Display + Search

Today I have a new direct report starting and am very excited.  Bright, articulate, driven and already adding value even before she starts.  I was putting together a “how to succeed” document for her and figured I may as well share it.  Utterly presumptuous.

I currently lead sales in the East for MediaMath.  I’ve led consulting teams, operations teams, yield teams, biz dev teams and, now, a sales team.  When building my teams, I’ve looked for the same thing: Driven multi-talented athletes who love marketing.

If you want to succeed in this business, here is my list of how to do it:

Execute – you simply must get things done and that is job one. Don’t get caught up in the hype.  Don’t spend days at Adtech and drop the ball on driving your results.  The conferences and lunches and drinks are the gravy.  My daughter has to eat her dinner before she gets to watch her shows and you have to sell some stuff before you can go to iMedia.  I’ll admit to having fallen into this trap at times.  Also, don’t try to be too smart.  Smart without getting things done is an unemployed person that people meet at Starbucks in Astor Place to “pick their brain”. Better to be described as “wow, he’s an animal. he just gets things done”.

At Cause – this means that you create what happens.  Things don’t happen to you, you make them happen.  As far as I’m concerned, every campaign of any of your clients is your problem. As Smokey the Bear says, “only you can prevent forest fires”.

Intellectually curious – you should find this stuff incredibly interesting.  All of it – from learning how client-side cookies work to having lunch with an SVP of Analytics at a major holding company.  If you find yourself sitting in the wrong panel at AdTech and it’s about SEM for local businesses, you should think “well, I’m here, how is this relevant for me? What can I learn?”  Specifically to sales, while I appreciate guys who just love the sales game and love to close, if you don’t love the industry and don’t appreciate a real marketing challenge and how to solve it, you simply won’t do well with me and you won’t understand how I expect you to sell.

Make sure it’s hard – If it’s easy, you’re not doing it right because anyone else could do it and then who needs you? Trust me, winning six championships was hard for Michael Jordan.  So if you feel like it’s easy, work harder.

Absorb like a sponge – related to being curious, you absolutely need to absorb what people are saying and doing in the industry (and by the way – those are two different things).  What are they thinking? How does it impact my company and me?  You do that in a number of ways.

• Conferences – meet with people and LISTEN.  I learn more at conferences than people learn from me (and that is my goal). Sure, I make contacts to follow up on, but I spend most of my time listening and learning.  I ask questions rather than pitch.  I also focus.  I see people at conferences with their laptop open, managing email or IM’ing a colleague or whatever. Sure there are times when there is something you simply have to get done (see “Execute” above) but multitasking rarely works because you end up doing both things poorly.  So if you’re going to be at an event, focus on the event. I no longer bring my laptop to conferences for just this reason – I couldn’t focus.  And if I find myself focused on my email or something else, I leave, because the conference is not earning my attention.
• Network – meet with people in the business. I’m shocked by how many people simply don’t leave the office and don’t meet with people outside their company or even their department.  Sure, there are internal and external jobs, but an hour at lunch with a counterpart from another department or a partner, customer or competitor can give you a whole new perspective of what you are doing.  If you’re internal and you don’t think you know anyone, use Linkedin. Find someone.  Even if you’re Ad Operations you can join Admonsters and meet people and then go to lunch or grab a drink.  Knowledge sharing like this helps you and the company.
• Read – many people in this industry love to tell people what they are thinking.  After all, why on earth would anyone have a blog or tweet?  Read, absorb, and relate it to what you do.  There are a few publications and people you simply must read regularly:
  • Adexchanger – this is a must on a daily basis.  John Ebbert does an incredible job bringing together all that is important in the industry.  I insist that my team read this daily and they should expect to be tested on it (“yes, Gabe, this will be on the final”).
  • Exchangewire – also a must focused more on the European market but lots of US as well.  Don’t be a typical American – there is a lot of interesting stuff going on outside of the US and it’s applicable to you even if you “just” do AdOps or “only” sell in the Midwest.
  • Blogs – there are a few bloggers you have to read when they post and I focus specifically on guys who are (a) doing stuff and (b) have different lenses than me.  Here is my short list (in alphabetical order):
    • Darren Herman
    • Rob Leathern
    • Jonathan Mendez
    • Jerry Neumann
    • Mike Nolet
• Twitter – note, I didn’t say Tweet. I said Twitter.  It’s a firehose of information and, if managed correctly, you can let the world do your research and collate for you.  Actually tweeting is a personal decision driven largely by personality and motivation.  You can’t be made to tweet and twitter accounts where someone is made or encouraged to tweet tend to show it.  But you should have a twitter account, follow the right people, and spend 10-30 minutes daily seeing what is being talked about.  As for who to follow – start with Adexchanger’s list in the sidebar and follow people who are followed by people you find interesting..

Don’t believe the hype –  Chuck D was right. Take everything that is written or said with a grain of salt.  We all have our own perspective and motivations.  Recognize that a post on a company’s blog is their perspective, not necessarily the only perspective.  Articles in imedia, clickz, mediapost also are written from a specific perspective and tend to be position pieces supporting a specific product or service (oh, and so do Q&A’s on adexchanger).  That doesn’t mean you shouldn’t read them, just the opposite. Understanding these perspectives is critical. I read and re-read any articles from guys at other DSP’s and networks to learn what they are thinking. Where are they going? What are they telling my prospects that I need to counter?  But they may not be “right”.

Be thankful – I consider myself extremely lucky to have literally fallen into this industry.  Two weeks before I got my first job in advertising I interviewed for international internal audit at AIG.  Let’s just say I’m much happier meeting with agency heads than I would be reviewing claims processes in the Philippines (and I think about that often).  People on my team should appreciate that they are in a high growth, super exciting, fun industry.  We should also remember that, at one time, automobile manufacturing was high growth with 100s of startups.

Carry on.

We return to the 2011 marketing plan for Carroll’s Florist.  Apologies for the delay, particularly to my cousin Chad, but snow, ice and a teething 14 month old got in the way.

Last time I said we’d be laying out the plan, but I want to really walk through the process of getting ready.  The next few posts will be a combination of how we are planning 2011 while also making sure that the pieces are in place to make our efforts a success.

First, let’s review what we‘ve done.  We did some test buying in late 2010 and the results were good enough to decide we wanted to do more of this.

A quick summary of 2010 buying:

• Facebook:  1.3MM impressions; 469 clicks; $377 spent; $0.28 eCPM; $0.80 CPC
• Google Adwords: 6,077 impressions; 38 clicks; $74 spent;  $12.17 eCPM;  $1.95 CPC

First, if you think these numbers are small, the population of Staten Island is approximately 0.1% of the US, so if you want to think of this as a national campaign, just act like the numbers are in thousands.

We spent a little less than $500 in about a month.  Unfortunately, we really couldn’t track success.  The Carroll’s Florist website doesn’t have a real goal event and most business comes in via phone or actually walking in the door.  That said, anecdotal “feeling” of the staff was that business had picked up slightly and we were able to track requests for information on the website and that seemed to increase.  But again, we just don’t know. What scared me was we spent nearly $500 (and remember that would be $500,000(!) if a national campaign) and we didn’t really know if it was working. We just thought it might be.  If this was a national brand, the CMO would be looking for a job right now. Just before Christmas I halted buying until we could fix this and make sure that we were spending money effectively.

So the first thing we need to fix is what is called Attribution – and this is the focus of this post.  For any marketer, proper attribution is one of, if not THE, most important thing to solve.  Attribution is the process of attributing results to marketing efforts.  Those results could be new customers, purchases, leads, whatever.  For example, yesterday I searched for “Canyons Discount Lift Tickets” and purchased a set of tickets after clicking on a paid search listing for go-utah.com.  The marketing manager at ski-utah.com would attribute my purchase to Paid Search. More specifically, she would attribute it to the search term “Canyons Discount Lift Tickets” which would flow into her models and drive her to increase her bid price on that term.

However, while my example was pretty easy, attribution often is not. What drove me to look for discount lift tickets? Why am I staying at the Canyons? Why am I skiing? Why am I going in February?  Depending on the marketer, all these questions need to be understood because many other efforts may have driven me to do that search and convert. If those efforts are not properly valued then the marketer may not dedicate the proper budget (or cease altogether). If those efforts actually do drive value and they decrease, the marketer will see a drop in her results and not know why that happened.  A death spiral ensues.

The point here is that attribution is critical for Carroll’s Florist.  While we may think that our ads have been driving some increase in business, how do we know?  We need to set up a process that allows us to capture this information so that we can analyze it.

We will do this by improving our web analytics, by implementing dedicated and unique phone numbers and by providing unique discounts or special offers to customers.

• Web Analytics – There are two major steps to any web analytics implementation. First, get it implemented properly on the site, and, second, set up a system for tracking what you do.  When we started this effort we actually had no web analytics so we had to fix that.  Carroll’s had implemented Google Places (great!!) but mistook this for Google Analytics (not great).  It took us about 15 minutes to fix this and our web analytics data began flowing through.  Now we need to make sure that the results of our marketing campaigns flow properly through to our analytics when customers come to the site (I call this campaign tracking).  This was not done in 2010 and we need to fix this before we launch any new campaigns.   TO DO: Fix campaign tracking
• Unique Phone Number(s) – A great deal of Carroll’s business comes in via phone. Before this effort started it was nearly impossible to communicate to the florist via the website (the site repeatedly encouraged phone calls).  While we made sure customers could get in touch with us via the site, we still need to recognize that a lot of people are going to want to call, and we need to be able to track that.  We’re going to implement unique phone numbers on the site.  There are companies who provide unique toll-free numbers that can be used for different purposes and allow the marketer to learn where the customer came from. We will obtain several numbers and place them in various points on the site so that we can track different efforts.
• Special Offers – advertisers have been providing special offers to customers who tell them where they came from  since at least the 1800s.  When the local deli gives you a discount for saying that you saw their ad in the Sunday Paper, they are tracking their marketing.  We need to figure out the same. I’ll be assigning Chad with the job of figuring out what he can offer customers for telling him where they came from.

That is my quick list of what we’re going to implement in the short term.  If you have any suggestions, comment below, and I’ll probably add them

Coming up soon we’ll lay out the marketing channels we plan to test, what campaigns we plan to run, and what we need to do to the website.  Among a lot of other stuff we need to get going.  Spring is around the corner and Weddings are the bread and butter for the Florist, so we need to be ready soon.

Off and on for the last year or so I’ve been talking about Display and Search. All along, I’ve wanted to actually put all this theory into practice. To get my hands dirty and really see if what we’re talking about actually works and is true.  So in 2011, I’m going to do that and, if you like, you can follow along.  I’m actually going to design and run a digital marketing program, run it, and share it all with you.  Full disclosure, I got this idea from Adexchanger’s post on how his remarketing campaign went (so thanks John!)

Now no big advertiser would allow this to happen. It’s all too proprietary and super-secret.  So I had to go local.  Luckily, besides my day job selling MediaMath’s TerminalOne, I also act as informal digital CMO of Carroll’s Florist in Staten Island (I get paid in beer).  My cousin, Chad, owns and operates this 60 year old family business.  It’s a successful and profitable business that has all the building blocks to do this (eg, a website, some budget to spend, a pretty well defined target, etc).

So we’re going to run marketing for the florist and show you what we’re doing and what happens.  This will have the added benefit of letting us really explore what’s out there in local and social.

In late 2010, we started to do some basic online marketing (Adwords and Facebook).  We fixed some things which were broken.  For example, Google Analytics was not tracking properly and there was no real obvious flow on the website.  As we got close to the holidays we stopped spending, though, because the florist was just too busy to accurately track results of what we were doing.

So here is what we’re going to do:

1. Build the plan (next week or two)
2. Lay out a schedule – what will we do and when
3. Continue to work on some of the issues we face. A big one being that our true success event happens offline (generally at the store)
4. Take your input – if anyone has ideas or input, please comment.  We’d also be more than happy to test your product, if we feel it can help us meet our objectives and you’ll let us disclose the results.
5. Execute

A few housekeeping items:

• We’ll only use channels that are available to a small local advertiser
• We’ll disclose results but I wont get into the real nitty gritty (eg, what keywords worked).  We do have competition, you know!

So we now need to build the plan. Stay tuned.  That’s due February 1.  If I don’t get it in, I don’t get my beer.

Digital marketing suffered a loss last week. I hope we didn’t lose it for good.  You may have missed it. Facebook eliminated conversion tracking.  The explanation was something along the lines that they didn’t feel it was relevant to the social nature of their ads and they think it’s more important to track their social metrics like “shares” and “fans”.

Hmmm, what!!??!  What’s the whole point of what we’re doing?  Our goal is to get people to actually buy something!  So we need to measure it.  Not to maximize some interim step like becoming a fan.  I am a fan of Tankfarm Clothing and I haven’t bought anything from them, ever. I am a fan of Panera Bread and my “fan-ness” had nothing to do with me eating at Panera Bread, and never will.

Another excuse I’ve seen is that Facebook advertisers weren’t placing the pixels properly.  To that I say “So what?”   Major agencies and advertisers still mess up pixel placement.  Train them. Explain why it’s important.  Give them tools to QA the pixel placement.  Google’s hundreds of thousands of search advertisers seem to place pixels properly.

Let’s think about what might have been.  Had Facebook been a little forward thinking here and really embraced conversion tracking and made it something unique and designed specifically for social media.  Worked on it, evangelized it, encouraged their advertisers to learn how to use it.

Admit it, we all wonder if this social marketing really does anything.  We believe it should and probably does, but we don’t really KNOW.  Especially us quants who want to see numbers that prove it out.  We know search works all the time.  And we know display works most of the time and if measured properly.  But we don’t KNOW if social works.  And please, don’t tell me that your recent social campaign generated 150,000 fans on Facebook.  Awesome, good for you, how did that drive ROI for you or your client?  Hmm?  I love the Old Spice Ads.  I’ve shared the videos and searched for them and their derivatives.  But I still buy Gilette products when I go to Duane Reade.   Sorry Isaiah Mustafa

With conversion tracking in Facebook, we actually had the opportunity to learn if this social stuff really worked.  Some of the comments I’ve seen say that since we can track click-based conversions from Facebook using site-based web analytics, that it was redundant in Facebook.  But this misses the point.  First, it’s not just the click, especially in social(!), it is consumers’ interacting with your brand and building a relationship.  Second, by measuring on the site you lose all the granular data about what actually happened on Facebook and in the social graph. You simply know that someone came to your site from Facebook, you don’t know why.  If done properly, Facebook conversion tracking could have told you that.

The power and promise of social marketing is that one interaction can spawn a wildfire.  In both search and display, one consumer interacts with the ad or not and then it goes away.  You get what you pay for.  In social, one interaction between advertiser and consumer can be transformed into hundreds of relationships with consumers via sharing and liking; favoriting and friending.

The question is – what does all that even mean without true measurement of the bottom line effect?   Your product’s facebook page may have 150 fans and may have been shared 300 times, but did it result in sales?  If a page is shared, does anyone buy?

What’s tragic here is we could have been able to measure all of this. Facebook could have enabled measurement within the Facebook advertising UI of the true effect of every ad purchased.  Not just the impact of users who click on the initial ad, but the downstream effect of all of that sharing and friending and favoriting.

Think about it.  We could have been able to determine if friends that user shared the ad with converted. We would have been able to track if people who became fans went on to perform an action on the advertisers site, whether they did it through a link on Facebook or just typed in your URL and went there.  Social advertising, which is now a bit of an “art” could have become more of a science and the quants would have come rushing in and begun optimizing Facebook ads to maximize their downstream social impact.  What fun!

Wouldn’t that be amazing? And incredibly compelling?  Couldn’t that have moved a large number of advertisers to spend in social?  Could social have been the new search?

But no, we won’t have that.  We’ll have to continue to live in the darkness.

Facebook, please reconsider this. You won’t let us use 3^rd party ad tags in your system, so we can’t do this ourselves. Even if we could, we would not be able to get to the granularity you can give us.

And if you won’t try again, there is one more chance.  There is another dominant social network where people share things and brands want to “talk” to their customers.

Hey, Adam Bain, you’re new at Twitter, please add conversion tracking to Twitter ads AND please go all out and allow marketers to track events after every type of interaction with a brand.   Events after tweets, retweets, replies, reads, everything!! Now that would be awesome.   The quants would rejoice and the dollars would flow in.

Google Instant is cool.  But here’s what I want.  I want the ability to measure the impact of search suggestions on subsequent action.  Basically, I want view-based attribution on instant search suggestions.

What I mean here is that if I type the first letter into Google and a major brand pops up as the suggestion – will I then later search for that brand, use that brand, buy that brand?  At a minimum, I’m aware of that brand.  Do the rich get richer? Do big brands start to takeover through liminal or subliminal reminder while we type our searches?  As I type in my search for “bars east village”, flashing before my eyes is the results for “Bank of America”.  Will I now be more likely to use BofA?  At a minimum, I’m more aware of them, right?

By the way, despite the fact that Google Maps ended my usage of Mapquest, I think it’s kind of nice of Google to let Mapquest appear when people type “M”. I would prefer it to be “Mannino”, but Mapquest being there is a nice bone to throw to the dog whose ass you just kicked. And Mapquest remains the suggestion until I get to “maps” at which point Google Maps takes over.

Back to the main point.  We need a way to measure this impact.  Let’s call it “Instant Search Attribution.”  What’s most interesting here is that, in this case, search is becoming more like display(!).  Think about that for a second.  All along, we’re trying to make display more like search.  To make it perform as well, to make it as easy to manage, to make it as automated and algorithmic.  And lo and behold, Google does something that makes search more like display.

This means that search could now have all the same challenges that display faces.  The biggest of these is attribution.  Many search people will tell you that the click is the most important event and is all that should be measured.  It’s somewhat self-serving but it’s no more self-serving than display people who say “clickers are bad” simply because there is some click fraud out there and display struggles to generate clicks.  And let’s be honest, everyone is right – clicks are most important in search, and not that important in display.

But now, suddenly, clicks might not be the most important thing in search.  Well, not really, but at least clicks have lost a little of their lead and other events (eg, impressions) have gained a little ground.

So what we need is a way to measure this.  Only Google could enable this since the first event happens on google.com.  And the answer is not for Google to just enable this within Dart or Google Analytics.  Google proclaims they want an “open and inclusive” display ecosystem.  Plus, they obviously very much want to grow the display market.

There needs to be a way for search advertisers to drop a pixel when their ad shows up as part of a suggested search and an impression is counted.  The pixel should be able to be from any Google certified ad server or pixel server (remember, the open ecosystem).  This would allow advertisers to evaluate the value of being tops for certain keywords and terms.

This would increase spend in search. But it would do something more. It would begin to shift the mindset of search buyers towards recognizing that impressions without clicks do have a value.  Suddenly, view-based attribution would mean something to the search marketer.  This would do wonders for the convergence of search and display and result in more dollars going towards both.

Returning to the subject of how platform-based buying can help publishers.  Last time I wrote that proper attribution and the elimination of user-duplication would benefit publishers despite being provided by buy-side platforms.  But it’s a little unsatisfying because both of these solutions are being built for the buy-side and buyers will dictate how they develop and evolve.  They only indirectly impact publishers.  Yet buying platforms are building tools that can be directly used by publishers to get smarter and monetize more effectively.

One of the biggest challenges facing publishers today is that direct sales forces still think they are selling the site.   Sorry, sales guy, the world has moved on.  If you’re a salesperson for a website you are selling (and your prospects want to buy) the various audiences that comes to that site, whether that audience segment is related to your site or not.  There are a number of ways to sell audiences that come to your site and probably the least creative and most challenged in the future is to just sell the site itself.

Unfortunately, most publishers have near zero visibility into the many audiences that visit their site.  Advertisers, on the other hand, appear to know everything about what they are buying (or at least think they do).   This instance of asymmetric information drives a lot of the tension in the advertiser-publisher dynamic.

But the same tools the advertisers are using to make these decisions can (and are) rather easily be made available to publishers.  These tools give publishers visibility into their audiences and allow them to better sell and monetize their inventory.

Let’s walk through an example.  Kozmo.com is a publisher and sells out 60% of its inventory at $5 CPM.  The other 40% goes for $1 via networks and exchanges.  That 40% is either selling uniformly at $1 or, more likely, advertisers are looking for specific audiences they like and paying more for them while letting the rest is going for under $1.

To oversimplify for the example, let’s assume there is only one audience segment that buyers want – Cat Lovers, of course – and buyers are willing to pay $10 cpm for it, regardless of the site the user is on.  We’ll also assume Kozmo.com has nothing to do with cats.

Bringing back Mr. Omniscient Marketer from the prior post, he can see that 20% of Kozmo.com visitors are in the Cat Lover segment.  However, because Kozmo.com is so focused on selling the site and doesn’t know which of its visitors are Cat Lovers, ¾ of that audience is already sold as part of the direct sales efforts for $5.  Can someone say “NIGHTMARE!”  This is tragic for everyone involved.  The buyer cannot get the audience it wants, the publisher is actually selling impressions at $5 when it could have gotten $10.  And the Cat Loving consumer cannot learn about the latest in self-cleaning litter boxes(!).

But there is a solution. Give the publisher the same visibility into the audiences on its site that buyers get.  This creates a whole new character – it’s Mr. Omniscient Publisher!  WooHoo!   Platforms (like my own MediaMath) can leverage extensive data partnerships and integrations to provide publishers with a view of who is on their site.  Publishers can then make decisions about what audiences to sell directly and what should be placed on exchanges or with networks.

Let’s rebuild the revenue line for Kozmo.  Obviously, $10 > $5 so Kozmo will want to sell the Cat Lover audience first.   Once that is sold, the original direct sales force inventory can be sold at $5.  Yes, there will probably be a little bit of loss due to frequency caps and buyers wanting first impressions, but $10 is two times $5 so there would have to be a lot of direct sales attrition to make this a worse deal.  Let’s assume that instead of selling 60% of total inventory, the direct sales force can now sell 55%.   The rest goes to networks and exchanges at $0.75 cpm (lower because cat lover buyers are now buying direct vs being blended into the remnant price).  Overall eCPM went from $3.40 (60%*$5 + 40%*$1) to $4.94 (20%*$10 + 55%*$5 + 25%*$0.75).  Big win!

Admittedly, this is an overly simplistic example. The % of visitors from an unrelated audience segment is exaggerated as I would imagine the % in any one segment rarely approaches double digit percentages.  It’s important to point out that even if the segments are not big enough to sell directly (and they likely are not) the publisher can still benefit from audience intelligence. This intelligence can either help it negotiate with networks or it can be used as an input when using a Supply Side Platform like Admeld, Pubmatic or Rubicon (the SSPs have tools themselves which serve this need).

The CPM’s are also a bit extreme to prove a point and make the math easy.  The point is that if publishers know more about the audiences on their site, they can make better decisions and monetize more effectively.  With greater intelligence, direct sales teams can be better informed when they approach advertisers, get off their heels, and serve their advertisers better. Data will stop being a scary word to publishers.  Relatedly, direct sales teams can take the offensive when they find advertisers buying their site via indirect channels (a subject well documented by Ryan Maynard of Pubhelix here )

So look at that, with just some simple insights we radically changed the publisher’s ad stack (at least philosophically – no more “direct always goes first”) and increased revenue by 45%.  The next step would be to allow platform-based buyers to optimize on performance and pick and choose individual impressions out of all of your inventory (not just what is deemed “remnant” and thrown on the exchanges).    With that, publishers may find even more incremental eCPM lift.  But since I was told my last post was a bit long, we’ll leave that for another day.

We’ve heard it all.  Oversupply.  Glut in inventory.  Commoditization.  This exciting new era in digital marketing will be the death of the publisher.  Providers of quality content simply won’t survive.  All people care about is pushing for a lower eCPM and driving the cost down.  Yadda, Yadda, Yadda.

I go to a lot of conferences and attend a lot of panels (basically I don’t turn down free beer) and representatives of the buy-side always seems so charged up while the sell-side looks like they woke up to learn they managed the Pittsburgh Pirates (=====>>).  But it doesn’t make sense.   Digital spending continues to grow, budgets continue to shift online, and display spending is predicted to grow at a strong rate over the next five years.   Meanwhile, the number of people online has plateaued.  More dollars divided by same number of users should equal rising revenue and profits for publishers.  Right?  Well, actually, no, that’s not what is happening.

Publishers are seeing downward trends in pricing and are finding that buyers don’t actually care about them or their site, they simply want to find a specific audience and, barring porn and malware, they don’t care how or where they get it.  And the chorus screams “Commoditization!”

I don’t actually believe there is commoditization of display media.  And while the symptoms of it do exist, I believe this can and will be solved.  Commoditization implies that all individual units of a good are the same and are capable of mutual substitution.   I’ve seen site level performance reports and the data simply does not bear out that all media is created equal.

Rather than commoditization, what we have here is insufficient tools for buyers and sellers to really value impressions.  It all looks the same, or at least I can’t figure out how it’s different, so I’ll assume it’s all the same and not worry about it.  That’s how buyers are thinking.

This is where buying platforms come in. Platform based buying will benefit media sellers and publishers as much, if not more, than it benefits advertisers.  This seems odd and totally wrong, right? The true savior of the sell-side is actually focusing on the buy-side?  Yes, because buying platforms solve many of the issues that are causing this feeling of commoditization and this experience of ever-decreasing prices.

What qualifies as a buying platform?  It’s really any entity that can actually take control of and, more importantly, provide actionable insight into an entire media plan.

Buying platforms solve problems facing both buyers and sellers.  Let’s first look at the challenges that buyers face and dig into how platform-based buying solves, or at least minimizes, these issues.

Duplication

First, there is massive duplication of users across channels within media plans.  Currently, agencies and advertisers buy display media from a number of sellers.  The magic number for the agency seems to be about 10 or 20.  More than that and it’s simply too hard to manage.  Less than that and there’s both limited reach and also a perception that you’re not really doing enough.

What’s the problem with this duplication?  Let’s walk through an example.  Assume that Mr Omniscient Marketer has determined that the optimal number of impressions to get someone to buy is 3 impressions per user per week.  In this example, analysis has shown that less than that simply won’t generate the awareness needed while more than three will simply be wasted money since the user will either have already been convinced to buy or will never buy.   The marketer goes out to 10 sellers (networks, publishers, whatever) and does 10 separate buys and implements a frequency cap of 3 ads per user per week on each.  The campaign runs and what invariably happens is that a large proportion of users see far more than 3 ads per week – because they are addressed by more than one of the channels used by the marketer.

An obvious solution to this problem is to give each channel a frequency cap of 1, because then, theoretically, this will reduce the number of (and cost of) over-addressed users.  Unfortunately, this also increases the number of under-addressed users, which is just as bad.

Solving for this is challenging for the marketer, and they know it and that is why the holy grail for the display marketer is universal frequency capping.  We rarely get in a conversation with an agency where this topic doesn’t come up.  But this is not new and if you’re reading this you’re likely in the choir I’m preaching to.  Most discussions about exchanges, platforms, DSPs and networks all talk about the value of universal frequency capping to the advertiser.

What is new is that the lack of universal frequency capping may be even worse for media sellers.  Counterintuitive?  Yes, a bit.  Why would sellers want or need something that limits the number of impressions bought?   Wouldn’t this exacerbate the oversupply problem by decreasing demand even more?   The answer is… it’s complicated.

The buy side controls all the data and dictates how much they spend, at what price, at what goal, etc.  So the fact that far too many impressions are being served to some users is already factored into the buying process and economics.  They simply reduce their target CPA and CPM.  Buys aren’t “optimal” but it doesn’t really matter to the buyer since they are optimizing under these constraints and making it work.  Continued shift of spend to digital shows that advertisers have largely accepted this and are moving in.  But it’s painful for publishers.

Let’s put a little math behind that.  If a customer is worth $100 and we expect to convert 0.01% of the people we address (which is about right in display) then we can afford to pay $10 per thousand unique consumers.   If we have an optimal frequency cap of 3 impressions per consumer, we can afford a $3.33 cpm.  However, if we actually serve 10 impressions per consumer, now we can only afford $1 cpm.  If we serve 20 impressions, we’re down to $0.50.  And so on.

Now take that math and add in a few more complicating factors which cause smart buyers to become even more conservative with how they value impressions.  First,  add an uncertainty adjustment by the buyer because they simply do not know how many impressions they are serving to each users (so they assume the worst and overcorrect on the high side) – this lowers allowable CPM.  Second, take in the inherent difficulty with attributing display conversions due to last-click attribution and known inability to measure the true impact – this lowers the CPM further.  All of this results in buyers being unwilling (and unable) to pay a high price for impressions.

Platform based buying actually solves this for sellers.  Universal frequency capping is difficult across display channels.  It requires integration into the buy decision (eg, real-time bidding or, a better name, impression-level bidding). Without that it can result in a great amount of exhaust (extra impressions you’ve bought but don’t want to serve). However, true frequency capping isn’t really necessary.  Aggressive frequency management should be sufficient to have a significant impact on “right-pricing” display media.  Frequency management comes about when the buyer has insight into frequency, reach and duplication and can then use available levers to optimize media channels.  It’s not an absolute cap but it gets the job done.  For example, if Network 8 and Publisher 12 are not adding unduplicated or under-addressed users, they can be dropped from the plan and the dollars reallocated to Network 3, 4 and 5 and Publisher 7 which are reaching under-addressed users.

So now we’re buying less impressions at a higher price.  But the dollars coming into the system didn’t grow. So the problem for the publisher isn’t solved yet.

Attribution

Our goal isn’t to make all publishers rich. It’s to halt this feeling of commoditization and to enable publishers that are actually driving results and adding value to be compensated for that value.  This is where attribution comes in, because the challenge for the buyer doesn’t stop at frequency; correct attribution is also critical.  Though it has a huge impact, let’s ignore cross-media attribution issues (eg, display vs. SEM vs SEO vs. organic vs. offline vs. word of mouth, etc).  Here let’s just focus on attribution within the display media plan.  We want to make sure that each marketing channel is getting credited with and rewarded for the value it actually adds.  This actually is not difficult, it just requires the right tools.

Poor attribution drives poor decisions and results in dollars being allocated to the wrong channels.  A baseball analogy to poor attribution would be evaluating a starting pitcher based solely on Wins and Losses while ignoring the fact that it’s much harder for a pitcher on a bad team like the Royals to win games than it is for one on a good team like the Yankees or Red Sox.   This issue affects all types of publishers and networks.  Smaller networks or publishers may actually be driving good results and have a lower eCPM but may not have the reach to drive results in the manner that looks good to a poorly designed attribution analysis (ie, they are pitching well but they simply don’t have a good Win-Loss record, which is how they are being measured).  Similarly, premium sites may be driving results and be cost effective even with their premium price, but poor analysis causes them to look bad and to have to drop price.

The solution is to evaluate all your display media buying through one lens and while most ad servers can provide lenses into attribution, they do not allow the buyer to execute against that insight.  Platform based buying with superior insights informs the buyer who is doing what and what they are worth.  This insight means that display impressions are no longer a commodity, they are no longer undifferentiated.  A buyer can truly evaluate what works and what doesn’t and pay accordingly.   Suddenly, the medium-sized network or the premium financial site is hearing positive feedback, the advertiser is asking for more volume and prices rise.  Before you know it, we might even have a charged up publisher on a panel.

Next:

We’ll talk about the challenges facing publishers themselves and how the tools buying platforms are building can be used to help publishers compete and thrive.

Eyeblaster released a research note -  “Search and Display: Reach beyond the keyword” (get it here).   It’s a brief note with very interesting findings, particularly for display folks.  Using Eyeblaster’s proprietary “Channel Connect for Search” (CC4S) product, Eyeblaster tracked cross-channel campaigns and determined the source of the conversion.

The note starts with a really good overview of the purchase funnel as it pertains to display and search (using a very relevant analogy to the yellow pages).   The funnel diagram shows how display and search can impact the funnel.  In the diagram, Display is shown to drive the entire funnel from Awareness to Intent to Purchase while Search only drives Consideration and Intent.   I’d suggest that Search can and does influence “Favorability”.  Once a consumer is aware of the need, she will search for generic, higher-level terms, and search can show ads that will drive brand favorability.  For example, I might search for “BlueRay DVD player” and could then see a search ad for LG and view LG DVD players favorably.  Additionally, while it isn’t clear if they mean this in the diagram, display is proven to drive repeat purchase (via retargeting).  Search is unable to directly drive repeat purchase.

As to the results of Eyeblaster’s study, it was found that:

“Overall, for customers who used both search and display, 72% of conversions arrived as a direct result of the displaychannel. Only 23% of the conversions were a direct result of the search channel. 5% were the result of display ads that were followed by a search.”

Eyeblaster caveats that the share of display vs search is “the result of  budget allocation decisions made by the advertisers.”   It’s probably safe to assume that Eyeblaster clients lean a bit more towards display than search, particularly those using CC4S.  The extremely low number of “search after display” conversions differs pretty dramatically from prior studies I’ve seen, which I believe has to be due to the types of campaigns being used in the study.

The report then digs into the share of display and search conversions by vertical. This is great data and you can see a wide variety in share by industry vertical.  Entertainment, CPG, Financial and Careers lean far more towards display while B2B, Travel and Retail skew towards search.  I’m actually surprised to see careers sitting more towards the display side, since jobs seems like a much more search-driven category, but the data tells a different story.

What would be really interesting here, given Eyeblaster’s leadership position in driving display engagement, is to get a sense of the breakout of display conversions by attribution type.  How are these display conversions being attributed?  It would be great to break these conversions out into post-view, post-click and even post-engagement.  I’m petitioning for a follow-up Research Note.

The key takeaway here is that display is an extremely important part of any marketing plan and can drive conversions on its own.  We in display intuitively know that and evangelize it every day, but it’s great to see a study from an industry leader showing some numbers to back it up.  Next step is to dig even further into these numbers and really nail the point home.

Could there be any more compelling case that display drives search than the fact that Google is running display ads to drive to Youtube videos about Google’s search capabilities?

The videos are on Youtube here – http://www.youtube.com/searchstories

John Battelle first talked about this in his SearchBlog.

Most relevant here is that Battelle saw Google doing roadblocks on nytimes.com.  He states:

“It’s truly a brand campaign: Google is not selling anything here other than its own brand – that ephemeral sensibility that resides between its customers’ ears.”

Now we know that every campaign, even a “brand” campaign, has a goal.  So what’s the goal here?  How would the dominant search engine with 65% market share in the US measure and evaluate success?

Since I haven’t seen screenshots of the display ads, I can’t evaluate if there is a chance the display ads could drive view-based activity (vs. focusing solely on driving clicks).  But the primary video on the youtube site , called “Parisian Love”, has nearly 700K video plays in two months.  I am no expert on how many video plays is impressive, but that seems like a pretty good number.  Side note – these videos are pretty cool – I’m immersed in the space and I find them compelling yet I could see my mom or sister also finding them interesing and informative – nice job!

It’s interesting to think of the way Google could have used data to drive results in this campaign.  First, assuming the goal is to increase search share, and in an interesting case of reverse retargeting, they could have simply cookied their own searchers and either excluded people who either never or rarely searched on Google.  They could also have cookied people who didn’t search in an optimal manner (these videos do a good job of demonstrating the lifecycle of searching for things and coming to solution) or those who didn’t seem to find what they needed.  Finally, while they could never do this because they “do no evil”, they could have cookied people who have the Google toolbar but searched on Bing.

Regardless of how they are measuring success – this campaign is the ultimate example of using display to drive search.  If the master of search is using display to drive search, who are you to argue? Thanks Google.

Took a break from blogging to welcome a new addition to the family.  Little Jackson, our newborn son, is doing very well and, as you can see, has a very excited big sister.

I hope to get back to a more semi-regular schedule with the blog.