be like, “Oh, how can I fix that with taxonomy?” How fixed do taxonomies stay or how dynamic are they as a business grows and changes?

Nicole::
Yeah. That’s a great question. I think a lot of it depends on just what relies on the taxonomy in a lot of cases. If a lot of systems rely on the taxonomy not changing, then even if it needs to change, it might not. But I think as a rule, taxonomy should be fairly fluid, because I think they’re based on whatever they’re classifying. And if the corpus of things that are being classified changes, then maybe that data model doesn’t suit it in anymore.

Nicole::
This is an extreme example that’s not too common, but sometimes a new product comes onto the market, like when those single wheel motorized scooters became popular a few years ago, it’s like, there was no category for those. So it’s like all of a sudden you need to create a category for something that’s more drastic than what usually happens, but those changes are always happening. Things are always being invented or new materials or whatever. So you need to be able to go with that.

Larry:
It seems to me I’m just thinking off the top of my head here, it almost seems like a new thing would be easier because you just find its place for it. But it seems like change might be harder than the introduction of a new thing. And I can’t think of an example, but I know I can feel the concept of people’s perception of a thing changing or something to where it might end up being reclassified.

Nicole::
Yeah. I totally agree.

Larry:
Yeah. Something else we talked about before we went on the air and this hearkens back to what I was talking about. And this is the title of my podcast is Content Strategy Insights. And one of the insights that we all have had as content strategists is that we’re really more therapists than we are word people or editors or whatever we’re doing, and dealing with people’s feelings a lot. I was asking you earlier about work culture, but there’s also the interaction culture. And maybe even if you’re not directly interacting with the vendors and how they feel about your categorization of something or things like that. How do you communicate that up and down however you’re communicating with them to get to a place where everybody’s happy, where their feelings are have been addressed, but you still have things organized the way you need?

Nicole::
Yeah. That’s an interesting question. I think Etsy, they have a lot more avenues for that than most other places. In that they have these seller forums on their website that sellers post in. And they have ways sellers to recommend categories and things like that. So there is this kind of flow of information from sellers to Etsy. And then in turn, if they make a lot of major taxonomy changes on the site, information about that will go out to sellers along with any guidance for how to use it. If they’ve created a certain attribute that you can only put certain values, they’ll communicate that information. But yeah. Again, some of the things at Etsy might be able to talk more about sort of the dynamic, but in my experience, everyone is never satisfied. So you’re always going to have sellers that don’t like the changes that you made.

Nicole::
Yeah. I don’t know. It’s a tough one, because you want to build trust with sellers, but you’re also in this position where you’re also seeing what people are buying and you’re seeing search queries and you know how customers are thinking about things. And so you’re coming to it with that context that sellers might not have. So I don’t know. I think the thing about Etsy is they’re genuinely trying to do right by the artist. And they want to sell artists’ stuff. They want to help artists make money. So I think there’s generally best intention there, but the sellers may or may not agree with the decisions made.

Larry:
Something you just said there that it reminds me of something I probably should have asked about earlier about the process of the research that informs your work. Like learning about those feelings that the vendors have, you said, there’s forums that Etsy specifically has, but there must be other ways. Are there UX researchers or other folks or product people who are talking to the customers and feeding that information to you or?

Nicole::
Yeah. Absolutely. Etsy has a really robust UX program in general. So yeah. You can probably find lots of talks on the internet that they’ve recorded. Yeah. They do a lot of user research and stuff like that. Yeah. Again, combing seller forums, there are a lot of… I used to be involved in the world of craft fairs actually. So I’m in a lot of Facebook groups for people that do craft fairs and things, who see people posting there. There’s a lot of different ways you can sort of gather anecdotal information from sellers.

Larry:
Yeah. I think we all have our listening posts that we routinely visit nowadays. Yeah. Wait, Nicole, we’re already coming up close to time. I’m always amazed at how quickly these conversations go, but I wanted to give you a chance if there’s anything left, anything that’s occurred to you during the conversation or that’s just on your mind about taxonomy or an ontology practice that you’d like to share with the folks before we wrap up.

Nicole::
I think I would just say, and of course I’m biased coming from the background I am, but the profession is becoming more and more codified, in the sense that you can study ontology in school now. And I think more and more new grads are coming out with this idea fresh that they want to be a taxonomist or an ontologist. And I guess I would just say, that’s great and that’s awesome. And I’m glad that exists. But at the same time, I would just say to other people, don’t be afraid of this field. I think anybody that likes research and is interested in learning about technology can thrive. And it’s one of the few places I think, where you can come in from a totally different career path and carve a space for yourself with unique skills. So I would just say don’t be afraid of ontology.

Larry:
Thank you. No, usually, this is another thing about content strategy that comes up all the time is impostor syndrome. And your advice would be like, you’re not an impostor. Your work is welcome here as anybody else. So thank you for that. Yeah. Well, thanks so much. I’ve got one very last thing. What’s the best way if people want to follow you? Do you have a social media account or a place you like to connect with people online?

Nicole::
Yeah. I do. Yeah. Twitter is probably the best place. And my handle is my first initial and last name with no vowels. So it’s N-B-C-K-N-W-L-F it’s like nbcknwlf, no vowels.

Larry:
Got it. Okay. And I’ll put that in the show notes as well for folks. Well, thanks so much, Nicole. Really fun conversation. I really enjoyed talking to you.

Nicole::
Thank you. I appreciate it.