How Attitudes Towards Ad-Targeting Vary by Time and Country

In the course of a recent project we had cause to measure the number of our geo-data panelists that had enabled Apple’s ‘Limit Ad-tracking’ feature, or Google’s equivalent – ‘Opt out of Ads Personalisation’.

Enabling this feature means that the Mobile Advertising ID (MOAD) is not visible to software vendors and app publishers, and that any prior behavioural insights learned as to that user’s affinities or preferences can no longer be associated with that device in order to personalise ads and content.

The state of this feature is by no means the only signal from users that they feel sensitivity towards data privacy, but given that those users are required to dig through their settings menu, find and set this explicitly, it’s a pretty good indication of how they’re minded.

Privacy this year continues to be a hot topic, post GDPR in Europe, with the launch of CCPA imminent and Google’s recent move to consign the cookie to history (the cookie being the web equivalent of the MOAD), so this is a timely conversation to have.

In this simple analysis, we calculated the proportion of our panel base that had chosen to limit ad tracking on a per-country basis – the results of which you can see on the map below. What’s interesting is that for a sample month in 2019 (November), the mean opt-out percentage globally was 9.8%.

Of note, Ireland (29.82%), the US (25.97%), Canada (21.37%), New Zealand (22.52%), the UK (22.16%), Australia (22.96%) and countries in Scandinavia show the highest proportion of users choosing not to share their MOAD. What do these countries have in common? Well, they are all among the most developed nations. Causation or correlation? Well, rather than speculating we can test this using established metrics that describe countries in these terms.

In this instance we compared the opt-out rate per country with three figures – GDP, GPD Per Capita (PPP), and the UN’s Human Development Index (HDI), which takes into account things like education levels, access to services and other non-wealth related attributes.

A simple Pearson correlation using the 2019 data reveals that GDP in itself is an unsuitable value as it is liable to be skewed by uneven economic distributions, ie. a country where 1% of the population accounts for 99% of the economy.

+-------------------------------+-------------------+
|            Dataset            | Correlation Value |
+-------------------------------+-------------------+
| Gross Domestic Product (GDP)  |       0.065415174 |
| GDP Per Capita (PPP)          |       0.358465392 |
| Human Development Index (HDI) |       0.380026111 |
+-------------------------------+-------------------+

Perhaps for this reason, GDP Per Capita (PPP) provides a much stronger relationship value – but if as we might suspect this opt-out rate is in some way related to different levels of education and media consumption then the HDI should line up nicely. As you can see from the table above, sure enough it does – where at 0.38 it’s the best-correlated value.

One other interesting way to examine the data is to observe whether this rate has changed over time. Limiting our analysis here to the countries picked out earlier, and grouping the opt-out proportion into quarters over the past two years of dataset history (2018 to present), we find the following:

In the most pronounced case we observe swing values of up to 20% in an annual period, and the regional coordination among some of these outputs might lead one to speculate that regional events (ie. the introduction of legislation or increased coverage) are able to influence perception and indeed action. What’s also interesting however is how quickly these trends revert to their original positions in the months following.

What we have learned from this is that despite this opt-out facility having been available to consumers for some years now, only 10% globally have chosen to take steps to withhold their MOAD. In the context of all we hear about privacy, does this figure surprise you? Well, it probably depends on where you are.