Foursquare Attribution continues to help customers industry-wide discover even more about their advertising campaigns. Get the rundown of how our product came to be, how it actually works and – most importantly – how it’s optimizing business strategies everywhere.
Foursquare recently launched Closed Loop Attribution, a new feature for our Attribution product that distinctly measures ad exposure against online and online-offline conversions. This first-of-its-kind feature offers a better understanding of how best to utilize ad spend, make optimizations in flight, and ultimately drive results.
Taking it one step back, we now want to highlight the methodology that powers not just our latest Closed Loop Attribution feature, but all of Foursquare Attribution. Our methodology is what started it all and has since led us to become the preferred choice by leading marketers to provide reliable, actionable insights over competitors in the industry.
Why we created Foursquare Attribution
Before diving in, it’s important to first establish the intent behind Foursquare’s Attribution product and the use case it aims to solve. Fundamentally, our objective is to enable causal statements about the effectiveness of advertising campaigns, allowing marketers to optimize their spend on media plans while reaching the intended target groups. This is done by connecting exposure and impression data that we receive with Foursquare’s panel of in-store foot traffic data (or online conversion data in the case of Closed Loop Attribution). For our clients, it’s important that we use quantitative tools to distinguish the targeting effect from the true causal effect of exposure; marketers are less concerned with seeing activity from individuals who were already likely to visit a business and more curious to understand the completely new incremental actions driven by the campaign.
Traditional Foursquare Attribution starts with our understanding of what visits are occurring in the real world. Foursquare’s perception of the universe is grounded in our proprietary Places of Interest (POI) data, where we map persistent device data to specific locations. This POI data is unique because our world map is based on actual locations – POIs – rather than geocoordinates. Bluetooth, Beacon, WiFi, compass, accelerometer, barometric sensor, and GPS data signals all inform a user’s journey in the real world. Coupled with stop detection signals, dwell time data, and deterministic human verifications, we can accurately predict when and where users’ visits occurred across a broader panel.
In order to provide Attribution metrics, Foursquare combines curated visits data with information about which users have been exposed to an advertiser’s media. We use equal fractional attribution to assign appropriate credit to users who were exposed to multiple ads served on different media channels. This means each impression that overlaps a visit or online conversion receives a fraction of the overall action. In turn, we’re then able to study the influence of a single impression against the intended action encouraged by the client.
Our approach to causal inference employs a two-stage hybrid to calculate lift. In the first stage, we apply regression adjustment to predict visits of exposed users in a hypothetical scenario absent of actual exposure based on unexposed users – this includes the creation of a synthetic control group based on a collection of static features. Then, in the second stage, we remove the residual positive bias from users’ historical visitation to the brand, ensuring that lift in the pre-exposure time frame is zero. This second stage follows a difference-in-differences methodology to adjust for users’ pre-existing rate of visitation to the brand prior to in-flight exposures.
To further ensure accuracy and stability of our Attribution metrics, Foursquare implements a set of projections and weights to correct for potential biases and potential skews within our data. In particular, the Census-weighting we do to scale Foursquare’s panel to be representative of the country’s population helps marketers better understand how their campaigns perform in the real world and the true impact they are seeing across their targeted audience in its totality. We also project visit volumes on the basis of match rates by platform and partners, accounting for the expected OS distribution across the broader impression delivery.
What sets us apart
Clients prefer and trust our methodology over competitors in the market largely because it accounts for exposed users’ historical visitation to a given brand. This differentiates us from competitors who provide metrics on net visitation to a physical store or an online website/app without controlling for loyalty customers; essentially, these metrics are not isolating the incremental value created by the media plan. Foursquare Attribution empowers marketers to understand how their media is driving net new actions as a result, which is a more compelling metric for agencies and brands alike.
Interested in learning more about Foursquare Attribution? Get started by requesting a demo here.