At Rockerbox, we’ve learned that focusing on click-through rates will lead to poor business outcomes in display prospecting campaigns. This is because there is little to no correlation between click-through rates and conversion rates.
For context, the above plot was generated from campaign data that we’ve seen across our direct response advertisers with more than 50,000 impressions. The campaigns themselves are generated from the insights provided by Hindsight — Rockerbox’s analytic platform. By using Hindsight, we’re able to build our campaign strategies around Recency Marketing (marketing against a user’s most recent browsing behavior). Since we are trying to get as many conversions (be it signups, purchases or anything else) for our customers, we look at all factors that could signal a positive conversion rate for a specific browsing pattern.
What we’ve found from this analysis is that a higher click-through rate doesn’t imply a higher conversion rate. In fact, we don’t find any relation between clicks and conversions. Even more, when we add a smoothed best-fit line we see a correlation that looks to be zero or even slightly negative.
Why don’t clicks lead to conversions?
For one thing, most users don’t follow the path of clicking on an ad to arrive on an advertiser’s site. Instead, they either go straight to the site (think opening a new tab and typing in the domain) or simply go to Google to search the brand’s name.
Furthermore, if you’re not strategic in terms of avoiding bot traffic – either through internal controls or external fraud detection vendors – a large proportion of your clicks will be generated by bots that travel around the web. You see this by looking at the chart below. Here shows that when we don’t actively work to avoid bot traffic in a controlled test, less than 50% of clicks actually result in the user arriving on the advertiser’s site.
This doesn’t mean that all clicks are invalid; there are still many users that click and convert. However, purely optimizing towards clicks won’t optimize towards conversions. Plus, it is an almost guaranteed path for optimizing towards fraud. Because of this, we’re very skeptical of clicks at Rockerbox unless they’re 1) validated by a conversion event or 2) generated after taking all available precautions to avoid fraudulent traffic.
So maybe our title isn’t entirely correct. We do look at CTR’s…but with many grains of salt.