Porch is a home improvement network that's built upon a professional graph of over 1.5 million people, 90 million projects, and $2 trillion. Read about the startup's innovative use of open and closed data.
Back in my twenties, I was apprenticed to a general contractor for 18 months and then spent a year renovating a house and working for another builder. In the process, I learned a lot about hard work, craft, design, discipline, materials, power tools, safety, teamwork, and staying warm in New England winters, among other lessons. I also learned that there's a gradient of professionalism and skill through the trades. Unless you have inside insight and relationships, it's difficult to know who to trust, and for what. While subscription-based online resources like Angie's List or Consumer Reports exist and review sites like Yelp can help, homeowners still could be better served in their neighborhood.
That's the challenge Porch has taken on, developing what CEO Matt Ehrlichman describes is the first "home improvement network." The service is built upon a "professional graph" of over 1.5 million people, 90 million projects, and $2 trillion of spending. Erlichman said that Porch has gathered insights about approximately 75% of homes in the US. Homeowners can search for professionals for free. Home improvement professionals can create free business profiles and then choose from a variety of premium services to gain more business from Porch users.
"The first core problem we're solving is helping homeowners improve their home," he said, in an interview in Washington, DC. "It's a way to not only get educated about the products around them and what they cost, but also find the right professional for their specific home that their neighbors have used in the past."
In a world where I've seen hundreds of smart public officials, developers, entrepreneurs, and civic hackers look for ways to use government data over the years, Porch's story jumps out. When I first read about it in The Wall Street Journal, it looked like a smart concept applied to an opaque market.
"It's remarkable," he said. "Despite being the largest purchase most people make in their life, they didn't have a good way to make decisions about how to maintain it. Technology has done a great job of making more and more information available, but the home remained perfectly opaque. There's so little info about what a project might cost, or should. So, we started Porch over a year ago and were lucky to raise a $6.25 million dollar seed round."
When I talked to Ehrlichman, I realized it was an important example of a data-driven company for several reasons.
First, the startup's founders went after a specific market (homeowners) and focused upon solving a problem they all have: finding trustworthy professionals to improve or maintain their houses. What Porch has effectively done is to make digital what had been primarily an analog system of word-of-mouth recommendations. It came to be because Ehrlichman wanted to scratch an itch, when he was in the process of building a home in Seattle. He found the process of getting educated in whom to trust for what kinds of projects to be so painful that he looked for a way to improve upon the process.
Second, after deciding on the problem, they went acquiring data that would provide the variable that would enable a more informed decision, gathering many slices of data not only about the people who work on homes but the projects they do and the homes as well. "We started by just aggregating data: public data and lots and lots of private data, through hundreds of partnerships," said Ehrlichman.
If you've read my previous coverage of Panjiva, which built a search engine for shipping based upon public data, this approach should be familiar: acquire and invest in a data science team to clean and structure public data. That data can then act as the backbone of a map of a market for goods and services, and then layer private data on top of it for additional value.
Third, after a map for a given market has been built on that information backbone, create incentives for the people or the businesses described in the "graph" to join, update, and maintain profiles for people searching for their services. When BrightScope created a professional graph for financial advisors using public SEC and FINRA data, for instance, they did much the same thing, charging advisors only to add pictures, links to LinkedIn, and other information. Making that data more liquid and searchable by name and location also provided a strong incentive for advisors to correct errors in the government data regarding their records.
Finally, and perhaps most controversially, understand that data is a strategic asset, and move to defend access to it as a differentiator from competitors, present and future.
"We learned what information you can get from the public data, and then we really focused our efforts on these private data partners that are out there," said Ehrlichman. "Before we even let the world know that we existed, in the summer, before we launched in September , we wanted to lock up the data. We wanted to come out and not be replicated. We see very clearly that data affords us the ability to do things that are not only unique but durably unique."
While Porch is currently focused on home improvement, the data it's sitting on will likely enable the startup to do other interesting things over time. To get a deeper sense of how they went about the process of adding more data-driven transparency to the home improvement market and accomplish something genuinely innovative, dive into the following interview with Ehrlichman, edited and condensed for clarity.
TechRepublic: What's the business model?
Ehrlichman: It's free to homeowners, and will always be free. Professionals come in for free as well. We have a subscription product as well for the professionals. If you look across the industry, companies have started to fray the consumers' trust. Everyone that pays is at the top. We have drawn a really hard line that that's not what we're going to do. When you're letting someone into your home, with your children and your valuables, there's a different bar of trust. It's always based on who your neighbors use and who's done great work, not who's paying.
TechRepublic: Does that mean showing sponsored search results, like Google does?
Ehrlichman: We could do that. When people search, we want people to find the best professional. It's in their interest. We can still get exposure to subscribers by showing their projects more often when people are browsing through the projects of nearby businesses. We can also include them more often in emails to homeowners about projects that are happening in their neighborhood recently.
TechRepublic: What sources are you using to review this work?
Ehrlichman: We have an endorsement function, similar to LinkedIn's endorsements. A homeowner can come in and, not just endorse generally, if they're good, but endorse a professional specifically for the type of work that they did. They might have been timely, well-priced, or they might have done a great job with tile. Whatever the case is, they can endorse them for what the person was actually good at doing.
As we dug into this, it's actually quite fascinating: there are professionals that are good, but at quite specific types of things. Maybe they specialize on million-dollar homes, or $200,000 homes. Maybe they're quite good at some things, but don't know what they're doing when it comes to putting up drywall. Again, that didn't exist before, in terms of figuring that out.
TechRepublic: Well, why hasn't this been done before? It's 2014. "Yelp for X" is almost a cliche in startup-land.
Ehrlichman: It really does come back to the data. This is one of the more fragmented industries that there is. There are millions of local, small business home service professionals out there. Not only is that industry fragmented, but the data about the projects that they have done has been even more fragmented. Many professionals don't maintain records in an electronic format. If they do, it's very messy. It's very unwieldy.
TechRepublic: What's changed today?
Ehrlichman: We used public data to get over the initial tipping point. Permit records and license records, nationwide, for 1.5 million professionals. When we launched in September 2013, we launched nationally. We used that kind of data early, to get a tipping point. We've done the hard work. There wasn't really anything magic about it. We had a team in place. We were very lucky to have raised a large seed round, frankly, when we didn't have anything.
Right out of the gate, they were working with the best professionals, in all the cities across the country. We found we could get these top professionals to come in and upload all their projects, and by doing so, use them as influencers to get all the other professionals to follow them, early on before we had a product that was launched.
On top of that, and this is something that I personally was going after, was hundreds of partners that have different slices of data about professionals. That really was what made this whole thing work: we created these exclusive and often perpetual partnerships with these companies that are out there, with different elements of data.
TechRepublic: What's in it for those partners?
Ehrlichman: A couple times, we licensed [their data]. Usually, it was free. We were able to bring the information in and create exposure [for the partners] through that information. Picture a big franchise, everyone from Orkin to others. The more information they bring in to the Porch platform, the more we know that, "hey, you worked for six neighbors of this particular customer," or "you worked on 100 homes in this neighborhood that are between $350,000 and $400,000 value and these types of sizes." The more information we have, the more confidence we can provide to the homeowner about that professional's experience. We've created an incentive for them to bring in information, because they get more business.
TechRepublic: Is this a next generation Yellow Pages? In years past, to look for someone in the trades, you might pick up the Yellow Pages and then ask your neighbors who they used. Is the analogy that on Porch, people are paying to get a bigger ad on a page?
Ehrlichman: I don't know. There are ways to build a really big business without abusing it, I guess I would say, from the monetization perspective, with a really ad-centric model.
TechRepublic: So, who pays you?
Ehrlichman: The professionals pay us. We have a number of products lined up that we haven't launched. Even if we only had a subscription product for professionals, the market opportunity in that alone is just massive. There's millions of these home service professionals. Every contractor, every plumber, every electrician, every painter, every house cleaner, every landscaper -- everything.
TechRepublic: What do they get in their subscription?
Ehrlichman: They get more business, from Porch, at the end of the day. We give them more exposure, so that we drive more to them.
TechRepublic: What does exposure mean, in this context? Is that an issue in providing recommendations, as some critics allege exists at Angie's List?
Ehrlichman: Exposure just means we're going to put you in front of more homeowners that are looking for your type of service. When they're searching for a professional, we're showing the options based on data. When homeowners are browsing, because they want to look at new decks that their neighbors have done and they want to figure out costs, we can provide the subscribing professionals access to those homeowners. We can also help those professionals get more insights, to know that these are the particular areas in the neighborhood where they are the most likely to be doing these projects. We can help them optimize their markets.
TechRepublic: Does that mean better targeted advertising and revenue from lead generation?
Ehrlichman: I think that's fair. We're not charging on a per-lead basis, like some of the other models are, but we can create exposure and more business for subscribers. At the end of the day, the biggest thing that these types of businesses need is more business, more of the right business -- not just flooding them -- that are the right types of customers for them, based upon what they've done in the past, and increased density. A lot of these people, like plumbers, electricians, and house cleaners, they spend a lot of their day right now driving around between jobs. If you can actually highlight them to neighbors and past customers, it will improve their density.
TechRepublic: In terms of the world of government data, how much was free and usable off the bat? How much did you have to pay for and clean up?
Ehrlichman: Right now, private data is a much bigger component of what we do than public data. We used public data early on, and it's important, there's no doubt about it, for permits, for licenses, for information on homes -- better data on home structure and size and value. There's lots of different public data available out there. We did not pay for it. The great thing about public data, generally, is that it's available, by definition. There's slices of it that have been organized that we paid for, certainly. License [data] is a good example: we had a team that did all the hard work to go to 47 states and be able to deal with all the different formats of licenses for all of those professionals, and be able to organize and rationalize and clean and master that data, across these states. Hard work.
TechRepublic: Do you have to go back and refresh that?
Ehrlichman: Every month. You have to get these feeds. I am a data geek, we have a team of data geeks. This is what we really like. This is kind of fun for us.
TechRepublic: A strategic asset?
Ehrlichman: Yes, and a sustainable one.
TechRepublic: What happens if someone else gets together another team of data geeks to mine public data?
Ehrlichman: That's exactly why we went after and locked up all of these exclusive partnerships for private data.
TechRepublic: That means that they'll give you this data but not somebody else?
I think a company that did this really well was LinkedIn: they understood really early that data was their asset, and focused on that, and they've been able to do things in the interface that are very different than anybody else can do. We're trying to replicate some of that in the home improvement industry.
TechRepublic: You've said you're able to refresh data feeds for 47 states. What have you found about periodicity and formats?
Ehrlichman: It's an incredibly hard data problem, because it's not standardized across states. It would have been beautiful for us before. Right now, I don't mind that it's messy, because we've done the hard work to clean it and process it for ourselves. Now, it becomes a bit of a competitive differentiator. Again, it was a lot of hard work to get all of these differently formatted data [cleaned], all of these different, complicated factors within the schemas that are being provided to us.
TechRepublic: Do you provide an API or bulk data feed of this cleaned up public data?
Ehrlichman: Not today. I think it's a really interesting question, though. In the future, as we get to this tipping point in this marketplace, and start to really build on that, it does create an opportunity to really further your competitive position. Facebook is a good example. They have their site, built off of their social graph, and they did start to expose that, and make APIs available to build on top of it. I think that was a genius move.
TechRepublic: But you couldn't export the social graph itself.
Ehrlichman: No, and it's smart. You have to be thoughtful about it, so that you continue to build your company and add value and everything else, but still create value for the broader ecosystem, so people can still use that information effectively. I do see a future where people will be able to build on top of information that we've gathered and cleaned, because it's valuable. We're not going to build everything. I think there is a path to make that available.
TechRepublic: You've formed a partnership with Lowe's, a home improvement giant. What does that mean?
Ehrlichman: We're rolling out to 139 stores. We'll be testing, iterating, and expanding further. [It means] signage all throughout the store, and kiosks throughout the store. All of the associates are trained with Porch, on their mobile devices. (Since this interview, Porch has rolled out to every Lowe's store in America.)
What happens today is that homeowners are in Lowe's stores, they're looking at products, and decide that they want someone else to do it, instead of them. They ask associates all the time for professional views. Aside from the very specific types of services that Lowe's provides and sells, they can't make them. They're using Porch to change that. They're really trying to retrim their business from being a home product company to a complete home improvement solution and partner. They're using Porch now: whenever a customer asks for a home improvement professional, an associate can go to the kiosk or use their phone to put in an address or location and do it right there.
TechRepublic: What other recommendations are used?
Ehrlichman: Endorsements is one. Seeing who your neighbors have used again and again is another.
TechRepublic: Is that self-reporting?
Ehrlichman: These come from contractor reports or are permit-related. When we get all this data from partners, we can line it up and see which professionals worked where.
TechRepublic: Regulatory data shows those relationships, building up what we could call a professional graph?
Ehrlichman: That's exactly right: homes they've worked on, what address, what cost, what day. That's shown by permit, licenses, and all the rest of the data from our various partners. We know what they've done. We know what types of jobs they've done and what they cost. We can highlight to homeowners what the average cost is for a type of project.
TechRepublic: What about consumer complaints? The home improvement industry has some shady characters in it.
Ehrlichman: Clearly, we will offer that. We have not launched negative reviews yet today.
TechRepublic: Do they exist elsewhere? Yelp, etc.?
Ehrlichman: There are certainly reviews. The challenge that we've found, what homeowners have found, is having that trust that the reviews are accurate. Generally, folks know that lots of the review sites are being gamed, where the business is reviewing themselves, their brother is reviewing them, their competitors are reviewing them negatively.
TechRepublic: The incentives are all there to do that...?
Ehrlichman: Exactly right, and there's no way to protect it. The word we use internally every day is "trust." We're trying to build a trusted network. The data allows us to do that, so we can have that check and balance between the professional and the homeowners. We know where the actual project happened. Since we've dealt with the professional, we'll be able to check that with the homeowner.
TechRepublic: So, do you validate the homeowner?
Ehrlichman: We want to do more work in our product around the verification of homeowners, that they actually live at that address, before we launch negative reviews. Once you start doing that, there is that incentive for competitors to blast other people. So, we want to make sure that as we launch that, we have a really strong understanding of verification of that home owner and where they actually live. So we would bring those two things to market together.
TechRepublic: Identity, security, and privacy trips up so many startups. To get to this next level, we have to understand who you are.
Ehrlichman: This is an interesting topic for us. We've drawn a couple of hard lines. When you come onto Porch, we never show the address of the home. First of all, when we bring project information in, we don't take the homeowner's name and their phone number. We don't want it. We use the address as a key, to get the type and the value of the home and know what that structure is. Even in the map experience, where you're looking at your neighborhood and the projects around you. We randomize within the neighborhood.
TechRepublic: At the block level?
Ehrlichman: We have a mapping of true neighborhoods that are out there. Where we don't have the understanding, we use a radius, an area. You can't come into Porch and see a specific home, but you can still see what's happening in your neighborhood, get a feel for who they've used.
The name Porch was actually chosen because of that: it's the public space of your private home. That's how we thought of it. It's that space where people are OK sharing.
TechRepublic: There's a lot of public data that has never been made searchable before, and then mapped, raising some potential ethical questions. Are you concerned?
Ehrlichman: Any building or structure in which a project happened, we randomize all of it. Imagine, in the gun example, if they had removed name and used an area, the controversy probably would have been non-existent. That's been a really important thing, because that was our very strong feeling, as we built small products and tested on users, that you're dealing with people's homes, and what they spent -- keeping that privacy was important.
TechRepublic: Is the major barrier to adding negative reviews having an identity infrastructure?
Ehrlichman: At the end of the day, what homeowners want is to hire the right professional. There's two ways to go about that: making sure you don't hire the bad one, or making sure you surface the best one. What we've started with out of the gate is making sure that they can find the best one for their home. If you know that your neighbors have used them multiple times, on a hundred projects on homes just like yours in your neighborhood, that creates a lot of confidence in people. We've seen on Porch, as we roll out into these Lowe's stores, people really like it and think it's cool.
TechRepublic: What would be the most important thing cities and states could do to improve consumer protection around the use of data or release of data?
Ehrlichman: Certainly: having a centralized understanding of who a person is that could be used for verification. What we lean on is Facebook, which is actually doing it well, in terms of real identity, but it's not perfect. If there was an API available that was pervasive, across everybody, that would be something that would be used by countless numbers of companies.