Case Study | Kidly | Microtargeting

Micro-targeting by Household Income

Kidly | eCommerce
Read More




Return on Ad Spend


Kidly offers parents a curated collection of “All Things Kids”. We wanted to show our ads to parents who were browsing on mobile while at home in the evenings as we’d noticed a lot of traffic around tea time. The question was how could we focus on the homes that were likely to have our customers? This was a few months before household income targeting came to the UK. Our answer was to take a ton of data from the Office of National Statistic and build a tool that enabled us to focus on neighbourhoods with greater precision than post code targeting.


We know that location strongly correlates with key economic indicators such as income as well as the likelihood of having a child so we wanted to find and use data to select locations to target. We hypothesised that geographical areas with a higher net disposable income would correlate with higher conversion rates and a higher average order value.

We wanted to identify and target only a few thousand households per location otherwise the average order value of a larger area, e.g. city, could be warped by the variation of household incomes within it. Our aim was to collect 3rd party data that would allow us to identify and target specific locations according to the average net disposable income within those areas.


Instead of paying for clicks, we started by collecting free accessible data from the Office of National Statistics (ONS). This data identifies 7,202 locations within the UK known as Middle Layer Super Output Areas (MSOA). These are more precise than city level targeting and include 2,000 to 6,000 households.

Then we estimated net disposable income for each MSOA according to the ONS definition of the term and also associated the MSOA with the likelihood of a household including a child.

Next, we processed the MSOA location data so that we could accurately target locations in AdWords. To do this we calculated the centroid for every single MSOA and generated longitude and latitude coordinates for this point.


After 9 months of data collection, we came to strong conclusions that helped us not only set expectations for Kindly, but it also helped us implement our newly developed Micro-targeting methodology on new clients.

1. Conversion Rate (CR) does not correlate with the net disposable income of an area (r=0.014). In simple terms: Rich people are not more likely to make a purchase.

2. Average Order Value (AoV) does not correlate with net disposable income (r= -0.1).

Conclusion: We should not target any locations proportionally to net disposable income.

However, using this data which is not available to us in AdWords, we discovered that…

1. Some MSOAS areas had up to 53% higher conversion rate.

2. Targeting these higher converting areas and excluding the low CR MSOAS dramatically increased the Return Of Advertising Spend (ROAS).

Therefore, in the last 3 months of 2017, we increased ROAS by 25%.


Via del Portone, 7
55045 - Pietrasanta (LU)