Home > Paid Traffic Sources > Native

Blacklisting Sites on Taboola - Statistical Significance (3)


07-29-2016 05:46 PM #1 smalcolm (AMC Alumnus)
Blacklisting Sites on Taboola - Statistical Significance

Hey guys -

How do you guys determine when to block a placement on Taboola? I know blacklisting is a big part of the game, but I can't decide when to block a site. For example, see the screenshot below. Do you cut the first placement or let it run a little longer?

Click image for larger version. 

Name:	Screen Shot 2016-07-29 at 10.44.42 AM.png 
Views:	124 
Size:	130.1 KB 
ID:	12235

Thanks,
Adam


07-29-2016 05:50 PM #2 bandito (Member)

Check this out - http://stmforum.com/forum/showthread...Banners-Part-2


07-29-2016 09:24 PM #3 vortex (Senior Moderator)

Quote Originally Posted by smalcolm View Post
Hey guys -

How do you guys determine when to block a placement on Taboola? I know blacklisting is a big part of the game, but I can't decide when to block a site. For example, see the screenshot below. Do you cut the first placement or let it run a little longer?

Click image for larger version. 

Name:	Screen Shot 2016-07-29 at 10.44.42 AM.png 
Views:	124 
Size:	130.1 KB 
ID:	12235

Thanks,
Adam
Thanks @bandito for suggesting my thread on placement-cutting!

If you want to cut placement accurately - that spreadsheet will do that for you.

If you can sacrifice some accuracy for efficiency - you can use rules of thumb, such as IF placement is in loss by X times payout, then cut the placement.

As for what the "X" should be set to - that would depend on how aggressively you're wanting to cut. I personally find cutting when a placement is in loss by 2x payout generally works pretty well. (Going by this rule of thumb, yes, you can cut that first placement.)

You can cut less aggressively if your camp overall isn't in great loss, and/or your current test budget will allow you to spend more money to make sure a placement is truly bad before cutting it, and/or you don't have a lot of traffic volume and don't want to limit that further unless you have to.

Another approach would be to TEMPORARILY cut placements very aggressively - for example anything with 0 conversions at 1x payout in spend, or anything with a loss of >1x payout - use the best-converting placements to test your creatives for cheap, and THEN use your best creatives to retest all placements that were cut before.

Every situation is different - but I hope I've listed some of the factors that need to be considered.



Amy


Home > Paid Traffic Sources > Native