IKEA

For some retailers, in-store foot traffic is critically important to enable brand experience and feet on the ground are simply more valuable than lines in a database. But there’s only ever so much available budget. For a long time, it felt that these retailers had to choose: online or offline?

Along comes Facebook. Social media may only have been widespread for the last few years personal touch has allowed retailers to invest heavily in activity that is simultaneously branding and DR.

Furthermore, there’s always been a logical relationship between social media and brand awareness and loyalty, which surely must increase visits and sales. We’ve had plenty of studies suggest as much, and gradually social has become an accepted, part of the marketing mix.

iProspect partnered with Facebook, EE and Vizeum (our Dentsu Aegis Network partner agency) to trial highly experimental technology, combining hyperlocal Facebook ad targeting with all-new tracking technology from EE to test the audacious hypothesis that social advertising as an isolate variable could be proven to drive in-store foot traffic.   

We chose IKEA Cardiff. Sale activity had been planned to coincide with the New Year and so we knew we’d have something to talk about to the local populace. Working with a relatively limited budget, our priority was to get the highest possible reach within favourable geographic parameters.

A geofenced area was established around the IKEA Cardiff store, connecting together multiple network masts and using triangulation to trace a precise outline within which we would measure people. Being able to track users from social to sale using mobile geofencing was in itself a huge step forward, a landmark moment for social media and retail alike.

Two separate groups were created: exposed and control. These were based on the phone number of the handset: numbers ending in odds saw the adverts; evens did not. The brilliance of this variable is that it is completely random: there is no pattern to number assignation whatsoever. This measurement was correspondingly 

integrated with the ads from Facebook to double-verify exposure. Everyone that was targeted/tracked had the same environment, so there was no artificial uplift caused by variables – anything that applied as a variable applied to everyone. This was conducted at sufficient scale, with some 173,000 tracked people in each group. We also created bubbles of hyperlocal targeting in surrounding areas: around a sister store in Bristol, the central and metro areas of Cardiff, Newport, Caerphily, Pontypridd and Bridgend. Lastly, to prevent technical cannibalisation we removed Wales in our on-going UK-wide branding campaign.

During the period 27th December – 11th January we served three Facebook posts by Isobar (our Dentsu Aegis Network partner and IKEA’s social lead) into the News Feeds of our exposed group whilst the control group saw nothing. We removed false positives from our dataset, leaving only people who we knew for sure had seen our advert and were definitely in-store.