Creating Structure out of Unstructured Data
Release Date:
Data is everywhere! And CX pros have structured ways to keep it all organized. But what about all the UNstructured data your customers provide? While there are some challenges in using unstructured data, it has the potential to bring a rich set of insights about your customer experience efforts. Host Steve Walker welcomes Ellen Loeshelle, director of product management, XM Discover at Qualtrics, for a discussion on how customer experience professionals can take advantage of the tools available to extract the value from unstructured data.
Learn more about Qualtrics XM Discover at https://www.qualtrics.com/discover/
Ellen Loeshelle
Qualtrics
Connect with Ellen
Highlights
Unstructured data is everywhere
“Every nook and cranny that you’ve got unstructured data in your business, send it our way. And so the survey, of course, is a mainstay in that cabinet. But reviews, emails, calls and chats make up an increasingly large percentage of the data that we’re dealing with. Social media, private messaging, chat bot data everywhere. It’s amazing. And it comes in so many different flavors that it’s important to use technologies that are well fit for that data type…”
Structured and Unstructured data go hand-in-hand
“…[structured and unstructured data] go hand in hand. And honestly, unstructured data without structured data is also equally powerless to some degree. Right? Because if I have a lot of I got a lot of feedback and people are saying, this is what I like, this is what I don’t like, but I don’t know who they are or the amount of money that they’ve spent with my organization or maybe where they’re geographically located or any of their personal demographic information. Then I also lack the confidence to make decisions on behalf of our customers because all of our customers are not created equal…”
Transcript
The CX Leader Podcast: "Structuring Your Unstructured Data": Audio automatically transcribed by Sonix
Download the “The CX Leader Podcast: "Structuring Your Unstructured Data" audio file directly.
The CX Leader Podcast: "Structuring Your Unstructured Data": this wav audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.
Steve:
We CX pros love our data and we like it to be well organized. But if we're willing to get our hands dirty, we could harvest a vast amount of data that our customers are just giving away.
Ellen:
Unstructured data really allows us to not understand the what but the why. When I take that information to my stakeholder and say on average, people are likely to recommend our business nine out of ten or heaven forbid, two out of ten. What's the next question that they're going to ask? Right. It's always going to be like, well, "why?"
Steve:
How customer experience pros can tap into their unstructured data on this episode of The CX Leader Podcast.
Announcer:
The CX Leader Podcast with Steve Walker is produced by Walker, an experience management firm that helps our clients accelerate their XM success. You can find out more at walkerinfo.com.
Steve:
Hello everyone. I'm Steve Walker, host of The CX Leader Podcast. And I'm glad you're listening. It's never been a better time to be a CX leader, and we explore the topics and themes to help leaders like you deliver amazing experiences for your customers. Data is everywhere and CX pros have structured ways to keep it all organized. Now, I know you're thinking, Steve, the existence of structured data would seem to suggest the existence of unstructured data, and you would be correct. While there are some challenges in using unstructured data, it has the potential to bring a rich set of insights about your customer experience efforts. Well, to help us better understand the value of unstructured data, we've invited an expert from Qualtrics to come on the show and tell us more. Ellen Loeshelle is a director of product management XM Discover at Qualtrics. Ellen, thanks for being a guest on this week's CX Leader Podcast.
Ellen:
Thank you so much for having me. Steve. I'm happy to be here.
Steve:
Well, it is a pleasure and this is a topic that is really, really timely for us CX pros. And I'm just so delighted that you're willing to come and share your expertise on this. You know, as usual, in case people don't know who you are, even though they should, why don't you just give us a quick background of your career and what got you to this point where you are an expert in the delivery of unstructured data platforms?
Ellen:
Yeah, absolutely. So as with most people's career, I think it's a lot of serendipity that got me to where I am today. So I studied languages, linguistics, cognitive science in undergrad and was really, really lucky to stumble upon Clarabridge right out of undergrad. And for those that aren't familiar, Clarabridge at the time that I joined ten years ago was a bit of a late stage startup focused on text analytics and natural language processing in the really burgeoning field of at that time what we called customer experience management CX today. And it was a great fit for me because I loved the technology, the products that we were working on. I started my career actually in services. I did two years implementing our technology, helping our clients who many of whom were early adopters in this space, trying to figure out what works, how to get the attention of their executives and their stakeholders. So I did that for two years, and then I had the opportunity to switch over into product and help drive the the core technology, the NLP technology behind Clarabridge for a number of years and then had the opportunity to graduate up and oversee the whole the whole department effectively. And then with the Qualtrics acquisition, I got to absorb even more and now also oversee all of the text IQ assets. So really this thread of unstructured data, of language, of human experiences has really been a constant for me throughout my career. And it's it's just been so wonderful to grow up personally and professionally in this space as it's grown up alongside me.
Steve:
So one thing you said there is really fascinating to me. So you were really more of a language person?
Ellen:
Yes. Yes, absolutely.
Steve:
So were you originally brought in to deal with some of the non English language part of it?
Ellen:
No, actually. And what's so interesting about linguistics, I think about it honestly more as a science rather than an art, because linguistics just gives you the tools to understanding the structure of language and the things that people are trying to communicate. And they weren't necessarily recruiters weren't necessarily looking for linguists, but it spoke to me definitely, and I think did give me a differentiated perspective on how to look at the unstructured data that people were submitting on behalf of their company experiences.
Steve:
No, this is the whole DNA of like the sentiment analysis and…
Ellen:
Exactly, exactly.
Steve:
…that's really, really cool. And, you know, I don't know if I shared this with you, but we were one of Sid's first customers.
Ellen:
Yes. Yes, I heard that. So I think you guys predated me, honestly, At Claridge?
Steve:
Yeah, I met Sid, like, literally, right as he was getting going. I think I knew him before. He'd had another business before.
Ellen:
Yes, he did.
Steve:
We had a really high profile Silicon Valley client who was visionary. And. And, you know, this is like in the late 2000, maybe 2007, 2008, 2009. And he really pushed us to find a solution. And so we partnered up with Sid, like really early on, and we're one of the first resellers. And then we, we actually stayed with Clarabridge until we partnered with Qualtrics. So we dumped you guys for Qualtrics. And then, lo and behold, look what happened.
Ellen:
So, yeah, we're all the gang's back together, right? It ends up the way that it's supposed to be.
Steve:
Well, I appreciate you sharing your story, and I think that's really neat how you came up in the business. But it's also one of the things I always remind everybody. You don't want to burn any bridges because it's a it's a long career.
Ellen:
And it is. And people come back around for sure.
Steve:
That's right. For sure. All right. Well, you actually use the term unstructured data in your opening introduction. But just again, for purposes of setting the table, why don't you… Because it's a term we use a lot, I think, in the nitty gritty details of it. But why don't you define unstructured data for us?
Ellen:
Sure. So unstructured data is the open ended part of feedback. So if we think about a survey, just for an example here, the survey is going to have some questions that's going to ask you to pick maybe from a scale or from a dropdown or radio buttons or something like that, where the answers are constrained by the person who set up the survey. And then most of the time, or in my case, hopefully all the time, it's also going to have a space that you can actually write free form about your human experience, right? What happened? What worked, What didn't work for you? That open ended box, that text field. That's the unstructured data, right? There's no limits to what you can say there, and it's not constrained by the person who created the database or wrote the survey up front.
Steve:
Yeah, I love that. So closed in questions and open end questions are sort of like research type stuff, but that open ends are how we get to the text and the unstructured data. And actually just a little tip for everybody out there in CX land. This is your your bonus of today's podcast. But when you're trying to engage another person in in conversation, it's always better to ask open ended questions than closed in questions. So like instead of saying, do you want to go out to eat, you say, Well, where would you like to go out to eat? Or so forth. So anyway, yeah.
Ellen:
And the same is true in customer feedback.
Steve:
Yeah, I digress. However. So. All right. Well, why is unstructured data so valuable for CX pros?
Ellen:
Unstructured data really allows us to not understand the what, but the why behind some of these experiences. So if we ask again, just to use the survey example, if we ask them how likely they are to recommend our business or how satisfied they were with an experience, they might give us a number between one and five, one and ten and seven, whatever our scale may be. But when I take that information to my stakeholder and say on average, people are likely to recommend our business, nine out of ten or six out of ten or heaven forbid, two out of ten, what's the next question that they're going to ask, Steve? Right. It's always going to be like, well, "why?" Right? And it's all if all we have to go on is, is that number or maybe even a label in some cases, we're not going to be able to bring that customer's voice into the boardroom or into even our conference rooms to make decisions on their behalf. We really, really need to understand what's going on in their head that led them to to that conclusion or to that number or evaluation at the end of the day. So unstructured data is really what helps us make our customer feedback or our employee feedback or whatever kind of feedback we're looking at, really make it actionable and make it make it more human, Right?
Steve:
Yeah. And it's a virtuous cycle too, right? Like your example, you said you follow up the closed in with an open end. But oftentimes when we're trying to construct closed end questions, we might do some qualitative, you know, like how do people talk about something? And then, you know, how would you craft that? So I guess where I'm going with this is how does unstructured data complement structured data?
Ellen:
Yeah, they go hand in hand. And honestly, unstructured data without structured data is also equally powerless to some degree. Right? Because if I have a lot of I got a lot of feedback and people are saying, this is what I like, this is what I don't like, but I don't know who they are or the amount of money that they've spent with my organization or maybe where they're geographically located or any of their personal demographic information. Then I also lack the confidence to make decisions on behalf of our customers because all of our customers are not created equal, right? They're interesting cohorts that have different preferences, different needs that we need to cater to. So really, the strength of our technology at formerly Clarabridge, now at Qualtrics and many other vendors in the space is the ability to combine both of those things together to say, Hey, we really want to know who you are, but we also want to know what makes you tick, right? What what actually is going to drive you to have engagements with us or leave us right for our competitors. And when you can do that dance between the two is when you start to really unlock the power of customer feedback.
Steve:
Yeah, it's a great explanation. Why don't we talk a little bit now about kind of the application of it. So give me a couple of examples of how this can actually go and impact the way we create our processes or or change how we're doing business for the benefit of improving the customer experience.
Ellen:
Yeah, well, there are a lot of, a lot of different ways that we can do this. One story comes to mind with a company that I worked with a long time ago. They they make kitchen appliances like small household appliances, and for a long time they had been reticent to produce pressure cookers because pressure cookers, people were concerned about the safety of pressure cookers in their homes. Right. Worried that they're going…
Steve:
Dangerous.
Ellen:
.. to blow up their kitchen. Sure. And so they had they they really had steered away from doing that. And they were looking at review data. So like Amazon reviews and other sorts of e commerce reviews, not only of their own products but also of their competitive products. And I love review data so much because it is rich with both structured data and unstructured data, Right. Of course you have that that text box where you can explain your experience, but you also often have information about what product, how much did it cost, what was the dimensions of the product, Who is the buyer? Right? We can verify the buyer. There's like of all sources in terms of quality of structured and unstructured and being unsolicited, being like the company didn't ask for it. It's the customer going and selling it. It's one of my favorites anyway. So they were looking at all of this data together and they had started to notice this trend where people were talking about Instapot or Instapot-type products in a different sort of light. So instead of saying that they were worried about the safety of, of pressure cooker type products or Instapot type products, they were starting to say they were not worried about it.
Ellen:
And so we had the ability to look at both the category of products that they were talking about, who these people were and the demographics of these people and who the who the buyers were potentially right. Often families, busy families who need to make meals on the go, but also look at the unstructured data and extract emotion, not just topics and sentiment, but emotion and effort and all of these other things. And that information actually helped this company decide to enter this market with this new kind of category that they had always steered away from. So being able to look not only at what is the SAT score or something like that gave them the confidence to make a business move that they never would have been able to do if they were only looking one internally at their own product data or only at structured data. It was really the combination of those things that that gave them the confidence to do that. And so we've seen examples like that all over the place, right? It can change the way you think about internal processes, external facing processes, products, services, really the whole gamut.
Steve:
You know, and I love that example you just gave for two really specific reasons. One is, you know, you just showed an example of where CX is innovative. You know, it's not just fixing a broken process. It actually gave them confidence to go and do something that could help grow the business. The other thing you mentioned that I should have gone to you earlier, and I think some of this is because of my background, but early on we thought of unstructured data simply as open ended questions. But in this case, you're grabbing unstructured data from not a survey but from a review site. Right? And that is unstructured data that in fact, it's unsolicited… I guess it is solicited. But, you know…
Ellen:
In that case, I think unsolicited, it can be either right sometimes either ask you to submit reviews. Right. And sometimes it's totally on your own. So it could be either.
Steve:
Yeah. And actually, the reengagement with Clarabridge 2.0 here now is part of Qualtrics. I couldn't believe how far you guys had come in, I guess what you call omnichannel…
Ellen:
Omnichannel. Yeah, exactly right, so…
Steve:
Taking unstructured data wherever you can find it.
Ellen:
Give us all of it. Right. Every nook and cranny that you've got unstructured data in your business, send it our way. And so the survey, of course, is a mainstay in that cabinet. But reviews, emails, calls and chats make up an increasingly large percentage of the data that we're dealing with. Social media, private messaging, chat bot data everywhere. It's amazing. And it comes in so many different flavors that it's important to use technologies that are well fit for that data type, right? There's a lot of investment that we made over the years to change how we're looking at language because the way that I communicate as a human, if I'm just right, emoting in a survey response or even in a review where it's just one direction, right? I am just letting out my feelings to the world, to the company or to the world is very different than if I were to call you Steven. Your you're my call center rep. And we're trying to work out a problem together, Right? It's not going to be as emotional or as sentiment bearing. Right. It's going to be a lot more transactional because to be honest, I'll probably want to get off the phone with you and you want to get off the phone with me so we can get get back to our day. So thinking about the structure of language and also the purpose of language was absolutely critical for how we evolved our technology to meet this industry as it was expanding.
Steve:
I want to talk a little more about the contact center, because, again, just in in my reengagement with you all over the last year or so since he became a part of Qualtrics, I think about like survey fatigue, you know, and how sometimes we we can't get the response rates well in a contact center you might, with XM Discover you might have 100% feedback on every call, right?
Ellen:
You bet. Yeah. I mean, the volumes are astronomically different and also the content is also very different. But it's interesting to see where those things overlap, where they're where where they differ. Right. And you can come to some interesting conclusions about what matters to your what matters to your constituents, what makes them tick, what makes them likely to come back the next day.
Steve:
So really, you kind of take unstructured data and you can make it structured.
Ellen:
That's exactly right. We create structure out of unstructured data. And I will tell you, the original impetus, the inspiration for the name Clarabridge was creating clarity and bridging that gap between structured and unstructured data. So it has always been at the core of our DNA. And I think increasingly as part of the Qualtrics family, it has given us way more tools to think about structure of data that we didn't ever have access to. Right? Suddenly, we now have tools to allow us to think about the whole person, right? With the XID profile where historically it was really more each record for themselves, right? Each moment or each interaction for themselves. But it's all about the people, right? Everything we do in CX is about humans, humans on the customer side, humans on the employee side, and often how they interact together to create these ideal experiences. And we're just so jazzed about the ways that we're going to be able to use unstructured data to understand people better and really to provide humans better experiences for each other.
Steve:
Yeah, well, I'm jazzed with you, too. It's it's exciting. And just for our listeners, again, you know, this is state of the art stuff that wasn't really practical five or ten years ago, but now today it is. And, you know, if you're struggling with, say, your transaction surveys within your contact center, you could take some of your unstructured data. You could match that up where you could get some structured feedback from data, and then you could write algorithms and employ the XM Discover platform to basically create really, really good, structured transactional data around the quality of your contact center. So again, I just I want to lay that out for our XM pros because, you know, this this is kind of mind blowing stuff when you really think about it.
Announcer:
Are you looking for a little recognition for your hard work? Well, here's just the opportunity. Applications are now being accepted for the US Customer Experience Awards. Finalists and winners will be named in 18 different CX categories, and you could submit an entry in multiple categories. This could be the chance for your team to finally get the recognition it deserves. To find out more and submit your entry, go to usacxa.com.
Steve:
Hey, my guest on the podcast this week is Ellen Loeshelle, and she is director of project management at XM Discover at Qualtrics. She's given us a fascinating discussion of how we could better use our unstructured data to drive better experiences for our customers and employees. What are some of the common pitfalls or the common challenges that you run into with unstructured data?
Ellen:
Yeah, there are a handful of things. I think data quality is always a concern. We actually were just talking about all these different data sources and they're not all created equal in terms of the integrity of the language that's there in the our ability to get good, clean, easy insights out of all of them. So certainly anything that is transcribed represents another layer of abstraction that goes between the speaker and the words that are on the page that can result in some misinterpretation. Even in emails, right, they're really junky with email signatures and pictures and all this other stuff that even when I receive an email, sometimes it takes me a few minutes to parse through it. So focusing on data that's easy to read, easy to consume is always where I recommend people start when they're thinking about unstructured data, right? Go for the sources that are available to you and that are clean and grammatical is just going to make everybody's job easier. One of the other major challenges, though, that I often hear is a lack of appetite. Often at the executive level. If your organization is not used to using qualitative data, open ended data unstructured data, it's hard to to create the muscle, to trust it, right? A lot of times people are so attuned to looking for specific numbers, quantitative things, statistical metrics that they're like language that feels very fuzzy, right? It feels squishy.
Ellen:
Like how do I actually trust it to make business decisions based off of it? And it can take some time to really evolve that thinking and get people on board. But what I normally recommend is start small, have modest objectives first, get a few wins and really appeal to people's heart. Right as as humans. Like that's that's where we connect. Right. So if you can find a few early wins that really help people empathize then like oh yeah, of course we have to do this. Like how could we ignore what our customers are actually saying to us? And then you build up, you build up that muscle, you build up that energy, and before you know it, it is part of your culture. And you could never imagine going a day without it.
Steve:
Well, I really like your idea about some early wins. And actually that ties in. And this is going to be a little self serving here for the podcast because your episode is going to drop right in the middle of our series we've been doing on the emerging evolution of CX. And I think that this fits right in there. And one of the things that I did throughout that series with each of our experts is say you kind of put them through a little exercise here and say like, okay, I'm a CX leader. I'm not really using my unstructured data very well. I got a call center, I got reviews on my website. What do I do first? How do I get started?
Ellen:
Yeah. So I think the first thing you need is to find a technology partner for you, and that's also obviously a self serving comment. But I truly do mean it is. We see a lot of organizations who try to reinvent the wheel within their teams to say, Oh, we've got a bunch of smart people, right? Let's have them figure out how to how to parse language. And like I love smart people and I love language, language nerds and those, but I would rather deploy those individuals on top of technology that already exists rather than having to have them spend their wheels reinventing. So anyway, technology I think, is incredibly important to allow you to ingest this data, start to make sense of it at low cost to you or low effort to you at least, because otherwise it's going to be so cost prohibitive to even get started to find those early wins that we're never going to get to that muscle building that that we want to do. So see what data sources already exist within within your ecosystem that are easy for you to get. Often public data sources, right? Like social or survey ones that are owned by your department are easy to obtain and then start to figure out what are those business questions that are making your program tick right and start to explore from there.
Steve:
You know, I love your little story about trying to do it yourself. And just for those of us from my generation in the market research profession, we used to have a department called coding, and basically it was people sitting at long tables and and they were transcribing the open ends into, you know, themes and organizing and basically collaborating on how it ought to be structured. And again, that's another place where software. Where has made that not such a relevant position anymore. But I love your concept. And yes, because of the advancements in things like AI and data storage and stuff, it's it's much easier to do this on top of a platform.
Ellen:
Totally. Right. And we deploy our people to things that they're good at, which is understanding people and reasoning on on top rather than being computers themselves.
Steve:
And then communicating back to the operation what the impact ought to be on the business or how we could make it better for our customers.
Ellen:
Yeah.
Steve:
Ellen, we talked a little bit about sentiment and just unstructured, but what is kind of the next level of that? What else can you do with this past sort of the entry level of just making sure that you're leveraging your unstructured data?
Ellen:
I'm glad you asked, Steve. My my provocative and somewhat hot take on this is, I think topics and sentiment are kind of boring and I feel like I'm entitled to have this opinion because I lead products that build these things. But topics and sentiment to me are very much table stakes, right? When we they're kind of like satisfaction scores. And that sentiment is not something that makes sense to a lot of people, right? We have to sit down and explain to our stakeholders how they should interpret it. But when it comes to language, right, human to human, you and I get each other because we're both English speakers, right? We've been speaking this language for decades and we connect on this very different levels. So where I encourage people to go beyond topics and sentiment is to start thinking about emotion and effort and intense purpose, right? Empathy. These are all signals that we can pull out of text at scale to use to drive business change and to really connect with the hearts and the minds of the other individuals in your businesses. And those are the things that actually drive action. Right. And I'll give you one example real quick is if if if I were to tell you that you're the sentiment for your website was a negative two right, people didn't like it, you'd be like, Why? Why? What do I do about that? But if I come to you and say people found your website, the navigation on your website to be confusing, frustrating, and they're never going to shop here again because of it, you'd be like, Oh, that catches my attention right now. I know exactly what I need to do. I need to talk to the Web team. We need to reconsider our UX on this and we can move forward, right, and monitor it over time. But just a sentiment score, just the satisfaction score, they're not going to tell us that information.
Steve:
So I was thinking your comment about topics and sentiment are kind of boring. So I was thinking like later in life I've tried to learn Spanish, you know, mostly on tools and stuff like online stuff, and I can't really speak Spanish, but my vocabulary is pretty big. You know, I do interact now with native Spanish speakers on a pretty regular basis today. You know, we can understand humor even though we're not speaking the same or we're not native in the same language, we can feel things. And I think that's what you're talking about with kind of the next level of this, Right?
Ellen:
Totally. Right. We understand people, even if you don't share native language, you can feel you can share feelings with people, right? You can empathize with someone's experience by observing them or even just like kind of feeling the pitch or the inflection in the way that they're communicating. And I think that's a human quality, right? That's that's what allows us to connect with each other. Sentiment feels very corporate. Yes, Right. Satisfaction scores feel very corporate, emotions feel very personal, feel very human. And so that's kind of the theme I think that I've really run with today, is thinking about unstructured data is really about just connecting to people. Yeah, and that's what I love about CX, and that's really the evolution of the experience management space is how do we go from just thinking about numbers to actually thinking about the people on the other side?
Steve:
Ellen Loeshelle, thank you for being a guest on the podcast. But we have reached that point where I ask every guest to, you know, you've given us a bunch of stuff that we could take home and make some value with, but I always ask for, you know, the one tip, the take home value. This is the thought that our CX pros ought to take back to the office with them after they listen to this podcast and say, I could do that and help my organization move forward in its customer experience. So Ellen, local take home value, please.
Ellen:
So I think I'm going to riff off the discussion we were just having around connecting with people, and I would encourage all of you to think about the stories that have resonated the most with your stakeholders and your executives and what part of those stories, those customer stories have made them light up. Was it somebody's emotion? Was it an empathetic moment? Was it a certain level of effort? And those are the sorts of questions that you want to start asking of your unstructured data, right? So put sentiment to bed if you must. It's fine. Think about actually what's making people tick and start to spin the way that you're looking at your data and you're analyzing your data day to day.
Steve:
Thank you for that. Take home value. Thank you for the entire podcast. Ellen Loeshelle is director of Project Management at XM Discover at Qualtrics. Ellen, thank you so much. If anybody would like to continue the conversation, I know I found you on LinkedIn so I could find you Loeshelle is L.O.E.S.H.E.L.L.E.
Ellen:
You got it.
Steve:
And then they could probably find you a Qualtrics too, right?
Ellen:
Yeah. eloeshelle@qualtrics.com.
Steve:
If you want to talk about anything else you heard on this podcast or about how Walker can help your business' customer experience, feel free to email me at podcast@walkerinfo.com. Remember to give CX Leader Podcast a rating through your podcast service. Give us a review. You get some unstructured data out there that we could analyze. Your feedback will help us improve the show and deliver the best possible value to you, our listener. Check out our website cxleaderpodcast.com to subscribe to the show, find all our previous episodes, podcast series and contact information. You can drop us a note, let us know how we're doing. The CX Leader Podcast is a production of Walker. We're an experience management firm that helps companies accelerate their XM success. You can read more about us at walkerinfo.com. Thank you for listening. And remember, it's a great time to be a CX leader. We will see you again next time.
Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.
Automatically convert your wav files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.
Sonix has many features that you’d love including enterprise-grade admin tools, automatic transcription software, powerful integrations and APIs, secure transcription and file storage, and easily transcribe your Zoom meetings. Try Sonix for free today.
Tags: unstructured data XM Discover data Qualtrics Steve Walker Ellen Loeshelle