Dayparting initial test (7)
I'm wondering how you guys setup your campaigns for initial test about the dayparting. Do you run the campaign for all day, or just part of the day?
There is an option in PopAds where you can "throttle" the budget so you avoid burn the budget in a few minutes.
I've tested 2000 impresions per hour, for 15 hours splitted in: morning, afternoon, night. I'm not sure if it is the best approach 
Or do you think it is better to test something like, 5000 impressions per hour for example, between 6:00 pm to 11:59 pm?
This is a pretty general guideline, but typically the best converting hours will be during the day and evenings - particularly in the evenings after work hours. (I've read studies on this and have observed the same in my stats.)
So, if traffic volume isn't an issue, you can choose to daypart from the beginning - for example from 8am-12am, or even just from 5pm-12am if you have lots of traffic volume and wouldn't mind restricting traffic for better ROI (potentially).
That way you can test for cheaper. Just don't forget to retest all hours once you're green.
Using throttle would be another approach. That way you're not making any assumptions (e.g. that 5pm-12am are the best-converting hours). But this approach may cost you more. If I have so much traffic that I need to throttle, I'd rather do my offer/lander testing during 5pm-12am, get green first, then retest other hours - would probably be cheaper that way. That would be my guess. I haven't actually split-tested this so don't know for sure. 
Amy
Thank you Amy for the useful information
I was thinking what is better, if a few impressions during 16 hours with throttling (maybe we don't want traffic where most people are sleeping) or more traffic for example from 6pm - 12am, but the first approach is slow if you don't send enough traffic, as you say, should be more expensive because we are testing more hours, and those hours should be tested as it should be, sending the right amount of traffic.
In your opinion, how many impressions per hour/lander/offer should be good for an initial test? At least something ideal because maybe some countries can't send too much traffic per hour, but would like to know a good estimation, sometimes I feel I'm testing too slow and it is desperate! and if I send too much traffic, maybe I'm spending and sending too much traffic per hour and the budget is burned. I think taking the data for just 2 hours isn't enough for statistical significance.
Actually - if all you're worried about is statistical significance, I would say you don't need to.
If you're testing offers and/or landers and just want to find the winner(s), the fluctuations that will occur over time, should affect all landers and offers in a similar way. So what I would suggest, is to just run throughout the day if you want to collect data faster, and to check periodically and cut landers/offers as they reach stat sig. If you're wanting to test at lower cost, then just run traffic from 5pm-12am local time until you find a good offer+lander and maybe also the best-converting traffic segments (OS/Carrier etc.), and then retest other hours.
Amy
Hi jurtzi,
I prefer running campaigns for 24 hours. In most cases, there’s no relevant difference. It’s obvious that - during the day – you’ve got more impressions, but it doesn’t mean the campaign is more profitable. Moreover, bear in mind that certain numbers of conversions will be delayed, which influences your hour statistics. There were only a few unique cases where I saw a noticeable difference from day to night. Nonetheless – after proper optimization - all hours became profitable.
Regarding your question about optimal volume for testing: there are no straight, by-the-book rules. It depends on both the segment’s payout and conversion rate. I suggest you look into each specific campaign.
If the campaign’s been running for 1-3 days and you can’t optimize it, you should definitely increase the volume. As you can see, I don't do day parting so often. However, you should bear the following two things in mind in case you wanna do it:
1 - Decide the hours/days to target based on relevant data;
2 - Pick them based on profit/performance and not only on volume.
Remember: the trick is to find the balance between gathering enough stats for optimization and your willingness to optimize a campaign before it’ll makes you spend money needlessly.
Hope I’ve been able to help!
Cheers!
Thank you for your answers, I'm more clear and now it is time to move on, collect data and keep calm
Consider also different day from week. Different weekdays performs different but you have to gather more data
Home > General > Affiliate Marketing Forum