Advanced Machine Learning, Data Mining, and Online Advertising Services
In Feb 2015, we ran a display advertising campaign where our goal was to drive traffic to a landing page with a clear CTA (call to action) message. In this post, we describe how we executed the campaign and more importantly how we measured the campaign effectiveness.
The goal of our campaign was to drive relevant traffic to a landing page. We had a clear message on our landing page where visitors could read a whitepaper on AD Tech space after they took the CTA. Our CTA was to fill out a short contact form and after submitting that form visitor could view and read the article. The text of ad is shown below:Using Advanced Machine Learning Algorithms to Buy/Serve B2C and B2B Display Ads at Large Scale
See link below to find out how the landing page and CTA looks like:
Using Online Learning Algorithms for Computational Advertising
We served our ad post on a very specific website tailored to the audience who we wanted to target and measure their engagement with the ad and our CTA.
Here, we describe how we measure the effectiveness of the campaign.
The most obvious metric we should measure to find out how effective our campaign ran is to measure how many people took the CTA and converted. Our database conversion table shows that we had a Conversion Rate of 3.35% which is really good. This high conversion rate can be contributed to the right message and clear CTA and targeting the right audience.
Next, we wanted to measure the impact of campaign beyond just simple conversion rate. One important thing is that you drive this traffic to your website and you want to measure how much these viistors engaged with your contents after clicking on your ad. Did they bounce right away? Did they consume other contents? What was the distribution of their session duration etc? So, basically we want to measure and get insights into website visitors behavior after they came to your website.
First KPI we wanted to measure was how much traffic we got after running the campaign. For this, we took our website data for a period of 5 days after starting the campaign and compared that with 5 days interval before the campaign as the baseline. This way we could compute traffic visits lift of post campaign with respect to pre campaign.
Reader should note that there are several website analytics platformis such as Google Analytics that provide the measurement of these KPIs. The whole point of this exercise to go a bit deeper into those metrics and how we can compute them and provide some insights.
visits_lift = post_camp_visits_no/pre_camp_visits_no
Our measurement shows that overall lift in website traffic was 2.0. We can compute visits lift for a given page. So, would be important to measure the lift we got for the number of visits of the landing page. We found that the landing page viists lift was 5.28 after campaign which is significant.
It's also measure the effectivity of the campaign by calculating the bounce rate. Basically we want to measure how many visitors left the website after seeing the first page. So, one can simply calculates the lift in bounce rate post-campaign with respect to pre-campaign as below:
bounce_rate_lift = post_camp_bounce_rate/pre_camp_bounce_rate
Our calculation shows that post campaign our bounce rate droped by a a factor of 0.92. In words, our bounce rate decreased by 8%.
Last not the least we wanted to also measure the lift in visitors engagement with the website content after starting the campaign and see how much that impacted people to come and spend more time on the website and read articles. For that, we just needed to track visitor sessions and compute the overal time spent by visiotrs and mesure the lift as follows:
session_duration_lift = post_camp_session_duration/pre_camp_session_duration
Our mesurement for the amount of time people spent on the website and engage with contetnts after running the campaign shows a lift of 2.14.
To conclude this post, it's important to repeat again when you want to run a display advertising campaign you should make sure that you have a clear goal and strategy for your the campaign. It's also important to think about the KPIs and how you want to measure the success of your campaign ahead of time. That way you can iterate and optimize your campaign strategy in order to deliver the results your sale/marketing teams expect. And as we have shown above measuring the effectivity of a campaign can go way beyond measuring impressions, clicks, and even conversions.