LSA17: Local Marketing Data That’s Solving Attribution
March 1, 2017 | Contributed by: Mohannad El-Barachi
Over the past 6 months the local marketing industry has seen a proliferation of advanced features pop up on listings across the board. Yelp recently released a chat feature allowing consumers to chat directly with a store owner from a listing. Starbucks launched their mobile ordering feature, allowing consumers to order their beverage from mobile but pick up in-store. Google recently launched “Lists”: inching the search giant closer to creating a seamlessly automated experience. Google is hoping that a future shopping scenario will look like this: John gets in his car and scrolls through his “list” on Google to choose a location he’s saved. His connected self-driving car uses directions via Google Maps to take him right to where the restaurant booked by through his open table reservation. The industry is getting quite close to making this a reality!
But where are we today in terms of linking a search to a sale (what the industry is currently calling “Attribution” or “Online-to-offline”)? Attribution seems to be the word on every local marketer’s lips. Many local solution providers claim they can do it, we’re all aiming for it, but it also means different things depending on what technology is at your disposal and what problems you’re aiming to solve in local. Online-to-offline attribution in local is the notion of being able to “close the loop” from online actions to offline purchases: In a perfect world, this would enable businesses to tie an online action (an impression on a listing, a click for directions) to a customer actually entering your location and eventually making a purchase. The imagined scenario is the type of “closed loop” attribution model local is aiming for, but we’re not quite there yet.
Below I lay out some different types of data that heavy-hitting players in local are currently combining to build attribution models.
1. Search engine analytics: Listings are now more of a connecting point between consumer and merchant, offering accessibility in real time. Each platform usually has their own form of analytics so you can see how many impressions and how many actionable clicks your listings are getting. Google is king when it comes to attribution it collects from listings. If you don’t currently have a Google Analytics account, get one now. Some tools (such as SweetIQ’s Local Marketing Hub) gathers analytics from multiple platforms so you can amalgamate this data and compare and contrast.
2. Offline in-store data: It’s crucial to find out how customers are discovering your store but just as important is gathering all information on them once they are actually inside. Examples include installing a counter to track the number of people walking through your stores, using beacons to track in-store paths they take, and tracking average store
3. Geo-targeted ad data: Companies that sell geo-targeted ads (like Google and Facebook) are creating attribution models that track foot-traffic and how it correlates to whether someone was exposed to your local ad or not.
4. Proximity data: Companies like Foursquare and Uber are able to provide real-world behavioral insights based on data they collect on locations users are visiting, and in Foursquare’s case: how much time they spend at each location, and how frequently they go there.
The caveat with both #3 and #4 is that many companies like Facebook can only collect data from user’s who have agreed to share. While this is also true for Foursquare, users have to agree to location sharing at all times to use the app so it’s more seamless. Even if the app is turned off, Foursquare is still collecting data in the background on every location a user is visiting. The bigger problem with Foursquare is accuracy. There is some discrepancy between users actually visiting a store, or simply walking past it or loitering outside for a minute. As you can see none of these models are perfect, they still include probabilistic elements. All attribution models are built on a combination of two, three, or even all four forms of these data. For example, Sweet IQ is using a combination of online and offline data sources to show lift in foot traffic as well as ROI derived from local.
So which data set is best for you to help you build your model? First off, understand what it is you are trying to learn more about. Do you want to track lift in foot traffic? Ad effectiveness? Having these questions in mind when you start will help you find data providers to get you the right analytics. If you’re attending LSA 17 this week and you’d like to learn more about attribution and the data we use at SweetIQ to build our attribution model, you’re in luck.