Empowering a Client or Destabilizing an Agency?

Let’s talk about an advertiser who will spend several million dollars advertising online this year and is working with an agency.  Imagine that the agency is aggressive with their planning and they are good negotiators.  I know, it already sounds like an unusual agency, but its my example.  It’s not a small firm, but also not one of the mega-agencies.  These guys really care about their clients’ performance and they work hard at what they do. 


How does an agency draw the line and not fall prey to being at the beck-and-call of their clients?  Well they don’t.  Bottom line, big clients levy big demands.  But where does the line get drawn?  If every decision ultimately gets passed on to the client, it will seem as if the agency is afraid to step up and drive the relationship.  If the agency is willing to exercise their skills and expertise, take some calculated risks and accept responsibility for potential mistakes, they may not end up being emasculated by the relationship or end up with a client that drives all of the decisions.  Is this realistic?


But where does that line get drawn?  What about when an agency lets an overpowering client impact decisions about how the agency should operate?  This is precarious to me.  A client that knows what it wants is a good client to work with, it makes it possible to get somewhere in terms of ideas and results.  But an agency should know how they operate best.  They should know what technologies and what tools work best for them and what will make them the most efficient at their jobs.  What I don’t understand is how an agency can become so overpowered by a client that it will begin to yield its operating decisions to the client in addition to campaign decisions.  This is happening out there.  I encounter agencies that yield their ad serving decisions to their clients.  Not because of a price decision, but because of other points of evaluation.  I see agencies that are letting their clients drive decisions about what SEM tools to use and what email marketing systems to go with. 


Agencies that are skilled in online advertising should have preferences and experience in this space and should be making these recommendations to their clients.  No, maybe recommendations is too soft a word.   I am not speaking as a vendor who is frustrated with agencies that yield their ad serving decisions to their clients.  Believe me, its frustrating.  I am speaking as someone who believes that technical expertise lies in the hands of the person who uses it.  A client does not gain the insight and experience of SEM like an agency does.  A client does not gain the optimization capacity of ad serving like an agency does.  The agencies are the people who should become proficient with these tools and who should be developing preferences.  I don’t believe that an agency should yield the decision to its client as to how the agency should manage the operation.  I think the agency should reserve that decision for itself. 

To me it should still be about results.  There are many agencies out there that do what I am describing.  They are a “DoubleClick Shop,” an “ExactTarget Shop,” a “WebTrends Shop” etc.  And that has its advantages and disadvantages.  Technical expertise on a platform means an agency offers a client aptitude to optimize ROI throughout a campaign.  If a client pushes new technology on an agency, time is spent learning how to become efficient on the new tool set.  Interesting.  The agency spends the client’s money building a proficiency during a campaign or the client leverages an existing proficiency.  Client’s choice or agency’s choice?  This may all seem like common sense.  But there are a lot of agencies out there that lose sight of control over their operation and allow their dominating clients to make these decisions for them.  And in the end, the agency loses control, the client loses performance and then the agency loses the client.

_uacct = “UA-980395-1″;
urchinTracker();

Behavioral Marketing: In Focus – Targeting Success Stories and The Upcoming Holiday Season

Robert Moskowitz published an article today in iMediaConnection entitled: “Behavioral Marketing: In Focus – 5 Targeting Success Stories.”  Two of the stories have some relevance to a topic that I cover on this blog – equating behavioral targeting to event-based targeting and the opening the door to what behavioral targeting can mean when we consider customer re-targeting.

The first story I would like to talk about is Ecommerce – Sequencing Ads.  In this piece he describes your basic storyboarding of banner rotations with a twist.  The first time someone sees a banner, they see a full price ad.  If they click on that banner and visit the site they get cookied and are considered having experienced an ‘event’ (my definition not Robert’s).  Subsequently if that individual is encountered on the Web again through an ad campaign, s/he is shown a discounted ad upon being recognized as being someone that has been through the event.

So it is a spin on storyboarding insomuch that storyboarding deals with prospect-targeting and rotating ads in sequence without the event of having visited the advertiser’s web page(s).  But this pretty basic event-based targeting example is hardly what I would characterize as behavioral targeting.  We know next to nothing about this user other then the fact that they once clicked on an ad and visited the advertiser’s web page.  They could be a customer already, they could be a prospect.  Who knows.

Like other examples of what people consider “behavioral targeting,” all we have here is someone that has seen a web page before.  Had this person registered, logged on or somehow identified themselves through some incentive, and then left, and we later recognized them through an ad campaign, then we could consider it behavioral targeting.  We would know something about them and could target them based on some behavior or customer segment model.

The statistical results cited in this article were impressive, re-messaged leads represented 25% of the traffic and 50% of the revenue.  But only 7% of the banner impressions served were re-targets.  That means that only 7% of the audience composition was the same audience.  So all of the effort made in capturing people from the event, cookieing them and then getting setup for the re-target reached that audience 7% of the time.  This means that the media buy was not effective in getting to where the audience repeat-visits.  A lot of wasted effort.

When we work with people on customer re-targeting, we help them recognize that when there are sites – more so than networks – that consistently perform well month over month, there is a high likelihood that those sites are visited repeatedly by existing customers as well.  DirectServe campaigns can potentially find audience compositions of pre-existing customers to be as high as 25% or more when implemented.  So when you execute a campaign designed to capture people who have been through an event before, you should be media buying where they will repeat-visit.  Networks are a hard place to do that due to shear size and broadness of the networks composition.  Whereas sites that consistently perform well will likely be sites that are also visited repeatedly by your customers in addition to prospects to it is a win-win for customer acquisition and re-acquisition.

Customer re-targeting is a lot closer to behavioral targeting than event-based targeting is.  But I dare not confuse you by changing the definition for fear of getting lopped in with what is a failure-to-meet-expectation offering.  Customer re-targeting can be customer, registrant or applicant re-targeting.  It is the re-targeting of any known individual.  So if someone clicks on an ad, visits a site and identifies themselves, you can re-target them later based on information you now know about them.  How?  You build user segment groups for your targeting.

The second Story of Roberts that I think is relevant is Seasonal Retailers – Extra Incentives.  This example comes very close to what I am talking about, only it is confined to the 24/7 network.  Again it is event-based as it tags users who have been to a flower shop web site, makes the assumptions that they have shopped and then later targets them with ads following the assumption that they were previous customers.

Had this been true customer re-targeting rather than event-based targeting, the flower shop would have already tagged its customers with cookies at the time of the initial transaction.  Then when it came time to serve ads, they could recognize its existing customers – segmented based on a model of customer profiles – and message to its audience accordingly.  Instead of assuming someone is a customer, you could know they are a customer.  Furthermore, if the individual does not carry the cookie tag, meaning they are not a customer (or they deleted the cookie), then you can easily distinguish them and message to them as a prospect!

How does all of this become relevant to the holiday season?

Now is the time that people will be getting ready to shop online in droves.  A storm is coming to the e-tailers and they are scrambling to get ready.  If e-tailers can create a customer segmentation scheme now – such as a 5 or 10-point schema that describes customer preferences – and cookie customers as they shop between November 1st and December 1st, they could use those cookies for re-targeting online through their online ad campaigns from December 1st through December 24th.  Furthermore, all of the cookies that get written during the holiday season that segment customers for each e-tailer could be leveraged by the online ad campaigns for these advertisers throughout the rest of the year. 

Sounds simple doesn’t it?  It really is.  An ad server targets cookies.  If the cookies are written to reflect a customer segment target then they can be recognized and targeted.  If the advertiser writes the cookies and the ad server can read the cookies, then BOOM you have the integration of online ad serving and online marketing.  That is DirectServe.

_uacct = “UA-980395-1″;
urchinTracker();

How to Pick an Ad Server – Part V: Training

Training is HUGE.  The Big 3 will charge you for training.  This is a negotiable item.  Not always but sometimes.  At least initial training.  But the Big 3, whose interfaces are complex and hard to use for a beginner will charge you every time you need to train someone new.  In an agency, let’s face it, the trafficking guy or girl is usually the lower person on the totem pole.  A new hire.  So these people usually need to be trained and when they get promoted a new person needs to get trained.  Unless it is a smaller agency where someone wears multiple hats and continues to run the ad server.  Find out about initial training when you are evaluating ad servers but also find out about ongoing training.  Find out about what it will cost to train people as you bring on new users.  Find out about what it will cost to train people as new features get added and what it will cost to have updated training when you want a refresher.  I bring this up because we have a very different perspective on this than the rest of the industry.  I am not tooting TruEffect’s horn and I try real hard not to do that on this blog.  To me, an ad server gets paid when a client uses the ad server.  Sure costs need to be kept under control and training and customer support are sources of potential rising costs.  But training builds knowledge in a user so that a user becomes more confident and uses the system, hence more ad serving, hence more revenue for the ad server.  So why charge for training?  We don’t generally charge for training, certainly not initial training, and we try to host monthly training classes on an as-needed basis so that clients can have refreshers whenever they need them.  Any time there is a new enhancement or release, we train all clients.  If a client calls in with a question to their dedicated Account Manager they get trained on that issue on the spot.  Other ad servers will sell you blocks of support time or training time to do the same thing – hidden costs you should look into during the sales cycle.  Again, not tooting a horn here.  What I am doing is providing you with insight so that you know what questions to ask.


Another aspect of training is to find out what you will train on.  What I mean is, will you train on a demo sandbox environment or will you train on your own campaigns?  It seems relevant to me because after all of your analysis and research, when you finally select an ad server, you are ready to go.  In fact most clients I work with are ripping and raring to go.  In fact, they are late with a campaign and are needing to get going now!  So training is an immediate need and so what you will be trained with IS important.  DC, Atlas and Mediaplex will train you in what is called a sandbox.  It is a demo environment where you will use fake campaigns and will be shown how to do things.  I don’t personally know what Zedo does.  But the best way to train is on your own campaigns and on your own data.  When you get trained, ask to get trained with a campaign that you need to get setup so that you kill two birds with one stone.  Usually you don’t even have time to do all of the training in one sitting by the time you finish your decision and that should be okay.  If you have spent as much time as most people on the selection process, an ad server should be willing to still put a little more time in with you and not simply flip you over to customer support so fast. 


My clients get trained in real-time, so to speak, on campaigns that they need to get setup.  Their dedicated Account Managers share a hosted environment wherein insertions are built, digital assets are loaded and placements are assigned.  Schedules are built and tags are sent together by the Account Manager and the client.  The training is conducted all while the campaign is completed so that they client is able to do real work while getting trained.  It is a good use of your time that way.  I tell my buyers this when I am selling to them, whether you work with me or not, make sure you get this from whatever ad server you work with.  Don’t let your valuable time get wasted with training in a demo environment unless you have that sort of time.  Negotiate detailed training on top of the contract for a time when you have it in addition to campaign setup training.

_uacct = “UA-980395-1″;
urchinTracker();

How to Pick an Ad Server – Part IV: Price

Price.  Here is a biggie.  I have mentioned before that I see $0.025 and I see $0.33 CPMs.  I see $0.01 and I see $0.03 CPCs.  Atlas is notoriously the most expensive and Zedo and Mediaplex the least expensive.  Cheapest has its hidden costs in reliability of technology, lack of or quality of client support, inefficiency of scalable architecture, poor ease of use and a potential limitation in the number of sites that will accept the ad server’s tags (I haven’t touched that one before – but not all sites accept all ad servers).  The most expensive ad servers also does not mean they are the best. 


Prices are usually tiered, getting cheaper with more use.  The bands can run something like this:


• 1-25 million impressions/mos
• 25-50 million impressions/mos
• 50-75 million impressions/mos
• 75-100 million impressions/mos
• 100-300 million impressions/mos
• 300-500 million impressions/mos
• 500-1 billion impressions/mos
• >1 billion impressions/mos


Not everyone sees all of those tiers and every agency and advertiser will see a variation on these tiers and on the rates.  But that’s a pretty standard example.  Sharing with an ad server during the sales cycle your current impression volume is important.  And being accurate is also important because they will provide you with preferred rates.  But also be sure to share what you believe your growth will be as they will give you the higher tiers.  Don’t make the mistake of saying we get 50M impressions because you will only get tiered up to 50 or 100M impressions.  If you grow your rate will get stuck at that top tier and you could get better rates as you grow if you have the higher tiers in your contract.

_uacct = “UA-980395-1″;
urchinTracker();