[1] http://www.kaushik.net/avinash/multi-channel-attribution-mod...
> If FB can provide advertisers with the tools to more definitively say "this is making you money" and not just "oh, we summed up all these random stats and call it Engagement--look at the shiny engagement metric!" I'm confident they would find advertisers even more willing to open their wallets, shift spend away from the GDN and exchanges, and stop questioning or caring about click fraud since it would be a self-correcting issue (ie. high levels of click fraud would in theory result in lower perceived incremental lift if proper analysis was conducted).
Engagement metrics to me are much more about brand awareness then it is converting sales, so I didn't understand why you would point to engagement metrics when questioning tools that drive sales. Are you saying that you use conversion based ads on FB and aren't confident that their attribution window provides a highly accurate view of their performance?
I may have mixed my points up a bit, but basically I've noticed that they seem to lump quite a few things under "engagement" (an autoplayed video that plays for 3 seconds from someone scrolling in their feed is apparently a valid view). I'm using engagement in the broader sense, not the specific definition in the FB platform here, although it is pretty all encompassing there as well in terms of Likes/Comments/Shares (none of which are really the apples to apples at all).
My bigger beef is on the view-through side. Attribution windows alone don't dictate attribution weighting. You really need to be measuring incremental lift, otherwise you end up with lots of people seeing an ad and a ton of view-throughs, but a lot of wasted spend since many would have converted without the impression.
Sure, the user saw the facebook ad or clicked on the facebook ad. But... does that just mean that facebook is good at showing ads to the sort of people who already buy those products?
(I know that when I'm running without ad-blocker, I already own most of the products that correctly-targeted ads try to advertise to me. My argument is that the ads are merely following my preferences, not driving those preferences.)
http://blog.wishpond.com/post/63739323752/how-to-use-the-fac...
If you see an ad on TV, and then 2 minutes later receive an ad on Facebook for the same product, click through and buy then how much is the TV ad responsible vs the Facebook ad.
Same for Youtube ads. Typically (as a very broad rule) people don't click through to purchase from a Youtube ad. And yet lab (and other) studies show an increased propensity to purchase a specific product after Youtube advertising exposure.
As an ad buyer, how much do you spend on those TV/Youtube Ads, vs how much for the final click.
lcs's point is that sometimes that (paid) final click comes from a user who was probably likely to purchase anyway. So an advertiser would be better off paying more for wider exposure than for that final click (assuming a fixed budget etc etc).
This is a live problem right now, and a solution is worth billions of dollars.
But more broadly there are quite a lot of industry and industry-body sponsored ones, eg:
https://www.thinkwithgoogle.com/articles/online-video-ads-dr...
http://www.emarketer.com/Article/Mobile-Video-Ads-Lift-Brand...
https://scholar.google.com.au/scholar?hl=en&q=video+ad+expos... will find more.
I've not yet seen spectacular results for video, speaking as someone in the above category (some of the time) but my hunch has long been it can be very effective.
Clearly there are still holes in the methodology, but it's at least as good as I've ever seen done.
Any other hints you can share without violating agreements on the path to success and measuring success?