Are CPMs Rising? Will We Hit $19.5B in ’07? How to Spend and Get the Biggest Bang for the Buck. It’s in the Technology You Use.

Are CPMs on the rise?  Is it getting more expensive to advertise online?  Are more advertisers coming online, competing for exposure and driving up the rates?  Will the online spend projections for 2007 be reached or even exceeded as a result of more online advertisers or as a result of publishers driving up rates in response to the demand?  How do you take advantage of available technologies to become more effective with the media you buy to keep your ROI in check, your CPA down and your hair on your head!?!


Did you read the NY Times eCommerce Report today?  Ad Costs on the Web Are Rising, but Perhaps a Bit Irrationally.  Are they?  Haven’t they been for a while now?  But wait a minute.  Are we talking about rate card here or are we talking about actual deals.  Maybe both.  The deals are getting more complex.  The media plans, more comprehensive – web sites, networks, search, email.  How an advertiser leverages the technologies that each medium within the online space makes available can dictate how effective those dollars are spent.  But watch out!  Differentiating technologies that improve performance can cost and the incremental cost can blow your return.  So let’s look for what works.


The NYT does a good job of presenting a few different sides to the picture, because the reality is that rising rates is happening where the pressure is strong and the reverse is happening where it is not.  Where traffic is growing at a pace that is out-stepping demand (like with video) the price is falling.  But the premium pages where everyone wants to be, which continue to have increased traffic, is experiencing rising rates.  Search definitely has rising costs (see my previous post, Banners vs. Search) like on Google.  As more and more advertisers include online in their budgets, the rate pressure increases.  So yeah, rates are getting squeezed from that angle too.  Read the article if you want more examples.


So first let’s look at what you are aiming to accomplish.  Are you branding or going for direct response?  Promoting a message or generating leads, customers and sales?  If you’re direct response, keep it simple.  Email advertising drives actions but with low response rates so test, test and re-test.  Either do it in-house or enlist someone like Exact Target to do it for you.  Keep the budget in check and don’t have expectations that are too high.  Best thing you can do is reserve most of your email advertising as direct marketing to existing customers and then test-market prospecting to reputable lists.


If you read my blog, you know how I feel about Search.  It performs really well in the near term but can quickly grow to be less and less effective over time.  The most successful campaign will quickly become the most expensive if you don’t know what you’re doing or if you leave it on auto-pilot.  So you have to stay on top if it and know that you can’t stay with the terms that work for you now, long-term.  The rates will climb as they perform so set your thresholds and when you hit them, dump the terms and move on. 


Using a SEO provider will be helpful if you are buying in volume.  Ad servers that provide SEO integrated with their tools are a nicety since the reports are integrated with the banner campaigns, but remember you are getting a service that is outside of the wheelhouse of what ad servers do.  It’s always best going to people who are working at their core competency.    SEO is a technical science that is still rapidly evolving (believe it or not).  Google has changed the Adwords pricing model so now it costs more to keep checking the bids, so technically-advancing SEO companies will be gaining ground faster than ad servers for whom SEO is peripheral to their business. 


Banner advertising is a foundation of a campaign.  Now we can look at both Direct Response and Branding campaigns.  If you’re running direct response, it’s CPA all the way or remnant network inventory at remnant-priced CPMs.  Remnant inventory should not be getting more expensive.  More people going online every day and spending more time online every day means more inventory so buy aggressively. 


If you use an ad server with DR, you will gain the ability to A/B test creative and messages and optimize your campaign.  In turn, you will generate more leads from the networks and sites.  Obviously you’re not going to go with an ad server if you’re buying on CPA alone.  The ad serving costs will be astronomical.  But if you have CPM buys you should run them through the ad server and then run the same creative through your CPA buys. 


Try to always have at least some CPM portion buy even if you are a Direct Response advertiser so that you can be testing and improving your creative.  The reason for doing this is that networks who are delivering to you on a CPA are optimizing their inventory.  As your performance drops, your ad play drops.  You won’t even know when or why it is happening or which of your banners are producing the decrease in performance of plays vs. click-thrus but your lead flow will diminish.  So have a place to be testing creative so you can float optimized creative through your CPA buys.


With branding campaigns you are obviously more apt to be buying on a CPM basis.  Even if you are looking to generate some level of response, but not a typical Direct Response campaign (e.g., lead generating) you may be buying more premium inventory.  In this circumstance, you really need to be looking at the available technologies because you are entering the realm of rising CPMS.  You are going to be the most concerns with diminishing ROI.


So let’s look at ad networks.  Let’s look at behavioral targeting.  Let’s look at customer re-targeting.  Let’s look at campaign optimization and other forms of targeting.  Let’s look at storyboarding.  And let’s look at other ways you can leverage the ad server cookie file.


Before we jump into all of that, when you buy on premium sites, if you are not paying close attention to your campaigns, reporting frequently, rotating creative and optimizing campaign performance with a competitively priced ad server you are simply wasting your clients’ (or your own) money.  Ad servers are designed to optimize campaign performance so if you are buying in an environment wherein CPMS are rising, negotiate solid, competitive ad serving fees that decrease as your volume increases (no fixed CPMs people) and use the hell out of the ad server to maximize your campaign performance.


Ad servers provide a host of targeting and optimization capabilities like day-part, geo, storyboarding, limits and cookie-targeting.  Know what your ad server can do and leverage these technologies because they don’t (or shouldn’t) cost any extra.  Turn to your ad serving partner (not vendor) and express your issue with rising costs and get them to help you maximize how you use their product.  Don’t let them charge you for the training you need to become a more proficient user of their tools.  Optimize your campaigns so that you are maximizing your return.  Storyboarding can limit the frequency of an ad –play to an individual.  Great.  What about other ways to leverage the cookie file?  Can you define data in your ad server’s cookie for additional targeting?  We do it all the time. 


Ad networks should allow you the opportunity to see site performance data.  They may not give you performance on all sites, but you should be able to dive into the top-X performing sites.  When you buy on a CPM, get access to the data and maximize your exposure by managing your campaign. 


You have two choices with networks.  You can let them serve your creative or you can use an ad server.  If you use an ad server, you maintain control over creative optimization and you leave site optimization to the network.  If you let the network serve creative you are entrusting both to the network.  Do you have an ad server in place?  If you do, than you should be using it to optimize your creative on the network because the networks are (whether they admit it or not) ultimately optimizing the creative-site play combinations in a way that optimizes their inventory usage.  You can improve performance overall by managing your creative rotations yourself.  Then you can apply pressure on the network to optimize site placements.  Look at the reports you get from them on site performances and start culling sites that don’t work or negotiate a variable CPM for tiers of sites based on performance if you start to recognize a pattern.


Behavioral Targeting.  We talk a lot about BT.  Network BT is different than ad server BT.  TACODA or charge an incremental fee for BT but it is not super significant and it does improve your performance on their networks.  It is a technology that is worth taking advantage of for prospecting new people on the internet.  Test it and measure a campaign with it and without and you will be able to determine if it is right for you. 


You shouldn’t be seeing too much price pressure on the networks, at least not in 2007.  I say this because there is such a surge in the number of new networks that competition will be putting pressure on their prices.


Ad server behavioral targeting like DoubleClick’s Boomerang is an entirely different story.  The difference is that DC BT is looking for the DC cookie anywhere on the internet it can find it based on the same cookie/pixel combination that a TACODA or deploys.  They pixel the advertiser’s site, wait for events on the advertiser’s site such as certain actions like page views of products or purchase/thank you pages and then cookie the user.  If/when they encounter that user on the internet, they recognize the cookie, associate it back to the event(s) and allow for targeting of an ad based on the anonymous event.  Great conceptual technology but expensive.  DC is already a premium-rate ad server so when you tack on Boomerang you are looking at high fees.  Couple that with rising CPMs on publishers and you are quickly looking at the potential of a negative ROI.  You need to look at this very closely.  Many big-block advertisers with exclusive contracts with DC, or agencies with exclusive DC contracts, don’t use Boomerang because it has proven to not be cost effective.  Ask for their client list and you will get their biggest names.  Then ask which ones are using Boomerang and you will see what I mean.  Do your own analysis before you even bother testing it.  Atlas and Mediaplex too it’s the same story.


Customer Re-targeting is similar to BT, only it involves the reading of an advertiser’s first party cookie rather than the ad server’s cookie.  TruEffect, for example looks for an advertiser’s customers who are tagged with the advertiser’s cookies, which indicate a customer segment (such as shopping frequency or buying habits or preferences, etc.) and then allow for the ability of that advertiser to target that user accordingly with an ad through a campaign anywhere on the internet at any time.  This technology has no incremental cost in terms of the ad serving so it is a benefit of working with TruEffect over other ad servers.  Direct Response advertisers can use it to drive recurring revenue opportunities from existing customers.  They can up-sell, cross-promote or highlight products or services to individuals based on known shopping preferences rather than re-prospecting an existing customer through an advertising campaign.  Customer re-targeting is a great campaign addition to be used in combination with something like TACODA or prospecting BT.  This is an example of a technology that will drastically improve performance without increasing your costs, effectively increasing ROI.  DirectServe, as it is called, can also be integrated with Search campaigns as discussed in Search and Networks: Better Together – I Think So


So there you go.  CPMs may be rising.  We can see that the pressure to drive up CPMs represents an increase in the number of buyers (demand) and publishers seeking higher rates for their products will push the 2007 ad spend up and probably over the projected $19B mark.  We need to be well aware of how we allocate our budgets and that we are taking advantage of the technological benefits out there that will help us maximize campaign efficiencies so that we can keep our overall costs in check to maximize the ROI.  Costs are rising, ad serving rates are steady or even dropping if you are buying more inventory as an agency overall.  Don’t get sucked-in to value-added services that chip away at your ROI without some complimentary tests that will prove their value.  Vendors do want to demonstrate their value to you and the pressure you will be under to maintain ROI will mean you will have to get more aggressive in your negotiation with vendors.  Good luck. 

_uacct = “UA-980395-1″;

Ad Servers Add Video Serving Capability, Advantage-Buyer

In the race to stay relevant, DoubleClick, Atlas, Mediaplex (ValueClick) and TruEffect have all brought video ad serving to their arsenal of rich media ad serving.  DoubleClick made its November 6th splash with their media announcement following Atlas’ announcement a week earlier.  ValueClick followed-up with their announcement on December 4th and TruEffect released their video ad server to its clients at the end of the third quarter.  All four ad servers now promote their video ad serving capabilities to existing and prospective clients. 


So video is now another form of rich media that ad servers like Zedos, Poindexter/[X+1], Adjuggler and other tier-two ad servers will likely catch-up on with shortly as well.  The bigger question, however, is how many advertisers will use this capability in 2007.  We have all read the reports and seen the projections that video is the next big thing.  And it will be.  But how soon and how fast is harder to predict.  The cost to produce video ads is significant in comparison to other forms of rich media.  And to create multiple versions of creative to promote campaign optimization will result in production costs that will break some advertisers’ bank.  Video advertising in 2007 will likely be reserved for only the big-budget advertisers or those leading-edge early adopters who want to throw money at the newest technology because their target audience is going to be receptive to the new media (younger demographic, YouTube, MySpace, etc.).


But in the meantime the ad servers have brought video ad serving to the market and have added it to their offerings.  Pay attention.  The cost structure of video ad serving is entirely based on the bandwidth that is required to serve a video ad.  A properly priced service should correlate to the size of the files.  When you buy traditional ad serving, like banners, with file size limitations of 20K vs file size limitations of 40K you should see better rates.  If you are adding video file size limitations of 100K or 1MB, you will see it in better CPM rates.  When you negotiate your fees with the ad server, look at the bandwidth limitations you have and try to calculate the incremental impact that the file size is having on your proposed rate for video ad serving.  Be prepared to see all kinds of premiums thrown in for the newest, greatest thing – video ad serving – from the likes of DoubleClick and Atlas (probably Mediaplex too) so buyer beware.  But consider them ‘junk fees.’


A knowledgeable buyer should start with their current ad serving fees and look at the file size limits and negotiate based on the increase in file size limits that you will require to obtain video ad serving.  Don’t let the added fees get dumped on you.  The ad servers want to see these new services get used as badly as you want to use them.  Maybe more so.  Early adopters have buyer advantage.

_uacct = “UA-980395-1″;

Networks, Networks Everywhere

A wonderful thing about the holidays, the days and weeks leading up to them are like a rat race.  Working to get more done than possible with less time reasonable to do it with.  But when you get to the finish line, there is literally nobody to around to get any more work done with.  So I get to catch up on the reading, the blogging, the planning and all of the fun stuff that I have been putting off for the last quarter or so.


The first one that caught my eye was Jim Meskauskas’ article in iMediaConnection: Networks, Networks Everywhere.  Jim is colorful writer and I always enjoy reading his articles.  I wish I had not missed this one when it came out last month, as it had some great fodder for a conversation that I shared through blog posts with Dave Morgan as well as some other posts that I made on this blog.


Jim characterized the boom of networks over the past couple of years in a light as a love of something that will likely break.  “You can have your ads run in some great properties and, more importantly, reach great audiences. And it won’t cost you too much.”  Networks have always been known for the cheaper route to some potentially solid inventory.  More recently (as in years), some networks have turned on some technological advances like targeting (yes, BT) and content-specific selection that make them even more useful. 


You do have to weed through networks inventory to make sure that you are getting the goods.  As Jim said, “The web’s worst kept secret is that even the most popular sites have more inventory than they can sell… and so it either runs as bonus, a make good or ends up in a network.”  They keep the best valued positions for themselves and run the ‘remnant’ through networks.  But is that necessarily true?  Some networks like Tacoda or have exclusivity with a site and may get certain premium positions. 


Jim is also not shy about the fact that “enthusiasm to spend online doesn’t mean that all places become viable vehicles for advertising. Just how many ad networks can the marketplace bear?”  So therein lies the problem that I discussed in my post: Ad Networks and Advertisers.  How does an advertiser know which networks to go with.  At what points do they start just advertising on networks and stop advertising on sites altogether?  Okay, it’s the polar extreme but is it a trend?


Jim writes: “If as an advertiser I’m making a reach play, I risk burning out an audience with too much frequency even if the buy made was with multiple vendors. If as an advertiser I’m making a DR play, I risk wasting multiple exposures against too many disinterested members of the audience.”


Ad Networks and Advertisers I also talk about this problem.  Getting multiple exposure from different ad networks will throw off your balance of exposure.  Not all networks will let you select the publishers, positions, content.  Many won’t even tell you all the sites in the network.  So does network advertising have the potential of elevating above the board from remnant?  Will the value of branding ever extend beyond just a few networks even through more an more are coming online?  Jim asks what will the market bear?


Dave Morgan’s Ad Network Resurgence in MediaPost talked about the boom of ad networks over the last couple of years.  With his perspective and experience over the years he was very positive and saw the increase as a good thing for the industry.  Like Jim, I raised a concern about pressure on the market.


I asked Dave, “As to the idea that there will be more and more networks popping up, where does it end Dave? How do you keep differentiating so that a buyer knows the best place to allocate funds? Does an advertiser have to pick from Tacoda or for behavioral targeting? Or buy from both plus others to cover all bases as more pop up? The more networks you add the more diluted it becomes right? Don’t you think that a consolidation becomes more necessary? … I only ask because it seems like the more networks you have the harder it is for an advertiser to decide where to allocate funding.”


Dave didn’t respond to my inquiry on his blog, although he and Tom Hespos had some fun in follow-on posts.  For me, I am just trying to get to the bottom of the same questions as Jim.


In an market environment when you have too much pressure in one area, you have a reactive pressure in another area.  A ‘snap’ if you will to balance out the pressure.  It can be slow or fast but it happens.  In the case of the network boom, it is likely that a tipping point is approaching.  But what will that represent?  A consolidation?  A devaluation of inventory?  Will advertisers begin to withdraw from networks because of the overlap of audience?  Will publishers begin to apply pressure on the networks and drive up the CPM they demand for their inventory because of the competition in the space?  Something is going to force the over-abundance of networks back down when that point is reached.  Who will be left standing?  Many of us can guess on a few of them.  And even I am not bold enough to point fingers.  But the tipping point is definitely coming even if some of the most accomplished and experienced people in the space don’t want to talk about it.

_uacct = “UA-980395-1″;

Ad Networks and Ad Servers

The iMedia Agency Summit has prompted a lot of great topic conversation and some great articles that are worthy of comments, reactions and insight. 


A number of clients that we work with come to us with the conundrum of working with both web sites and networks when they advertise online. 


Dave Morgan has professed the rise of ad networks and he is not alone.  Tom Hespos recently wrote how the Brand Value of Ad Networks [is] on the Rise.  There is no doubt that the ad networks are stretching their wings and are trying to move away from being simple aggregators of remnant inventory.


Ad Networks are offering advertisers behavioral targeting.  They are enabling advertisers (at least some ad networks) to customize campaigns by selecting sites or at least site groupings for campaigns when buying by CPM.  Some ad networks are developing hands-on programs to promote branding strategies for advertisers that help the network move away from the pure CPC, CPA model and offer up the opportunity for greater revenue diversification for the network with a better value proposition to the advertiser.  In the end, the networks are trying and with time we’ll see who rises to the top.  See my post on consolidation of networks: Where Behavioral Targeting and CRM Meet, and Where They Can Marry. 


With time we will also see whether the savvier agencies buy into the idea of networks being able to really offer branding and awareness-capable campaign value or if in the end it is about direct response.  Time will tell but I think that the Summit produced some great arguments and there are some good discussions that came out of it.  I’ll tackle some more of them shortly.


Tom Hespos argued in his article that site duplication, when working with multiple networks, is one issue that you need to watch out for in managing online campaigns through networks.  Great point.  He also said the following which was the impetus for this posting:


Another way to minimize duplication is to look at cookie data from past buys. New tools are emerging to look at past campaign data for the purposes of figuring out which networks duplicated the most and for scenario planning with unduplicated reach in mind. Look to your ad server of choice for help with this one.


When we work with advertisers who are using our ad servers, we enable them to take control of their campaigns holistically online, across all the sites and networks they advertise on.  When you work with an ad network you have the choice of letting them serve your ads for you.  No problem there – it’s like letting a site serve your ads only they offer two additional features: (1) They can strategically optimize your creative for you across all the sites in the network upon which they play your ads and (2) they have solid reporting (a lot of them do).  But you can get both of these capabilities with an ad server.  In fact, the reporting with an ad server – by creative – will be far more robust because of post-click analysis and the grand view of the entire campaign (all sites and all networks you advertise for the campaign).  And the ability to centrally manage, rotate and optimize creative across every site AND every network simultaneously will cut way down on the headaches of having each network doing it for you individually.


Looking into the cookie data that an ad server collects for you with regard to your campaigns across ad networks … now that opens up a whole world of possibilities.  Thanks Tom.  If you work with a third-party ad server, what can you find?  What can you see?  First of all, what is available to you?


A third-party ad server covets the data they amass as it is core to the asset they build for profiling and targeting.  The one thing you will not be getting your hands on, is that data.  Ask Atlas for a copy of their log files!  LOL.  However, for a fee, you can get interpretations of that data. 


An ad server can look at the cookie log files and determine publisher, site, site section, banner plus log data like IP, browser type, time of day, etc.  But all of this is historical and can only help you keep tabs on your networks after the fact.  It is a lot of data-digging and will get expensive fast.


Here is another way.  I know that I am always espousing the First Party thing, but check this out … serve through a first-party ad server and use first party cookies.  You can have the ad server write the directly to the cookie: the site, site section and banners played in sequential order in real time.  Then when the lead lands on the advertiser’s designated web page, the advertiser can read the cookie – their first-party cookie – and can see exactly where the lead came from and how the lead was derived.  In fact, not only will the advertiser see how the lead was derived, but sequentially every banner-play that previously led up to the lead-generation. 


When you instrument the tags for the ad server with each network they can trigger markers into the first-party cookie-writing sequence so you will know whether site A came from network 1 or network 2.


This first part stuff, which TruEffect calls DirectServe has a lot going for it.  Check it out when you have time.

_uacct = “UA-980395-1″;

Marrying Behavioral Marketing and CRM with First Party Ad Serving

So in my last post I went through how behavioral targeting generates extensive acquisition marketing data that remains isolated from the advertiser’s in-house data repository.  An Advertiser must rely on the event-based BT network(s) that they advertiser on to collect data for them, then they have to pull that data into a central format – probably Microsoft Excel – to centrally analyze how they are acquiring leads in an anonymous format.  Once a lead is received and/or converted a CRM system can pick it up and it can be tracked by a combination of a site analysis tool and the CRM system.


In my last post, Where Behavioral Targeting and CRM Meet, and Where They Can Marry (XXXXX), I promised to offer up a better way to do this.  Specifically I want to examine how we can marry the acquisition marketing data sources and the customer data sources seamlessly.


I have talked in my previous posts about First Party Ad serving.  See: How Does Re-Targeting Work.  FPAS or DirectServe by TruEffect, enables an ad server to read and write cookies out of the first party domain, the advertisers domain.  So instead of all of the data being generated on the third-party network cookie, the data would be associated with the advertiser’s cookie.  The Ideal Online Advertising Campaign, Direct Response with Behavioral Customer Re-Targeting.  How to Get a Greater Share of Your Advertiser’s Budget extensively describes how a first-party ad serving campaign would be run by an advertiser, including across a network to generate leads whereby cookies are read and written by the ad server on a first-party cookie.


With FPAS the advertiser can still advertiser with BT on all the networks they choose.  No feature or capability is lost.  All we are talking about here is enhancement.  The next generation of technology, leveraging everything that already exists out there but using it differently.


Couple of scenarios.  First is customer re-targeting in the absence of BT (as BT is about prospecting new leads).  The advertiser can use FPAS to advertise on any web site and any network.  When a customer is encountered they are recognized and distinguished from a non-customer and a relevant ad is displayed.  Non-customers are treated like prospects and tradition third-party ad serving is conducted with features like cookie-targeted, storyboarded, day-part targeting and geo. 


When an individual clicks on an ad, if they are a customer, they are immediately recognized by the advertiser’s CRM system with the first-party cookie and treated accordingly.  If they are not a customer, they are also recognized because the ad server has written the first party cookie and associated the acquisition marketing data (all banners seen and all sites that the banners were seen on sequentially).  The advertiser can decide how to treat the new lead and associate the acquisition marketing data with the new record that will be created in the CRM system and treat that individual as a customer from that point forward.


Second scenario is customer re-targeting in conjunction with BT.  BT on networks can be conducted with the pixeling of an advertiser’s site.  The network will use its pixel-technology and cookie to select the advertiser when it is appropriate to promote its BT capability and then redirect to the first-party ad server for the selection of the advertisement which will be based on the recognition of the first-party cookie.  Multiple cookies can reside on the browser at the same time so this will work seamlessly.  The advertiser can work with as many networks as they wish and the First Party Ad Serving, customer re-targeting will work across every network, and every web site that they advertise on.


The end result is the marrying of the acquisition marketing data.


The First Party Ad Serving process writes data to the advertiser’s first-party cookie.  Any prospect that is encountered will have data written to the cookie.  Actually any individual that is encountered can have data written to the cookie if so desired.  The ad server can write which ads were seen and which sites they were seen on sequentially so that the advertiser can associate its online marketing efforts with that individual just like they do with other offline, traditional forms of marketing.


In my post, Maximize Return with a Marketing Model I comment on Charles Haggerty’s iMediaConnection’s article about multi-channel marketing analysis.  First Party Ad Serving brings forth the opportunity to integrate online advertising data with CRM, site analysis data with traditional customer knowledge.  The cross-channel view into an individual customer becomes that much more clear.  The creation of a new class, a new profile of digital customers is created with First Party Ad Serving.

_uacct = “UA-980395-1″;