Mobtown Labs

Attribution’s effect on The Ad Market - Demand Side

This is a continuation of my post on Friday discussing how the Supply side was affected by attribution modeling and, though I’m confident it will be a good thing for the ad market because I think online ads sell for too cheap these days, I can see how publishers are worried and advertisers could be too opportunistic. That said let’s examine the Demand Side.

Supply = Publishers (Friday’s Post)

Demand = Advertisers (This Post)

Demand

Attribution modeling is truly an Advertisers dream. Many publishers attempt to offer it like Doubleclick and Atlas’ Engagement mapping or the Advertising.com custom data mining offerings but there are inherent flaws because data is limited, the whole puzzle isn’t visible to the publisher, or advertisers are just skeptical that there could be a conflict of interest in that they are just trying to sell you more advertising.

Every marketer’s job is to buy advertising and make creative and then align the cost of doing so in an attempt to make it drastically cheaper to acquire new customers than what you are charging the customer. The problem with this is that without a solid attribution model, you never truly know what piece of media and creative is actually driving that customer. Furthermore it rarely is just one ad impression so the question becomes what combination of media over what time period (and what time interval) is the most cost efficient.

Couple that with the fact that you either need to pay for each individual ad (like a homepage) or you pay for a result and you have no control over or have no way to track how and when the ads were shown (like a pay per click search text link) and you have quite a messy scenario.

So in the grand scheme of things there is only upside for the advertiser with attribution modeling but there is a major philosophical flaw. Most marketers if they look at their marketing plan have customers that they think are for free. Natural search traffic, organic traffic to your site, customers walking in or calling in and just buying a product.

So I will be bold enough to lay down the Cardinal Rules of Attribution Modeling:

1. There is no such thing as free traffic

2. There is no such thing as a free customer

3. If you have free traffic and free customers, your attribution model is BROKEN.

I will end with a ridiculous scenario and prediction. I personally think that the demand side of attribution modeling could be the fix to ad market growth and economic growth. The reality is that we are doing less work and incurring less costs to acquire new customers, so therefore companies like Amazon, Google, and Yahoo can deliver sales to customers at a much lower overhead than newspaper, TV, and radio have done in the past. This concentrates dollars in marketplace pricing scenarios and those dollars are then compressed because these businesses use their low TAC to offer low ad prices.

Based on economic principles alone, if every advertiser knew what ads truly drove their sales right now, they would double down their ad dollars. That would double the ad market and pump twice as much money back into our economy and all problems would be solved. THE END!

Get Ready for the Coming Land War in Online Display Ads (adage.com)
Search Advertising (slideshare.net)
Awesome Map Of The Digital Display Advertising Tech Marketplace (adbean.net)
The rise of RTB and our investment in Triggit (sethlevine.com)

Tags: Google Advertising Online advertising Marketing Yahoo Doubleclick Pay per click Web Design and Development

Comments

So, where is the data and how fast can it go?

Image via Wikipedia

Working at a big company you get very used to how things are done. Data has a certain availability and you need to work around this and play within the confines you are given as any major infrastructure changes could take months or years and there is always the fear of killing the golden goose and switching costs from going to from one system to another. Basic rule of thumb is use what is currently working and innovate within the confines of it unless there is a major loss or need to change things.

I’m specifically talking about adserving technology, the databases that store information that is gleaned out of adserving and then the structure of this data to query and report on.

I got in a discussion with the CEO of a emerging data company who I have a ton of respect for recently whom also has an extensive technology background. Our discussion was around real-time data reporting and the feasibility thereof. Typically most adservers dump logs into massive long-term storage databases using either hadoop, neteeza, or even oracle to store this. There is definitely a maximum number of records you can insert per second, limitations on the structure of this data which could complicate how you pull it out later for reporting, and furthermore the more data you store in one place, the harder it is to pull out a tiny piece of it on a quick recall.

When talking about advertising, for an individual advertiser or marketer, you are talking about between 10 and 100 million display ad impressions per day along with tens or hundreds of thousands of clicks and tens or hundreds of thousands of page views on a daily basis so to break that down lets say you need to store:

100 mm impressions

300k clicks

20 k page view events

Per day totaling 100,316,000 event records

so that’s 4,179,833 events per hour

or 69,663 per minute

or 1,161 per second

Then you have to think about how that number will spike during certain hours of the day and then lets say you definitely want to design the system to handle a lot of advertisers so let’s say 1,000 advertisers….you are talking about being able to handle between 1 MM and 20 MM events inserted into the database per second.

So how do you do this while managing costs? And is it even doable?

One thing is for sure that you need to own the data source so importing data from third party adservers or publishers is off the table because the server to server transfer alone will add valuable seconds on to your process and if you ever plan to do calculations on that data like conversion rates, match processes, click thru rates, you are adding time on to process that, and we haven’t even gotten to querying that data out of the database yet…..we’re just inserting it. That said querying it out if structured properly will be a lot quicker and easier because you just have one or two users querying at a time per advertiser’s data set.

Anyway…..just thinking out loud about the problem at hand and the gap between real-time bidding and actually pulling the data based on real-time bidding into a reporting interface so a human being can actually look at a marketing problem and address it or examine it.

As a business side person myself, would love to see anyone else’s commentary on what database structures they have used and the feasibility of a project like this.

Bsac what keeps me awake at night final (slideshare.net)
Partition Wise Joins (blogs.oracle.com)
Algebraix Data Awarded Patent for Breakthrough in Data Management (eon.businesswire.com)
Data Explosion: Analytics Software Must Adapt or Die (readwriteweb.com)

Tags: Database Data Real-time computing Technology Advertising Business Rule of thumb Data Warehousing Mathematics Marketing

Comments

Taking the High Road on Attribution Modeling

Image by AWWS via Flickr

A year or two ago you started to see a lot of buzz in the marketplace about attribution modeling and analytics. Atlas and Doubleclick started coming out with products they called “Engagement Mapping” and agencies and marketers dug in! Finally someone would try to crack the nut of the age old question by John Wannamker:

“I know half the money I spend on advertising is wasted, but I can never find out which half. ”

So what ever happened with this?

You really don’t hear too much talk about Engagement Mapping anymore. Everything is more about “Audiences on Demand” or “Reatlime Bidding” or owning your own data warehouse and retargeting audiences.

Is this because people did the Engagement Mapping and figured out their attribution model and found the ‘half’ that is not a waste and is trying to buy that at the cheapest rate possible?

I don’t think so. What I think happened is that we started to dig into the Engagement Mapping and Attribution model tools and first of all realized warehousing all of that data and keeping it and analyzing it over the long term proved to be incredibly expensive, incredibly cumbersome to compute and quickly access and the Engagement Mapping product could not be easily productized and made into a nimble nicely packaged tool where marketers and agencies could quickly make changes and show results, they bagged it and went for the next best thing.

Essentially they took the high road on Attribution Modeling. Realtime Bidding and targeting audiences essentially productized and made Demand Side Platforms able to quickly and nimbly scale the ‘stuff that works’.

To better explain lets go back to John Wannamaker. He said that he knows half of his ad spend doesn’t work. Well we have come a long way and with technology now we are down to knowing that half of our ad spend doesn’t work, about 5-10% of our ad spend definitely does work and we can really do a lot of work and gain a lot of efficiencies in that 5-10% and I think that’s what DSP’s and a lot of the data targeting is doing.

Retargeting we know is a gold mine and for large advertisers, it’s reasonably scaleable. We also know that search data from companies like Magnetic.is is really valuable but we’re still finding out how to scale that and which keywords and from what sources will that work from.

These are great findings and our industry and marketers are much better served by knowing that these audiences are some new more robust pillars to stand on.

That said there is a whole lot more and I think we quit too early and there is a lot more innovation to do on Attribution Modeling. The data is there, the capabilities are there, it just needs to be figured out and the technology needs to be optimized because right now even with hadoop, cassandra, cascading, and some of the methods the guys at Flightcaster are using, data is still not normalized enough and inexpensively scalable enough. Also, the big dollars are being spent on the media company side (i.e. google, yahoo, AOL) and not on the advertiser or agency side because those dollars don’t exist.

I think as agencies expand their margins and as the DSP and Data space starts to cool and consolidate the next frontier will be Demand Side Attribution Model Software or DSAS or DeSAMoS or DemSAMoSo or maybe someone else can work on a cool acronym while I work on the product…….