Dayparting Methodolgies

[miketpowell]

Alright I've been grappling with this question and can't think of a way to solve it perhaps someone here has some insight.

Now most offers will covert differently at different times of the day with display traffic and it's easy enough to track and optimize at that level. However I'm trying to figure out what causes this and what ways I could test to figure this out.

A. The people that are more likely to convert for a given offer are more likely to be online at a given time of day/week.

B. Any given person is simply more likely to convert for a given offer during certain times of the day/week.

So the question is: Does the time of the day effect how likely a given person will convert or the chances that they are the type of person that will convert?

I'm trying to figure out a way to test to figure out to which degree each contributes. This will be of course different for different types of offers with there being a covariance between the two factors as well. I just want to find a good way to test it with data but can't think of one.

For almost all CC offers I believe B is a much bigger factor. Based on the fact that the day parting for display on an offer doesn't seem to be that dependent on the website IMHO.

 

[Jakey]

If you use p202/t202 check this site out. It's pretty sweet and I've had good luck with what it suggests, as it is more in depth than the default tool in p202.

Tools202 - Day Parting

Jake

 

[aeisn]

Unless you are able to target A and B individually, which I don't see how you could, it really doesn't matter.

 

[eVandals]

So the question is: Does the time of the day effect how likely a given person will convert or the chances that they are the type of person that will convert?

The answer is yes. You need steady conversion rates over time from a consistent source of traffic so the constants are there.

 

[jameszol]

For pure research purposes and statistical significance, this suggestion is probably not valid at all...but what you might be able to correlate with this could help point you in the right direction.

First, to identify the audience in A - I would review my analytics and cherry pick the websites that I know or am at least 85% sure that my target audience who I feel are more likely to convert will be. I would validate my theory by reviewing these websites at compete.com and Google's Ad Planner. Then if the ad platform allows, I would set up a day part campaign targeting those sites only to measure Audience A's conversion behavior.

For audience B, I would do the opposite: same day part but exclude the sites where I believe audience A is more likely to be.

With this method, Audience B might not be your target market, but you might be able to draw some conclusions regarding "any" given person converting at a specific time of day vs Audience A which are technically your target audience.

 

[AffAholic]

boy you guys sure like to use words & talk a lot...

 

[miketpowell]

Unless you are able to target A and B individually, which I don't see how you could, it really doesn't matter.

Sometimes I do ask questions that don't matter. I usually don't spend the time to write it up and put it on a forum though. So yes I'm asking this question because answering it could increase my targeting capabilities.

The answer is yes. You need steady conversion rates over time from a consistent source of traffic so the constants are there.

Thanks but you must have misread the question because "yes" is not an answer to an A or B type question.

For pure research purposes and statistical significance, this suggestion is probably not valid at all...but what you might be able to correlate with this could help point you in the right direction.

First, to identify the audience in A - I would review my analytics and cherry pick the websites that I know or am at least 85% sure that my target audience who I feel are more likely to convert will be. I would validate my theory by reviewing these websites at compete.com and Google's Ad Planner. Then if the ad platform allows, I would set up a day part campaign targeting those sites only to measure Audience A's conversion behavior.

For audience B, I would do the opposite: same day part but exclude the sites where I believe audience A is more likely to be.

With this method, Audience B might not be your target market, but you might be able to draw some conclusions regarding "any" given person converting at a specific time of day vs Audience A which are technically your target audience.

Thanks! That's exactly what I was talking about and that suggestion has given me a good idea of how I can go about tracking this.

I already have all the data to examine what you suggested here and looking over it and the conversion rates of "targeted" vs. "general" and it continues to support that time of day simply changes a non-buyer into a buyer or vice a versa. The problem with this though is now it relies on know which sites are targeted and which aren't... which is easy enough to do in extreme cases but not with more subtle differences as easily.

 

[Volume10]

are you direct with the advertisers? They may have additional data that may help your targeting/dayparting as well. We get data from our incentive advertisers and a handful of our rebill advertisers that allow us to at minimum give us a better shot at picking when and where to run ads to begin with.

Data like gender and geo info mixed with day parting may not be something your traffic sources allow you to target or something you want to dig that deep into but it def is something that we use.

 

[smaxor]

Run the traffic until you have enough statistical evidence. Then look at a graph of EPC's for the time frame you ran. It should be pretty obvious where it works. Might want to split out weekends and weekdays if you want to get more granular.

 

[mailinatorm2]

I have a software that does all that

 

[miketpowell]

Run the traffic until you have enough statistical evidence. Then look at a graph of EPC's for the time frame you ran. It should be pretty obvious where it works. Might want to split out weekends and weekdays if you want to get more granular.

Oh I get much more granular than that smaxor I do it by source as well usually for one and look for "pay check patterns". For example one offer I run I have narrowed down the behavior patterns of the group that converts pretty nicely. They are more likely to be on site A in the morning and that's when site A converts and move onto site B at night and that's when site B converts better.

That's a good point Volume 10 the data from the advertisers should be able to help me if used correctly to better create a "buyer profile".