I think they mean the research data won't be used directly to target ads or try to make sales to you (standard guideline for market research), rather than the second order effect of allowing them to improve their overall ad targeting which then affects you.
I think that just means DataRank is a market research tool. It won't mean companies that use it will spend less on advertising, but it will help them target specific demographics, perceptions, etc more efficiently. Not less spending, just different.
So am I understanding correctly that "market research" means selling data mining insights to advertisers? Do you expect this to be a significant portion of revenue?
(I don't have a problem with this- I'm just curious.)
> "I am not convinced it is that useful to have explicitly targeted audiences for advertisement. It may be better to understand and interpret an ad campaign, but generally, what works best when doing statistical prediction is not what is the most intuitive."
If you've collected marketing data that tells you your conversion rate among pet owners is 2% when your conversion rate among the general public is 1%, then you know that every ad you show to pet owners is going to have twice the impact on your bottom line.
Ad targeting enables you to collect this kind of data and to exploit it once you have it. If you find the right audience, your advertising dollars can go more than twice as far.
The study isn't published yet so I withold my judgement. Given that the results are beneficial to people selling adverts caution should be exercised. (Two of their big problems are that click through rates are abysmal and many people dislike ads. They benefit if advertisers believe they achieve "additional sales".)
EDIT: FOLLOW UP. I had a deeper look. The article is dated, "Posted in Business on February 3, 2017". I found one copy of the paper published December 29, 2015 and another July 26, 2016. Both same I think, but one author had moved from Google to Pandora.
The article text seems off beam. The paper describes a method to better estimate advertising impact, not directly what that impact is. Odd.
Well, they have sales data from before the ad ran, and then they get sales data from after the ad ran. It's not a perfectly controlled experiment, but if the ad has an impact (or doesn't) they'll be able to tell.
No, it's not. It never claims advertising doesn't work. It claims that data-driven targeting is a scam on multiple levels. That isn't to say advertising doesn't work, just that "data-driven" doesn't improve its results.
I read that line as "they have detailed data on how to maximize total ad views where losing some % of viewers is made up for by the increased number of ads watched by the remaining viewers."
And if they had selected for your feed based on "what will make you buy stuff in the ads" and not publish the results, that would just be "business as usual".
And yet that seems a lot more dangerous than this study...
> The more data points they have on each user, the more numerous results they can return to recruiters and more specific targeting they can do with ads.
The thing is that endorsements at least are pretty noisy data. What's the point of shooting for quantity at the expense of quality?
That isn't how it works. The study clearly states that the targeting criteria was not included in the analysis. If you look at any 2 advertisers running their own campaigns, and one is a savvier user of auction based ad systems, a 29% difference in CPM excluding targeting is common.
You're example is obvious because Newegg can't tell the difference between market research because you're about to buy versus research because you're about to sell.
But it's always amused me (in a not very amusing way) that immediately after I purchase something I'm inundated with advertisements from the same vendor trying to sell it to me again. Big data is supposed to be about finding statistical relationships between events, right? In what world would the probability of someone buying something be highest when they're holding a brand new one in their hands?
What it really makes me think is the web advertising folks are pulling a con on their clients. They're not actually doing the critical analysis they promise but simply sorting by linear distance in n-dimensional space and hoping no one notices how useless the measurement is.
I wasn't able to interpret your post in terms of results, i.e., in dollars or ROI, and I don't know if your results are representative of advertising overall.
reply