The use of digital data has been successful for countless brands, responsible for shifts in awareness and consideration. The use of online data in campaigns supposedly allows direct targeting of consumers, but just this month research from Nielsen stated that only 50% of ad impressions served in the UK aimed at targeting women actually reached women. So about as reliable as flipping a coin.
The results of the survey are troubling, but not surprising, thanks to low-target hit rates – and next in the firing line has been geo-data. Geo-targeting has long been the boast of many a digital media owner, and, when managed carefully, can be the keystone of digital campaigns. But using it successfully requires that available data is used in the right way.
The typical way geo-targeting is used is to “fence” a certain area and collect the device IDs of those within it, and then serve ads to those devices either at that time or at a later date. It could be anyone who has been near a brand’s out-of-home ad, visited one of their shops, or passed through a place associated with a certain activity (e.g. an airport, or a concert venue).
Its logic can’t be faulted – we are defined by the places we go; they indicate how we spend our time and our hard-earned money. However, as geo-targeting has become more widespread in digital media plans, people are starting to realise that without careful planning it is not all it seems.
The issue is the quality of the geo data available. When a mobile device user goes online and an ad call is registered, their device ID is recorded and is attributed a “location” signal. The latitude/longitude data relayed back to the advertiser can be up to 6 decimal places (d.p.) of accuracy – meaning within a metre – but the amount of 6 d.p. data being passed through the exchanges is very low.
This is all down to feasibility. GPS accuracy takes time to hone in and mobiles tend not to be fixed in one position for that long. If you think about when you open Google Maps or Uber, it places you within a circle with a diameter of about 20-30 metres at first, before narrowing down to between 5-10 metres.
Realistically, location data up to four decimal places is the best we can expect from exchanges – this is accurate to about 11 metres, which isn’t a problem for some purposes. However, we then run on to the problem of reliability.
Even if we get data that looks ‘accurate’, it isn’t necessarily reliable. The lat/long data we’re receiving may not actually indicate where the user is – some may be fraudulent, and some just incorrect. It is the responsibility of ourselves and media owners to pick up on patterns such as the lat/long being the wrong way round, locations in a line, centroids, and randomisation of location. All of these can be picked up by algorithms, but not all media owners and planners are aware of the significance of these issues or are as conscientious about dealing with them.
Geo data offers a highly valuable indicator of behaviour – so we shouldn’t give up just yet. Problems can be avoided by remaining diligent, and by working with partners known to be trying to combat ad fraud and mistaken co-ordinates, and only using location data of a certain quality.