Quant v. Audience: The Real Challenge and Opportunity is Using Both

Guest Post by:  Matt Patton Esq. Director of Optimization and R&D at DoublePositive

English: An audience in His Majesty's Theatre ...
Image via Wikipedia

Among the many nuances in the debate about which optimization algorithm represents the best RTB media buying strategy, the most prominent might be the audience retargeting v. audience agnostic approach.  Although exclusive to our business, this dichotomy is conceptually no different than two Wall Street analysts debating about whether a pure qualitative approach is better than a numbers-only quantitative method in purchasing securities.  My engineering training tells me that the only inferences that should ever be drawn as a basis for any decision should originate with the story told by what was actually measured.  But my legal training tells me what I think is the more correct approach, which is, as every law school professor professed so eloquently, “It depends!”

The truth is that measured performance numbers, i.e. impressions, clicks, conversions, and audience data, i.e. categorical audience profiling, both have their own benefits and drawbacks, and will almost always provide you different oscillating returns on your media investment.  On the surface, the only real difference between the two is that one is instinctual while the other is empirical.  With audience data, we infer that a group of users with certain online buying habits will purchase what our clients are selling based on the categorical distance between things previously purchased and what we are currently selling.  With fundamental performance data, we just care about which pieces of inventory have already yielded a statistically significant number of conversions regardless of who actually converted.  Thus it is blatantly clear under the surface that both methods offer the same benefit, which is a calculated a priori probability of a conversion from which we can derive an expected value of an impression, or what my mentor at Advertising.com used to call the “Crown Jewel”, of DR advertising.

The obvious drawback of both approaches is and will always be the cost of learning.  Because of the random nature of Internet advertising, we need to survey the biddable landscape before we can make any type of calculation, qual or quant.  In the quant world, it probably takes more time and thus requires a longer discovery period, while in the audience buying game, those “selective” impressions can get quite expensive when we are still learning which audience represents our sweet spot. Since there is no such thing as a free lunch in this business, both are going to require some upfront cash that will initially skew our calculated CPAs.

Instead of simply accepting this as truth, I want to treat this as an opportunity to thwart the economic black hole of overcoming a campaign’s inertia when starting from rest.  I see opportunity to utilize both approaches as a function of time and campaign maturity in order to get campaigns running more quickly and performing better long term.  I see complex regression analysis and time series plots that will tell us on a campaign by campaign basis of when and how to use audience targeting and when the stats tell us that the audience is worthless.  I see recurring real time cost benefit analysis algorithms telling us that we are paying too much for an audience or that an audience is underpriced.  I see this as a first step towards creating a RTB marketplace where being smarter actually translates to being more successful. For the first time, I see a scientific approach to an age-old art and an artistic approach to centuries-old mathematics.

Will this be accomplished and will it even make a difference? I leave you as a lawyer by saying “it depends”.

Related articles


Attribution’s effect on The Ad Market - Supply Side

When demand D 1 is in effect, the price will b...
Image via Wikipedia

So I recently engaged in a debate about how solving the attribution modeling problem could effect the ad market.  The conversation was broken into supply and demand where:

Supply = Publishers   (this post)

Demand = Advertisers  (the next post)


On the supply side solving the attribution modeling issue could affect publishers in a drastic way.  Since ad inventory is becoming so fluid as a result of companies creating marketplace environments out of their ad inventory, the value that an advertiser sees in an ad, could really affect a publisher’s business.

We already know that marketplace buying drastically compresses ad dollars.  We see that with advertisers who have pulled money out of newspapers and TV and put it into google.  Google takes these dollars and since they can deliver the eyeballs and customers at an incredibly inexpensive overhead cost, they allow advertisers to essentially set the price of each keyword ad.  Since there are so many keywords and relatively few bidders on each keyword and such low cost to deliver each incremental customer, this works extremely well so there is plenty of diversity for google to rely on their prices to continually rise as competition increases.

For display advertising it’s not the same.  A homepage 300x250 ad typically only has one advertiser per day and with 365 days in a year and between $50k - $2mm per day (depending on the site) that’s a lot of eggs in one basket for a publisher. 

So they are highly incented to keep advertisers from knowing the real value of their advertising and they can create scarcity in a high competition placement to drive up price.

The threat here is that if the small group of advertisers (it can’t be more than 365) all know exactly how much this placement is worth to them, they can hold the publisher hostage on price.

I think that gives an idea of two polar opposite sides of the spectrum for the supply side.  One area with a lot of diversity and little competition where having true attribution and REAL pricing will not threaten a publisher’s business model and another where little diversity and high competition if attribution is introduced could curb that high competition and cause problems for a publisher.

There is also the fact that REAL pricing and a sound attribution model could make advertisers identify what really drives incremental sales in which case all publishers win!

Enhanced by Zemanta


Targeting Sites vs. Targeting Audiences

Tower of Babel
Image via Wikipedia

I came across a post by Brad Terrell talking about the Appnexus Innovation summit and he highlights the topic of targeting sites vs. targeting audiences.

He goes on to talk about two different ad networks, their differences and gives emerical evidence of his thoughts:

Adam also illustrated the value creation potential of the audience-driven approach to targeting by pointing out how the enterprise value of a representative “old school” ad network, Burst Media, paled in comparison to the much higher enterprise value of a newer “audience-driven” ad network, interCLICK

 - Please read the whole post as this could be considered out of context and its worth reading the whole post http://loca.ly/couk6j

So I think about this from a entrepreneur's perspective and I completely agree.  Advertisers and Marketers cling to what is new and there is continual pressure to do a better job and take advantage of the newest technologies and methods possible.

That said, DON’T OVERLOOK SITE TARGETING.  Audience targeting if done properly with re-targeting or search keyword re-targeting such as with Magnetic.is can be very powerful and literally spin gold when it comes to bottom of the funnel and making sure you get in-market sales.  

Beyond this however lies the ocean of ad impressions that are still being boiled by the likes of the big ad networks and exchanges.  I recently had a conversation with the VP of Operations at one of the larges online ad agencies pitching my business and knowing that I used to sell to him I asked, well what do you think I should be buying.  He responded, that when you are doing direct response advertising where you need to hit a specific ROI to continue to invest, what you do is buy all of the ad networks and optimize the best you can.  He said DSP’s are great for getting that re-targeting audience and a couple other segments but largely a lot of the audiences do not work and there are so many of them that even if there is an optimal audience, you will waste a lot of money trying to find it.

From my experience on the sales side on any given campaign there were 2-3 audiences that really worked on any given campaign (one being re-targeting) and the rest of the heavy lifting was done by site optimization and frequency within site optimization.  In fact the most important driver is something that I rarely hear people selling on or creating business models on, which is site frequency and the fact that the first few impressions (1-3) are in fact the most valuable by a long shot.  Most publishers know this and technology lends to selling this easily but for some reason it is not sold out there on the open market.

Furthermore there is also a lot of evidence that brand name new sites and mail sites such as LA Times, NY Times, Yahoo Mail, AOL Mail and others similar create the most ROI impact if bought at the right frequency and attributed properly…..which leads to attribution modeling which is a whole other post which I probably need to write more on.  That said, if bought through an ad network or exchange often you get the higher site-frequency impressions and these sites don’t appear to work as well.  If bought directly, you have the burden of paperwork and exorbitantly high prices to get access to the right impressions of this media.

So because they have such a broad view of their buys I think a great new frontier for ad networks SSP’s, and DSP’s is to sell based on site-frequency.  They have this data and often know where they are in the publisher daisy chain and can aggregate supply across many sites and can efficiently convert this into a fluid buy.  

And, I would buy it…..

Related articles by Zemanta

Reblog this post [with Zemanta]

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

Math paper
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.

Related articles by Zemanta

Reblog this post [with Zemanta]

Real-time Bidding without Real-time Reporting

Image representing Invite Media as depicted in...
Image via CrunchBase

So Invite Media was bought by Google today.  To me that means Realtime Bidding is officially ON.  We all knew it’s valuable and the mechanism to do it has been validated.

But that still leaves wide open, what to do with it now that you have the mechanism to do it.  What is the use of real-time bidding if you can’t have real-time reporting and you can’t have a real-time dashboard to examine, analyze and control that. 

There is a critical problem there.  Decisioning on the price in real-time if you already know who a user is has been made readily available with DSP’s like Invite Media, however, what about real-time estimation on that impression and what about crafting your algorithm to learn and adjust things like frequency, ad sequence, and offer based on that real-time bid.  I’m hoping google will quickly string together Teracent and Invite to be able to do this.  Only time will tell.

But on the Media Buyer or Agency side where I now sit, other things come to mind.  Like if this data is flowing in and I can bid in real-time then I should probably get access to map this data to my marketing plan and report on this based on my marketing plan so that my marketing plan becomes a living breathing thing.

To do this you need real-time Reporting to match your real-time bidding!

Related articles by Zemanta

Reblog this post [with Zemanta]

LCD Soundsystem - New York I love you but you’re bringing me Down

Today is my last day at this job, so thank you for everything I’ve learned and the people I’ve met along the way.  This job was a true life changing opportunity and it made my career what it is today.  I will always be grateful for being given a chance, but it’s time to move on.

Thank you!

Reblog this post [with Zemanta]


Online to Offline Attribution Tracking - Stickybits API

UPC-A barcode
Image via Wikipedia

So I’ve been trying to keep myself contained waiting but it has finally happened….

Someone has officially linked the offline product sales tracking world to the online advertising world.

Yes, that’s right, I know you’ve all been waiting for it too!

Stickybits has opened up their api to the public http://code.google.com/p/stickybits/

Mark my words, this point in time is the opening of the proverbial floodgates of online to offline sale tracking.  Now any old schmuck who can get their hands on a python developer can put Ticketmaster, OpenTable, Live Nation, and any major retailer out of business.  I mean seriously….these guys ought to just file for bankruptcy now…cut your losses!!!

Why is this, might you ask?

Well for those who don’t know Stickybits is a company that issues bar codes and connects them to a url.  This url can theoretically be anything.  It can call an image, it can call a website, and it can also trigger a purchase or redemption through a URI call.

I will be curious to see what restrictions and what type of load Stickybits handles.  Also there’s the fact that taking a picture of a barcode on a smart phone is kind of unreliable and a clunky user experience, but all of that will work it’s way out the base framework for the business model is there.

I look forward to seeing the innovation that comes out of this.

Related articles by Zemanta

Reblog this post [with Zemanta]

The Next Big Ad Market Compression

Facebook, Inc.
Image via Wikipedia

The more I think about attribution the more it leads down the road of connecting the ads with the true direct response vehicle.  I’m not particularly talking about direct response advertising as it has evolved to today as that is really just optimization of the bottom of the funnel.

And contrarily brand marketing has still done a good job of keeping itself disconnected from direct response.  So what would happen if you started cutting that branding budget and relying on just direct response advertising?

Or what if you cut the direct response dollars and just tracked all the brand impressions back to the direct response mechanisms i.e. call centers, sales people, and online commerce carts.  We can do this.  So why don’t we?

Well I have to give google credit.  They passed up the brand dollars. In their earlier days they offered cpm advertising at the top of keywords.  They later cut this out and moved over to an all cpc model.

Now we are seeing facebook doing this as well.

So these are the two biggest marketing vehicles of our recent history and they are passing up the branding dollars which could have a major backlash on our industry.

We already saw the compression of ad dollars when google swooped in and captured massive ad budgets.  Well arguably they are getting the same results as when people would spend 10X what they spend on google on TV, Newspaper, and Magazines.  Now Facebook is coming in and offering the best targeting our industry has ever seen and is passing up the brand dollars.  

I think we will see another wave of ad dollar compression industry wide from the facebook price change backlash.

Also, I think it’s time for disruption at the big agencies and starting to use some of the tools that we have with ad servers, scoring, data warehouses, and call tracking to connect the brand dollars with the call centers, sales channels, and ecommerce carts.  

Maybe a new smarter view thru metric will be the catalyst for this…..

Related articles by Zemanta

Reblog this post [with Zemanta]

Revenue Model Cop-outs - High Hopes for Twitter

Image representing Twitter as depicted in Crun...
Image via CrunchBase

I kind of hope Twitter has a cool and useful revenue model.

I was thinking about this today after talking to a friend of mine at the wharton school about the life sciences startup world vs the tech startup world and how a lot of the life science startups take a lot more money and a lot more time to produce an exit or value whereas you could have a web app startup company 3-5X over in the time it takes to evolve a drug or procedure or piece of technology in the life sciences field.

This lead me to think about Myspace and Facebook.  Myspace, first to get huge and basically just plugged in ad networks and then sold brand advertising and then tried to cobble together targeting methods.  Facebook, now with an ad platform with pretty solid targeting and a lot more scalable for small to medium advertisers.

But are these really useful and interesting.  I mean Myspace has a massive database of bands and comedians and entertainers that use their site as their homepage and their portfolio.  Why haven’t they evolved enterprise tools for these users that are more useful.  Facebook with its apps and api and facebook connect, why have they stalled on peer to peer payments and are going toward the ad model.

Is the ad model the easy way out?  Are ads and ad networks a conduit to not having to innovate?  Is facebook and myspace short selling themselves when it comes to being useful to the world and it’s users?  Are they stopping short of disrupting companies like Ticketmaster, Paypal, Bank of America?  Could Facebook go head to head with Bank of America?

Now what for Twitter?

From ReadWriteWeb quoting Peter Kafka:

  • Ads will be tied to Twitter searches, in the same way that Google’s (GOOG) original ads did. So a search for, say, “laptop”, may generate an ad for Dell. The ads will only show up in search results, which means users who don’t search for something won’t see them in their regular Twitterstream.
  • The ads will use the Twitter format — 140 characters or less — and will be distributed via the third party software and services that use Twitter’s API. The services will have the option to display the ads, and Twitter will share revenue with those that do.
  • Twitter will work with ad agencies and buyers to seed the program, but plans on moving to a self-serve model, like Google offers, down the road.

Read full ReadWriteWeb Article http://loca.ly/ade2jo

I mean surely they will make money from an ad platform but really…..is this the best that they can do?  Isn’t there an opportunity for them to put Reuters and Associated press out of business and revolutionize real-time news and real-time headlines?  C’mon now, lets be useful!

Related articles by Zemanta

Reblog this post [with Zemanta]