Ad Effectiveness & Time of Day

Original post found at blog.codefund.app

CodeFund has historically had a difficult time holding on to advertisers. We want to change this. I want to share some data with you. The names have been changed, but the data is real.

Here is a chart that shows our paid impressions by advertisers for Wednesday, Jan 9, 2019:

You can see that we have five paying advertisers. This graph tells us a story. Note that the bottom timestamp is UTC. At the top of the chart, you will find the times in PDT (San Francisco time).

Here’s what I learned from this chart (for PDT timezone):

• Advertiser A spends their full daily budget around 9 am
• Advertiser B spends all of their funding between 5 pm and 7 pm
• Advertiser C spends very little; however, the spending mostly happens between noon and 4 pm
• Advertiser D is spending their maximum budget every day before noon
• Advertiser E is spending their full budget between 5 pm and midnight

Many advertisers have told me that ad clicks that come from the US and Canada provide more actual conversion and sales than any other country. Based on this information, I have added a dotted red line across the graph to indicate (what I believe) the likelihood of the ads converting into a converted customer:

Now, looking at the chart, I can see that we are not providing the quality to our advertisers as we could be. We have added a few knobs and levers to fix already.

Here’s a screenshot of a portion of the campaign form:

Here you can see that there are several options for campaign targeting:

• Countries (soon to be US states or regions)
• Programming languages and topics
• Exclude programming languages or topics
• Only display ads between 5 am to midnight visitor time
• Only display ads on weekdays

However, this is not good enough to really ensure that the budget is spent during the right hours.

I strongly believe that we can really make an impact on the overall quality of ads if we switch our pricing model to be based on the visitor location.

By doing this, we can still continue to display ads during the lower converting hours. But in this case, the budget will not be spent by 9 am PDT. I believe there’s a strong demand for traffic in the US and Canada between 9 am to 5 pm PDT.

We also discovered during this deep dive into data that there’s a bug in our system which is allowing paid ads to mix with fallback ads during the day. This could be the reason that Advertiser C is not spending their daily limits.

I’m looking forward to measuring conversion improvements for our advertisers with the upcoming enhancements and see if we can provide more value to our publishers with happier advertisers.