Today’s marketers can pinpoint precisely where customer activity occurs based on data as simple as a hashtag, a location mention or a photo.
Marketers also have a multitude of sources to mine to understand where that customer activity occurs. Take images, for example. The Atlantic referenced a 2014 Kleiner Perkins report on internet trends — the well-known annual Mary Meeker report — that estimated that people upload 1.8 billion photos a day — 657 billion photos per year. Since 2014 when the research was published, beacons, smart devices and other sensors have come online to add meaningful metadata to those images.
Location, Location, Location
It’s no surprise that marketers want to analyze location to decide where to invest their budget. However, to accomplish that analysis requires advanced modeling techniques using spatial statistics — sensor-generated data and images that identify a geographic location. Marketers are increasingly turning to these techniques to support local SEO and SEM campaigns.
Spatial statistics have typically been used to create topology and interactive maps for social issues (e.g., highlighting gentrification trends in neighborhoods) and for science studies (e.g., how climate change impacts geography). The use of spatial statistics for social science work has created open source opportunities for marketers to map customer interest and activity by coordinates. One open source approach involves importing data into an R program and creating models based on it, such as regression against search data or mapping out data.
Images add another layer to help understand customer activity and customer sentiment. Most smartphone cameras — with the proper permission enabled — can take pictures that capture longitude and latitude and add a timestamp of when the image was created. Called geotagging — the process of adding location metadata — this technique creates information that is store in the Exchangeable Image File (Exif) format, which can then be imported an application for data modeling.
How Marketers Use Geotagging
Marketers can use this metadata when building models in R, Python or other languages. The data can be analyzed in a regression model, for example, to determine if there is a statistically valid relationship between where photos are shared and other data such as a hashtag use, beacon activity or website conversions linked to geographic locations.
R programming, for example, uses a library called ExifR that extracts Exif information and places it in a data frame. A data frame is a container for a matrix table, much like an array in an object-oriented programming language. You can put any type of data into a data frame if the data has the same type within the columns.
Marketers wading into R programming waters have several sources to help them learn how to use location in their predictive analytics and machine learning models, even if translating the ideas to code seems cryptic at first. For example, many analysts have started to share ideas and explanations on R-blogger, a popular site used by data scientists. Those with more serious developer chops will find ideas on GitHub.
Analyzing geotags is part of a larger local SEO strategy. Marketers are starting to see how images relate to activity at a retail location and how a slew of nascent tactics can augment analytics solutions, such as adding location information to images before uploading to social platforms, which allows algorithms to use that information against queries, to assess if specific images should be associated with results. In fact, Mary Meeker noted in her most recent report that savvy retailers and crafty brands are “finding ways to make images, video, data, algorithms, and voice” work together.
Investigating image metadata is a worthwhile part of your marketing strategy. The analysis can help you determine how seemingly disconnected activity and image-sharing at a location relates to leads and sales.
Learn how you can join our contributor community.