Machine learning is a hot buzz word in software at the moment. It has had other such moments in recent memory along with its more ambitious sibling, artificial intelligence. But these technologies have seemed remote, academic and monied. More Wall Street than Main Street. Locu, a startup that grew out of MIT’s Sloan and CSAIL schools and was recently acquired by GoDaddy, is aiming to change that.
Locu started by powering up-to-the-minute menu listings for hyper-local sites like Open Table, CitySearch and TimeOut Chicago. Not much to it, right? Wrong! The tricky part about providing a service to distribute content generated by small-business owners is getting them to actually generate the content. There are restaurant groups and chains, of course, but most restaurants are fairly self-contained operations that don’t have the luxury (or need, really) of a dedicated marketing or IT department. So how do you get overworked restauranteurs to input their menus and keep them updated?
The answer to this question is where the machine learning comes in. The obvious answer to how you get anyone to do anything is simple: you make it as easy as possible. And it can be even easier than that. You can do it for them. This is the problem space that CEO and founder Rene Reinsberg led his team through as they built out this functionality. Making the content acquisition part as easy as possible turned out to mean two distinct workflows.
The best way to get high-quality, structured content from users is to set up very efficient input mechanisms that guide the user through the process and render the results in a format that can be used in a wide variety of display media. The problem is that as well-designed as one can make such a process and as beautifully as one can format the results, the user still has to sit down and enter in all of the content. So although structured user input makes it easy and efficient for the back end of the system, the burden placed on the the front end (the user) makes compliance problematic.
The best scenario for users, Reinsberg discovered, is to be able to accept their menus in whatever format they have them, a word doc, a pdf, a web page or even a scanned image. But how do you handle all of these forms of unstructured content on the back end without resorting to expensive and error-prone human labor? The answer, for Locu, has been a combination of data extraction, machine learning and crowd sourcing. This process, which Locu calls “input magic,” returns structured document for approval within 24 hours (and usually, Reinsberg tells me, in as little as an hour or two.) And it turns out that the work of extracting data from supplied documents is very similar to the process of crawling the web itself for data.
So in the process of solving a problem for its customers, Locu has developed a set of robust tools for extracting and structuring menu and other types of data from the web. This means that in addition to being able to help restaurants promote their own menus and daily specials, Locu has developed the capacity to give its customers comparative data about competing restaurants—even those that are not Locu customers. The image below shows what an API call on “pizza” returns in San Francisco in terms of a histogram of relative prices as well as a geographical distribution of restaurants color-coded by price.
While building out these capabilities, Locu identified a weakness in the internet strategies of most small businesses. Consumers now use multiple channels to find restaurants and other local businesses, but making sure your business is listed in all of these places online and keeping track of and updating your listing information is a time-consuming task that usually goes undone. Most of what SEO consultants do for their clients is just to help them set up this basic “plumbing” so the real marketing work can begin. Working on the “web scraping” problem gave the team a lot of insight about how search engines (and hence consumers) find things. Locu has turned these insights into the capacity to give small businesses an easy way to make sure that their information is current and represented on all relevant platforms.
The diagram below shows how locu takes a business’ location information, hours of operation, menus and other services and product photos and distributes them seamlessly to CitySearch, Facebook, Foursquare, OpenTable, TripAdvisor, Yelp, the Yellow Pages and more. This in itself would just be simple plumbing but for the fact that when you first sign up for an account and input your business name and location, Locu automatically crawls the web and shows you all of the information that is currently out there on all of these services and gives you a streamlined way to make corrections and additions to them all simultaneously from a single set of input fields.
The result of all of this work on discoverability for small businesses caught the attention of GoDaddy, the world’s leading web hosting company. Last August, it acquired Locu for a reported $70 million and named Reinsberg VP/General Manager of Discovery Marketing Products. Today, Locu announced that it is making a fully featured version of the service available both through GoDaddy for its customers and directly through Locu for others. Get Found, as the product is called, offers basic discovery services at $4.49 a month. An “essential” tier, for $17.99 a month adds the more active components of pushing your company information out to search engines and listing services as well as the menu (or product list) publishing features and instant updates of all content. The “ultimate” tier, for $26.99 a month, adds the insight comparative data feature as well as the ability to publish content and promotions through social media.
GoDaddy has long been known for cheap web hosting, helpful customer service and very aggressive up-selling of second-rate add-on services. The company’s new management seems intent on changing that last perception. It is still clearly up-selling, but in the case of Locu, the add-on seems first-rate.
Machine learning and the hybrid human/machine workflows that emerge as optimal solutions to many data-centric problems are a key differentiator between mere “plumbing” and what I like to call “smart plumbing.” There is a lot of value that can be delivered just through creating a means to manage data in complex systems. But unless the means of creating these systems is generated by the data itself—in other words through machine learning—they will quickly ossify and be supplanted by the next advance in plumbing. One of the biggest technology risks for small businesses is to get locked in to platforms that do not continue to evolve. What may initially provide a competitive advantage can easily turn to a competitive disadvantage if the chose solution does not keep up.
What gives me confidence about solutions like Locu is that they evolve from the ground up though an understanding of the dat they are trying to process and the customers they are trying to help. Locu started with restaurants but has discovered that, with minor variations by vertical, most location-based small businesses will benefit from similar capabilities. When consumers are looking for something they want to know basic information: where is it, what is it, how much does it cost and when is it available. Answering these basic questions increases the odds that a customer will choose you, especially if information about your competitors is incomplete or inconsistent.
This kind of basic information infrastructure is just the starting point for discovery marketing, but as Reinsberg discovered (and GoDaddy has concurred), most small businesses don’t even have this in place. In Locu’s version of the up-sell, on top of this foundation you can add the capability to actively publish and update your content. Especially important here are menus (for restaurants) and product/service price lists for other businesses. As vulnerable as this can make a business feel because it means that it is easy for competitors to be aware of your prices, this kind of transparency and access to information is exactly what consumers want.
Of course, once a business starts actively promoting themselves based on price and product offerings, the value of comparative data become obvious. In this way, Locu’s tiers of service follow a very natural progression of how a small business can use the management of its data to drive growth. For GoDaddy itself, acquiring all of Locu’s MIT smarts is indeed a very smart move for driving its own growth and for getting tech-savvy customers to reassess the quality of its add-on services.
– – – – – – – – – – – – – – – – – – – –