Sahana Murthy is the General Manager of Loveland Technologies, where she manages the team, the product, and the overall corporate & marketing strategy. Prior to Loveland, Sahana had extensive experience working with software companies and startups of all sizes across USA & India. Over the years, she has worn multiple hats with roles ranging from a software developer, developer evangelist, product manager, and product marketer to most recently leading a startup as its COO.
Sahana was interviewed for GeoHipster by Mike Dolbow.
Q: I’m told your current job is your first in the geospatial area. Tell us about your journey to get here and what drew you to this gig.
A: I have been in tech for over 10 years now. Started off as a techie working on proprietary software for a large tech conglomerate. I was lost in a sea of employees and techies, never fully knowing what value my work was adding to the larger scheme of things at the company. There was very little creativity and very little autonomy to my work. And then, one fine day, I discovered the magical work of open source and tech startups. That totally changed the trajectory of my career path and interests, in general. I haven’t looked back since. I have been working with tech startups in different industries and technologies from AI/ML to SaaS/PaaS to now a GIS software & data company. With every new gig, I have looked for a different technology and product suite and that’s what drew me to Loveland. The world of GIS and an incredible team that has tirelessly put together a comprehensive dataset of 143 million parcels across the US. No small feat!
Q: Loveland aims to be the place that all sorts of folks seek out for information on land parcels, and you’ve pieced together a nationwide dataset with pretty amazing coverage. Did you know how much of a “holy grail” this was for geospatial folks before you joined Loveland? What kinds of reactions have you gotten from new users?
A: Yes, we are the “Go-To-Source” for all things parcel data. 🙂
I sort of knew how important and valuable this dataset was before I joined the company. But I think I now truly understand how hard it is to :
acquire and collect this data
Standardize it across the board. Every county’s data is so different from another. So normalizing it – cleaning up the data for easier consumption – is easier said than done.
provide it in 5 file formats to ensure customers get what they want
cut through the bureaucracy to obtain the data from counties
find and integrate valuable datasets that can augment our already valuable data – like the USPS vacancy dataset and buildings footprints data.
The reactions we receive from users have mostly been about the quality and coverage of our data, the price, and our transparency. We are the most affordable parcel data vendor with high quality coverage. We communicate all of our data updates to customers on a monthly basis. So they are always aware of what we are adding, updating and improving. Our origins lie in “democratizing data” and we believe we are doing that well.
Q: I myself know what kind of an immense task this is, since I’ve been in charge of compiling similar data for Minnesota’s 87 counties for years. What’s the hardest part to automate?
A: Well, we’ve built a state-of-the-art ML model that lets us clean parcel data to perfection in a matter of seconds! 😉
No, but seriously, the hardest part to automate is normalizing and cleaning the data! We do rely on a very robust set of open source geo tools like PostGIS, QGIS, and GDAL. There’s still a lot of manual work backed by the judgement and experience of our parcel team.
Nearly every one of the 3,200 counties in the US manages its data independently. There’s very little consistency from place to place even for basic columns like “owner”: it could be “deedholder, “firstname”, “fname”, “propow”, or pretty much anything else. And you’d think that a field called “Parcel ID” might be unique — but it rarely is.
Generally, counties within states are more like each other than counties outside of those states, but not necessarily. Also, cleaning the data can be quite tricky at times. Working with this data makes it apparent how much human error can be involved, especially when it comes to casting columns that should be integers or double precision from text. Date field conversion can also be quite tricky at times, depending on the formatting used by a place. A lot of rules built up over time plus the keen eyes of our data team help us keep quality high.
Q: I know there are a lot of counties in the US that sell their parcel data in an attempt to recover costs. Do you think these counties are aware that you’re repackaging their data and selling it yourself? If so, have you faced any opposition from local government officials?
A: Some counties have been slower or more reluctant to openly share their parcel data than others, but the overwhelming trend we see is towards more data accessibility. Over the last five years, many more counties and entire states have moved towards open parcel data, and we haven’t had any problems in displaying, sharing, and providing services around it. It’s not uncommon for us to have cities and counties as customers and they all want easy access to the data.
As time goes on, we anticipate that the public facts contained in parcel data will be ubiquitous, and the value that we add to the base data through organizing, standardizing, and adding additional data from other sources, including machine learning, will be where a lot of the value is. This is also a value add that we can provide back to counties, so it’s a win win.
A few years ago the team did some research to see how much county assessors who do still sell parcel data are making from it, and who they are selling it to. The numbers were very low and the people they are selling it to are often resellers who mark it up and resell it. We’re definitely not alone in the space of selling parcel data, and we work hard to be good, positive actors who are helpful to local governments and the public.
At the end of the day, parcel data is made of public record facts about how the country is subdivided, owned, taxed, and used. The public nature of the subdivision of land in the US stretches back to the earliest days of the country and big public programs like the US Public Land Survey which started in 1785. We see ourselves as providing a missing service by bringing all these local datasets together into a big picture of how we own, use, and inhabit the country. The reaction to that has been positive.
Q: Are you scraping any REST endpoints like OpenAddresses, or are you more frequently downloading from open data sites and then loading into your database?
A: The nationwide trend towards making parcel data open to the public has been important to us. We do download data from data portals and scrape from public REST APIs. Our team of parcel prospectors is both incredibly nice and incredibly talented at collecting the data. We haven’t yet gone to digitize counties that are still paper-based — maybe someday.
Q: How does the team decide what attributes to standardize on across the country? I’ve seen parcel datasets with some 90+ columns in them, but you seem to be flexing what you serve based on the sources. Is that hard to stay on top of?
A: We roughly based our schema around what the State of North Carolina uses to standardize their parcel data. We try to use columns that are broad, but also relatively unambiguous. There are occasional movements towards a federal standard, and that would be amazing.
Most places don’t have that depth of data as North Carolina has been working on their statewide dataset for a while, and helping local assessors get up to that level of detail. If a place doesn’t have as many fields, that’s what we work with.
As far as custom columns, we generally try and take a maximalist approach. For clients working exclusively within a very local scope, you never know what column is a must-have. The benefit of having the schema is that at a larger scope it allows one to be pretty flexible in data exporting without porting over a ton of custom fields aggregated from various counties.
Q: Congrats on getting your data into Carto’s Data Observatory. What does that mean for your company?
A: Thank you. We are thrilled about our partnership with Carto. We love Carto and everything they have done and built in such a short span of time. They are a benchmark in the GIS world and to get our data into their product suite is absolutely amazing! What this means for us, I guess, is it’s a testament to the quality of our data and coverage. We have always known that, but when we have trusted members of the GIS world like Carto and Mapbox believing in our data, it’s great validation for all of our work! Our lean team of 14 works tenaciously to improve our product every single day and now everybody is starting to see that.
Q: I guess you can tell I love talking about parcel data. Let’s diverge, though: what kinds of things do you like to do in your free time? Any hobbies?
A: I’m guilty of varied interests! Although as a mom to a hyperactive 2 year old, I guess I don’t get to indulge in hobbies as much as I used to. 🙂
I love to cook, sing, and read. I’m sort of an amateur food blogger on Instagram right now.
Travel has always been a passion. Growing up, I maintained a travel log and dreamed of a career in travel some day. Of course, those days are behind me, but the passion for travel still continues.
Growing up in India, I had made a vow to travel to every state in India. I think I am at a 95% completion rate. It’s not so much about checking off the bucket list but more about discovering the rich cultural heritage and differences in every place. India is so diverse. People speak and write a different language, just 50 kilometers apart. And the only way to truly experience that is by traveling to every nook and corner. And of course, discovering the range of local foods is absolute heaven for a foodie like me. The hope is to do the same here in the US, and eventually worldwide.
Q: Since you’re new to geo, you probably haven’t read a lot of GeoHipster interviews. What comes to mind when you hear the term ‘geohipster’?
A: I actually have read some since I started with Loveland. I love the work that many of your featured geohipsters are doing. Although when I read those, the irony is all too evident to me. I’m an imposter GIS person at best right now (that will change in a few months I reckon). 🙂 My team however is GIS through and through.
The term ‘geohipster’ to me is someone, anyone – techie, non techie, GIS or non-GIS who’s passionate about their local geography and land grid! It could be an urban planner, a surveyor, a cartographer or someone like our CEO – Jerry Paffendorf who’s dedicated the past ten years of his life to change the property landscape of Detroit and continues to do that still, in his own way, outside of his work at Loveland.
Q: What kinds of advice would you give to folks in startups and SAAS looking to diversify into a field like geo? What advice would you give yourself the day before you started?
A: As someone who comes from a long line of tech startups and SaaS, the one thing I can guarantee is that working in geo would bring in just the same amount of challenge and excitement as any other industry. Typically, that’s what most folks who prefer working in startups are drawn to – an exciting, fast paced industry. Geo has not disappointed in that regard. Also, GIS is a very large umbrella, brimming with opportunities. The surveying and mapping market in the US is about $9.3 billion and growing, as of 2019. The sky’s the limit. Cooler things are happening and being built as we speak with droning technologies and use of AI & ML on aerial imagery. It’s an exciting time to switch over to Geo.
For me personally, since we are the “parcel/cadastral kind” in GIS, the advice I would have given myself is to prepare for working with the public sector. Until now, I’ve only worked in the private sector and managed customers in private sector verticals. So working with counties and the public sector is a different experience that I’m now learning to navigate. But, the exciting part about my job is that I get to straddle both sectors.