I teach search engine marketing at a local university, and I love hearing the perspective of young people on trends in digital marketing. One question that has come up in more than one class is, “do people really click on search ads?” When we discuss this, there has been general consensus among my students that younger people don’t click on ads.
When this question came up again recently, one student shared this article to help make her case: Younger users rely on snippets and knowledge panel, often don’t click, survey says
The article is interesting, and generally supports my students’ POV, but it is also based on survey data. As the 2016 US election tells us, how people answer surveys doesn’t always match what they actually do. To investigate further, I decided to do a little quantitative research. I analyzed search ad CTRs by age for 50 different advertisers. I then segmented the data into age ranges to see how CTRs compare.
Here are the average click through rates I found by age range:
| Age Range | CTR |
|---|---|
| 18-24 | 7.64% |
| 25-34 | 7.38% |
| 35-44 | 7.85% |
| 45-54 | 8.00% |
| 55-64 | 8.18% |
| 65+ | 8.05% |
| unknown | 6.97% |
The group with the lowest click-through rate, 25-34, was 0.8% less likely to click on an ad than the 55-64’s, who had the highest click-through rate. On the one hand, that’s a meaningful difference. If I’m running a search campaign, a 0.8% increase in CTR is a big deal. But on the other hand, a 7.38% click through rate is far from nothing. In the world of advertising, getting 7.38% of viewers to engage with your ad is a fantastic result.
The table above compares mean click through rates, but I also thought it would be interesting to compare the spread of the data. Maybe certain types of search ads get consistent CTRs, regardless of age, but others are more variable?
To investigate, I constructed this box plot, with the mean values placed as labels on the boxes:
Looked at this way, there is surprising consistency among the spread of data for the three younger age groups, and the three older age groups all have more spread, and a top quartile boundary close to 10.5%. This indicates that some businesses do see more variability than others.
In conclusion: yes, younger people do click on search ads, but it is also true that they are less likely to click on an ad than an older person.
other considerations
- How often do people actually know that they are clicking on an ad? Perhaps younger people are more likely to recognize ads, and therefore less likely to click on them.
- Are younger people less likely to click on ads simply because they purchase less overall, and ads are mostly commercial in nature? As purchasing power grows, it stands to reason that we research buying decisions more. Anecdotally, I had one pair of running shoes that carried me well over a thousand miles in college. Now, in my early 50’s, my kids mock me for my closet full of shoes.
- My data doesn’t tell us anything about the “why”. Young people are bound to perceive search ads differently than older people. I am in the 45-54 age range, and I find search ads as much less invasive than the ads I grew up with. A young person is likely to have a different perspective, but we can’t infer from this data that the lower CTRs correspond to differences in attitude. We would need to perform a controlled test to do that.
methodology
This analysis is based on data for 50 advertisers in a variety of industries, all based in the US. The sample size for each advertiser was > 30,000 impressions and the date range for each was Jan 1 – March 31, 2019. Only search ad impressions on the Google Network were counted. The overall mean for each age range was calculated as a simple (non-weighted) mean of all advertisers. The box plot shows the distribution of mean values for each advertiser in the sample. The margin of error for the overall mean values is 0.03%
Nico Brooks is a fanatically analytical principal at Two Octobers, and does occasionally click on search ads. When he’s not doing that, he loves helping clients find meaningful trends in data.
