This is not unlike the “lean startup” approach described by Eric Ries in his book, The Lean Startup. In it, he advocates the “build-measure-learn” cycle, which puts experimentation, measurement, and learning at the heart of the organization — the goal being to figure out how customers respond to a product before dedicating larger resources toward its development.

The reason why this approach is popular among comedians and start-ups alike? It’s far less risky to go big when you know what works based on prior first-hand data.

We will now discuss a similar framework for approaching display advertising from a “lean” marketing mindset. By using real-time bidding (RTB) as an efficient means to experiment with campaigns and gain insights prior to scaling with direct buys, marketers can figure out the most successful elements of a campaign prior to allocating larger budgets.

Let’s dive deeper into the four strategic phases of this lean display advertising approach:

Click to enlarge

Phase 1 – Build Your Test Campaigns

“Never stop testing, and your advertising will never stop improving.” —David Ogilvy

The goal of the first phase is to create various campaigns aimed at a variety of potential audiences to see how each responds. This is where baseline data is gathered for further analysis. By creating a series of well-crafted test campaigns (i.e., experiments), it’s easier to learn from them in the analysis phase.

There are several keys to the testing phase, the primary ones being:

• Tracking
• Segmentation
• An ample variety of ads

In terms of tracking, you want to properly measure your campaigns. Measurement is extremely important when it comes to analyzing and optimizing your campaigns, which are the keys to extracting the highest value from them. It also sets the stage for intelligently scaling up your campaigns in the future.

To do this effectively, make sure that you are tracking as many variables as you can. At the very minimum, you will want to track conversions or desired actions (preferably with an associated value). In an ideal scenario, you’ll want to track not only conversions and revenue, but the impact of campaigns on brand searches, view-through conversions, bounce rate, landing page engagement, and even lifetime value.

With respect to segmentation, this simply means properly defining the scope of each campaign. You can’t assume that audiences in different regions, age groups, or income levels will respond to the same ad, on the same website, in the same way. This is why segmentation is important: control the variables to make for better data.

Lastly, having a wide variety of ads within your campaigns is the key to enabling good performance comparisons. If you create many nicely segmented campaigns, but they all use the same standard banner ads, you miss the opportunity to split-test the impact of multiple ads with each of those audiences. Having variety enables you to increase the amount you learn from your campaigns.

Phase 2 – Analyze The Data

The second phase is where you measure, analyze, and learn from the data that was collected in the first phase. In order to extract insights from your campaigns, digging into the data on different levels is crucial.

Some key areas to analyze in this phase are the engagement and impact of:

• Your ads on a placement level, site level, and in aggregate
• Visitors from specific regions, on certain devices, at certain times, etc.
• Specific websites and placements in general
• Specific audiences in general

When analyzing the data, you should be able to discover which combination of campaign elements and audience factors show the most potential, and which need to be pruned in the optimization phase.

Phase 3 – Optimize Your Campaigns

Once a proper set of data has been collected and analyzed, it’s time to cut the weak elements from your campaigns and allocate more resources to the elements that are working. This is the goal of the optimization phase: to improve your campaigns by using the insights learned from the previous phase and applying them in a meaningful way.

One example of how to improve a campaign is to disable or prune certain elements — such as ads, placements, or sites — if they are found to be under-performing. This allows the elements that perform well to receive a larger share of resources, thereby improving overall campaign effectiveness.

Another way of allocating more resources to elements and campaigns that work is by increasing budgets, bid prices, and frequency caps (incrementally) in order to drive more volume.

Optimization should really be viewed as a continual process of learning from the data and improving campaigns over time. Once you have optimized campaigns on your hands, the next logical step is to scale them up.

Phase 4 – Scale Your Winners

“What enables the wise sovereign and the good general to strike and conquer, and achieve things beyond the reach of ordinary men, is foreknowledge.” —Sun Tzu

It is only after you have acquired real-world data (foreknowledge) that the time becomes right to invest the effort and dollars into scaling your campaigns with direct media buys. Here is where the complementary nature of RTB and direct buys reveal themselves to the advertiser.

We already know that buying display ads with real-time bidding (RTB) is more efficient, so your initial testing and optimization are best conducted on agile platforms that enable you to exploit the benefits of RTB. Armed with your insights, you can now confidently approach publishers to reserve more inventory.

Imagine the following conversation with an ad sales team, where you say:

“Your inventory is performing well for us via RTB. We are paying around $2.50 eCPM, but we would be willing to pay a slightly higher rate, or commit to a certain order size, to reserve more inventory directly.”

The publisher’s ad sales team might then say:

“We’ve only been putting a fraction of our inventory on RTB, but for a monthly commitment of $X dollars at $Y CPM, we could probably guarantee Z million impressions per month for you.”

Savvy publishers are already well aware that RTB is great channel for prospecting direct sales. If a specific website or publisher is performing well for an advertiser through RTB exchanges, there’s a good chance they want more of the same inventory. Knowing the inherent volatility in RTB volumes, approaching publishers for direct buys (after testing, of course) only makes sense.

A note of caution: this may not always be a practical course of action. While exploring direct buys, you may notice that CPM rates exceed a level that is feasible for reaching your performance goals — so always keep your key metrics in mind.

In any case, buying display ads via RTB is an ideal way to gain knowledge and test out the performance of specific publishers and audiences, without blindly committing thousands of dollars on high-CPM direct buys. Once you’ve discovered valuable placements, you can then approach the publisher directly to reserve your own inventory on a guaranteed basis.

This strategy enables advertisers to build impactful display ad campaigns, all while minimizing risk and wasted resources, and maximizing the amount of learning in the process.

The good news? It doesn’t have to be this hard. An evolving market has far more benefits than disadvantages, if only you know how to wrangle them. Personalization is just one example of this, especially within site retargeting.

Before you scratch your head and think that personalization is already in full force within site retargeting, take a step back and let me explain. I’ll walk you through the evolution, current landscape, and actionable steps you can take to truly weave personalization into your site retargeting for unmatched ROI and impact.

Understand The Evolution

Site retargeting is a powerful idea that began as something really elementary. Initially, the process went like this:

Buyer visits site. Buyer leaves. Buyer is served a relevant ad.

Simple. Basic. Effective. Of course, we quickly realized that this powerhouse needed to be let loose — cue, audience segmentation on the website. The next stage of the process went like this:

Buyer visits site. Buyer’s behavior on site is collected, including categories searched, products viewed, cart abandonment, etc. Buyer is bucketed into a segment with other buyers of similar actions. Buyer is served ads based on the bucket he/she was relegated to.

At this stage, site retargeting basically evolved from serving the same ad to everyone in the ocean to segregating the audience into smaller ponds. It made some sense, but we knew that more could be done. So, we began tapping even further into the world of personalization, and the process evolved to look like this:

Buyer visits site. Buyer is segmented into a bucket of like-buyers. Buyer is served personalized creative based on product(s) viewed, recommended products, product interaction, etc.

With this latest evolution, creative aspects were tailored to fit specific visitors’ behaviors — to very successful ends, I might add. The shortcoming? For most companies, personalization stopped there. The site retargeting train was making serious progress, but those segmented buckets we talked about earlier slammed on the brakes — hard.

If you’ve read my posts before, you’ll know that I’m a fan of unstructured data rather than segmented data. Personalization of creative is great, but there’s a huge opportunity to take it even further by treating each individual visitor as a “segment of one,” thus ensuring that bidding and optimization decisions are also personalized. And — you guessed it — that means looking at and using raw, unstructured data rather than their bucketed cousins.

Extend Personalization To Produce Better-Optimized Bids & Creative

As I mentioned briefly above, most site retargeting companies are continuing with the same ad frequency, same bidding and same optimization across bucketed data elements — which makes zero sense.

For example, if a male buyer visited a high-end retailer’s website and checked out two very expensive luxury watch brands, ads targeting him should be very different from those targeting a male buyer who visited the same website and looked at a couple of inexpensive exercise watches.

The important takeaway here is this: if these two buyers were each to visit a brick-and-mortar store and then leave, which of the two would you allocate more money toward trying to get back?

There will almost always be one, clear-cut choice that will equate to higher ROI, and understanding buyer behaviors on a granular level (via unstructured data) gives the insight necessary to optimize bids and ad creative far more intelligently. This is key for increasing ROI and performance levels.

Next Steps?

By now, I think I’ve hammered home the idea that personalization is great, but that it must also extend to better optimization of bids and ad creative. The next logical question becomes this: how do you make that happen? Here are some dos and don’ts:

Don’t: Use Separate Companies For New Audience & Existing Audience Retargeting

In theory, I get why you might do this. But, separating your new audience buys from your site retargeting (or existing audience) buys is a recipe for misinformed decisions, attribution challenges and weakened power behind your efforts. We all understand the idea that maybe you chose company X because it’s full of experts in site retargeting, and then paired it with company Y for your new audience buys because that’s their specialty. This sounds ideal, right? Not so much in practice.

The browsing history of each prospect before, during and after a site visit makes your audience buys smarter across all display channels. If you rely on the information collected through one vendor and view the data from the other as an entirely separate piece, you’re missing out on a heap of valuable insight. It’s time to bring new audience buys under the same roof as existing audience buys and allow them to strengthen and reinforce each other.

Do: Embrace Micro-Segmentation

Many retargeting companies are waxing poetic about taking a website and shrinking it down to the smallest segment possible. Yes, drilling down to the smallest segments is the goal, but stop to think if this is really the best way to get there. Instead of trying to break audiences down further and further into niche markets based on your site, why not use a platform that utilizes data that’s already broken down to the most granular level?

Unstructured data, anyone? A platform like this already has this “pre-visit” interest and intent data on your audience before they arrive at your site, so you can personalize from the ground up instead of breaking down existing audiences.  It’d be in your best interest to explore these platforms first.

To reach the highest level of impact with site retargeting and get the best ROI, look for a partner that can build, bid, report and optimize based on unstructured data. Viewing audiences as individuals allows decisions to be made based on more information than what was viewed. You end up with historical information as well as behavioral details on what they did after they left. These data equip marketers to target based on the nuances of why certain customers convert and others don’t — making them smarter and more successful.

Image via Magnetic

More and more Web browsers are blocking the use of third-party cookies by default, a development which has sparked a fierce debate in the digital ad industry.

Firefox is the latest browser to make the announcement, forcing marketers, ad technology companies and ad agencies to discover another means for reaching consumers with targeted advertising.

This controversy began with Safari, but because Safari had a mere 5% market share, it didn’t ruffle many feathers. However, over time, other browsers have followed suit including Microsoft with “Do Not Track,” and now, Firefox.

Even more troubling is that on mobile – the fastest growing Internet access method — iOS/Safari has over 60% market share, and that number is just too high to ignore. It’s time to face the facts: cookies are being kicked to the curb.

Cookies Aren’t Used For Insidious Purposes

Digital advertising is a $40 billion industry in the U.S. alone, and about half of that spend requires using third-party cookies to locate and target relevant consumers. Third-party cookies aren’t used for anything particularly insidious. They are needed to target reach and frequency for online campaigns, for example. Without cookies, ads can’t properly be tracked for frequency and could accidentally be served to the same user hundreds upon hundreds of times. Basic stuff…

Cookies are fantastic largely because users can dump them and start anew. They are also great for advertisers because they effectively track the performance and targeting, providing the basis for campaign analytics.

What Happens When The Cookie Crumbles?

So what is going to happen when cookies go away? Do you remember the maxim, “There’s no such thing as a free lunch“? There is a tacit agreement between consumers and publishers: the Internet is free but you’ve got to watch our ads (or pay for our content).

As you may imagine, no advertiser wants to deliver you ads you don’t care about, so understanding what you are interested in is the tradeoff. And, with $40 billion dollars in spend at stake, and a compound growth rate that is the envy of the rest of the economy, the ability to target and measure digital ads simply isn’t going to go away.

According to a statement from Randall Rothenberg, president and CEO of the Interactive Advertising Bureau, “Without third-party cookies, they (users) will not be able to participate in the existing industry system for privacy protection.”

The truth is that targeted advertising is too large of a business to be shut down because browsers ban third-party cookies. The end result will just put advertisers and the digital industry in a position to figure out other ways to reach consumers with relevant ad messages. In theory, if all third-party cookies were blocked, a huge portion of this industry would disappear overnight. But, that won’t happen.

Consumers Will Have Less Control With Cookie Alternatives

For a glimpse of the future, let’s look at mobile. Who was the first company to stop allowing third-party cookies? Apple, which owns Safari. Apple also blocked the Unique Identifier that enabled advertisers, vendors and others to identify mobile devices. What did Apple do next? It released its own proprietary version of the cookie: The Apple ID for Advertisers, which is persistent (unlike cookies).

As traffic continues to move away from the traditional desktop and laptops and onto tablets and other mobile devices, the industry will need to devise various ways to identify each of these devices to anonymously target relevant users. And, whatever this technology is, it will end up being adopted on “desktops” and all other devices that access the Internet.

However, there is one big difference: it will be more difficult for consumers to control. Each company will use their own method of identifying and targeting consumers, and those consumers will no longer be able to easily erase these identifiers like they can with cookies.

Achieving The Opposite Of What They Intend

While companies like Mozilla, Microsoft and Apple may think they have the user’s interest in mind when deciding to not accept third-party cookies by default, they are actually doing the opposite. A default setting says nothing about what consumers really want.

The growth of mobile computing will provide the publicity cover. With each browser setting different standards and approaches, combined with the growth of tablet and other smartphone usage, users will find themselves moving from a simple life where they knew what was going on and had control  in their hands, to having their control fractured into a myriad of inconsistent choices.

Topic targeting used all by itself can also be very powerful, especially for branding, new product launches or to get on a wide variety of sites relevant to your topic. With topic targeting, if you were releasing a new book on computer programming, you could easily target the topic for “programming” or, if your book is about scripting languages, you could tell Google to only show your ads on sites and pages that had content about scripting languages.

Keywords Vs. Targeting

In Google’s Display Network, with keyword targeted AdGroups, Google goes out onto the ad network and uses your keywords (along with your text ad copy and landing page) to reach the audience you are trying to target.

If you use the keyword [tennis] or a set of keywords that targets tennis, Google will show your ad on sites that have content relevant to tennis. What many people don’t realize is that through topic targeting, you can also set up AdGroups that target Tennis.

How It Works

There are multiple ways to add Topic Targeting to your AdGroups or to create a new Topic Targeted AdGroup.

To create an AdGroup for Topic Targeting:

To add Topic Targeting to an existing AdGroup:

You can also use AdWords Editor to add topics. The AdWords Help section has a great walk through and lists all the available categories here. In AdWords Editor, just go to placements, add placements and then add the topics directly using this format: category::Sports>Individual Sports>Racquet Sports>Tennis

To monitor how your Topics are performing, login to AdWords where there is a tab for Display Network and then a sub-tab for Topic Targeting.

Combinations

And now let me tell you about how Topics can work as part of a team. Topics are very efficient and effective when you combine them with other AdWords strategies like keywords and placements.

Keyword Example: If I were to go after banking (category::Finance>Banking), I might have one AdGroup with a keyword or set of keywords targeting ATM locations with an ad for free ATM withdrawals. I might set up another AdGroup with the same topic for banking but use a keyword or set of keywords targeting online bill paying with ads for free online bill paying when I set up a checking or savings account.

Placement Example: About.com and Ehow.com are both content-rich sites that cover a variety of topics. If I were a company that sold children’s over-the-counter cold medicine, I could use About.com and Ehow.com as placements and combine them with a topic such as Pediatrics (category::Health>Pediatrics) to get in front of a large audience of mostly mothers with small children.

Combining your remarketing audience with topic targeting can eliminate some of the more random traffic associated with standard remarketing strategies. If you want to refine your remarketing strategy to only show your ads when your potential customer is in the right mindset, using topics is a great way to accomplish that goal.

Tips, Tricks & Challenges

Topic targeting works better if you limit your topics to one per AdGroup. If you really need to combine topics, keep the list small and try to group them together so that they are all similar (like all news related).

Too many combinations can kill your AdGroups. If you try to use topics with keywords with placements and then throw some audiences in the mix, you may find that you are not getting many impressions or clicks at all. It takes a little bit of experimentation to find the right combinations with topic targeting, so keep tweaking.

Exclusions

Topics work great as exclusions. If you had a website aimed at recipes using dates, topic exclusions could keep your ads from showing for dating and relationship websites.

Regardless of how you integrate topics into your AdGroups, you are adding power and refinement to your AdWords strategies that very few people are using.

The process that a marketer must follow, however, to reach that end goal is very different depending on the approach, and each has its pros and cons. To help you understand which approach may be best for you, we’ll begin by contrasting some of the characteristics of traditional direct buys and RTB, and what they mean to brands, agencies and marketers.

1. Targeting – Websites Vs. Audiences

If there is one thing you need to remember, this is it: The fundamental difference between direct buys and RTB is the shift from buying ad impressions in bulk (direct), to auctioning each impression off individually to the highest bidder (RTB).

With direct buys, you are essentially buying impressions in bulk, in order to have your ads seen in a specific context (e.g., on ESPN.com). You have the ability to filter the audience that sees your ads with targeting rules such as geography or browser type, to name a few of the basics, but you’re still ultimately targeting your ads to a specific website.

This works especially well for brand advertisers and agencies that are very sensitive to the placement of their ads, and are willing to pay premium prices to secure such inventory. Brands are also afforded more freedom on the creative level when working directly with publishers. Rich media formats, like page takeovers and other types of custom brand integrations, are currently a luxury only possible with direct buys.

With RTB, each impression is profiled and evaluated in milliseconds during the auction process (while a page loads). You can target ad viewers at a demographic, psychographic and behavioral level, but the reach of RTB enables you to do so across a wide array of sites, rather than on just one, making it possible to target audiences at scale.

So, instead of being limited to buying ads on ESPN.com to reach your audience, RTB allows you to buy ads on (almost) any site that “sports fans” may visit. You also have the ability to take a more agile approach to campaign optimization, since each impression is being bought individually, allowing for more efficient performance and control.

Everyone wants exposure to the right audience, which is why this approach works well for almost all advertisers.

2. Supply – Guaranteed Vs. Non-Guaranteed

Another fundamental difference between direct buys and RTB is the level of certainty that your ad campaigns will receive the volume you want or need.

With direct buys, you agree to buy a heap of ad inventory at a fixed CPM rate that the publisher will deliver in the future. In that sense, the inventory is “guaranteed” or “reserved” for you. Barring any external issues, you will receive the impressions you agreed to purchase at the outset.

This works well for advertisers and agencies that have specific exposure goals and require a high level of certainty that campaigns will deliver. In exchange for paying a higher rate to the publisher, you get certainty of campaign volumes and avoid the naturally competitive landscape of RTB.

With RTB, as the name implies, you are in an auction with a multitude of other advertisers, all bidding at different rates (explained later) for each impression, in real time. In such a dynamic environment, the ad inventory is considered “non-guaranteed,” due to the unpredictability of the marketplace.

When you don’t know what other people are bidding, there is simply no guarantee that you will win the impression you bid on. Furthermore, guaranteed buys usually have priority over RTB. This means, if the demand for guaranteed inventory on a particular site increases, the supply available on RTB for that site correspondingly decreases.

3. Workflow – Manual Vs. Programmatic

Another difference between direct buys and RTB is the workflow of launching campaigns.

For the most part, direct buys consist of a manual process that involves hours of human effort in planning and execution (pdf download). It requires reaching out and making initial contact with the publisher’s sales team, negotiating and planning the “insertion order” (a contract outlining the terms of the ad campaign), emailing ad tags back and forth, and so on, all in preparation to launch.

The publisher ultimately controls the flow of the campaigns using their ad server, which means that there is a natural delay when it comes to campaign control and reporting. This asynchronous process is not only prone to miscommunication and human error, but also requires hours of human time on something that could be handled instantaneously with a programmatic solution.

Then, there is the issue of management complexity. Take the traditional media buying process – negotiating a direct buy with a publisher, and working through all the setup – and try multiplying it by ten or twenty. Now, imagine the complexity of having to run them simultaneously, like an agency would, and you quickly realize the overhead required for management alone. Not to mention the process of negotiating with various sales teams, which is an art in and of itself.

In contrast, RTB is a primarily programmatic process driven more by user interfaces and algorithms, and less with phone calls, emails and contracts. There are still manual elements (such as ad quality review, tech support, and billing), but nothing close to what is necessary when dealing directly with publishers — let alone a group of them.

As a result of the programmatic process, more of the campaign components are real time (i.e., nearly instantaneous) in nature, from controlling the flow of campaigns, to reporting and optimization.

To be clear, even though RTB is a programmatic channel, that does not mean that direct buys won’t eventually enjoy a programmatic future. There are several ad tech vendors out there giving publishers the ability to offer programmatic access to inventory on a guaranteed basis, all while maintaining control over pricing and ad quality. These technologies have yet to reach critical mass, so for the time being, we should probably keep direct buys in the “manual” workflow category.

4. Pricing – CPM Vs. eCPM

Another primary difference between direct buys and RTB is the nature in which inventory is priced. This difference stems from the fact that with direct buys you are buying impressions in bulk; whereas with RTB, you are bidding on individual impressions separately.

Direct buys are almost always priced in fixed CPM rates, where the inventory is sold in bulk and all impressions are essentially priced the same (e.g., $10 CPM, or $10 per thousand ad views). This pricing model has been the standard since the inception of the banner ad, and doesn’t look like it will be going anywhere in the foreseeable future.

With RTB, each impression is auctioned off. Since each impression is priced individually (and since a cost-per-impression metric would be wildly impractical to advertisers from a reporting perspective), the de-facto metric for RTB pricing is effective CPM or eCPM.

Another way of thinking about how these two pricing models differ is by using an analogy like apples. Buying ad inventory directly from publishers is like buying giant bushels of apples. You pay a fixed price for each batch, and receive various levels of quality within the bunch.

But with RTB, you are essentially bidding for each apple (impression) based on its individual characteristics. This means that you still end up with 1,000 apples at the end of the day, but the overall cost for the batch will be a dynamic value derived from all the individual prices you paid for each – hence the term effective CPM.

5. Accessibility – Barriers To Entry

The final difference we will cover between direct buys and RTB is the accessibility of each approach. Given the “traditional” nature of direct buys, they typically have much higher barriers to getting started compared to RTB.

The first barrier faced by marketers performing direct buys is the sizable minimums in ad spend required by most publishers to get started. In general, you can expect a commitment of at least $5,000-10,000 for a direct buy of guaranteed inventory.

On much smaller sites, you can get away with paying flat rates of a few hundred dollars. For larger publishers with attractive inventory, you won’t get any attention unless your budget is in the 5- to 6-figure range. To many small- to medium-sized marketers, this can be a non-starter.

Remember: this is considered “premium” inventory in the eyes of the publishers, so don’t be surprised when you hear the CPM costs and minimum spends. The truth is, when going directly to the website publisher, there is an unspoken rule that all prices are negotiable, so website publishers will often quote surprisingly high rates. It’s not uncommon to receive rate cards anywhere from $10 to $75 CPMs (and beyond!) from many ad sales teams. Such rates can be completely impractical for anyone but large brands.

Serious media buyers also need to consider the costs for an advertiser-side ad server. If you are managing direct buys from multiple publishers, you want to be aggregating and auditing your own numbers, in case of reporting discrepancies. You also want a central place to log in to and reduce the complexity of logging in to multiple publisher ad servers.

I can elaborate on this point in a future article, but for now, it’s important to understand that with direct buys there are extra costs associated and technical learning curves to overcome that may not be obvious at first thought.

As opposed to direct site buys or buys through large ad networks, which have higher barriers for marketers to get started, buying RTB inventory through a DSP (depending on the one you choose) has a much lower barrier in terms of financial commitments and operational management. Obtaining an ad server also isn’t necessary, since DSPs provide the ad serving and publisher integrations on your behalf. So, for the marketer, there is far more control and far less friction in the media buying process in a programmatic RTB environment.

Moving Forward

While buying ad inventory directly can be inefficient (not only from a price perspective, but also operationally), it does provide a number of benefits, which make these hurdles palatable to certain types of advertisers. RTB overcomes these pricing and operational problems, but introduces its own: guaranteed volume.

Hopefully, we now have a clearer picture of the differences between direct media buying and real-time bidding. Keep in mind, it’s quite possible to use both approaches in a complementary way, but it requires a data-driven mindset and a methodical approach to optimization and scaling. Moving forward, we will discuss these methods so you can apply them to your own display ad strategy.

Instead of being strategic about your ad buy, you’re forced to take your vendor’s word for it that the consumers you’re targeting are actually being targeted. This is unfortunate at best and reckless at worst. After all, these are the interwebs. There’s data everywhere!

Is the lack of transparency a result of a true blind spot in technology? Or is your vendor holding back key data on your campaign due to a lack of technical sophistication of their own or in a shell game designed to separate you from your hard-earned marketing budget?

Whatever the motive, here are four types of campaign data you probably don’t have visibility into… but should.

1. What Data Is Being Segmented?

When you buy targeted ads based on interest and/or intent, you’re probably not driving terribly far down the long tail of keywords, sites visited, content consumed and more. Let’s say you want to compete for luxury clothing shoppers who searched for Michael Kors. It would make a lot of sense to differentiate between those who searched for Michael Kors sale and Michael Kors couture. These are two distinct user groups that require distinct messaging, bidding, and optimization to convert to a sale.

Unfortunately, in most RTB platforms, you’re not buying data segmented along common sense lines. You’re buying bulk data where search queries, sites visited and content consumed get lumped together as Michael Kors or worse, “luxury shoppers.” When data is packaged in this way, there’s a significant loss of performance and information that’s valuable to you, the marketer.

2. When Are The Ads Being Served?

Although attribute and intent data are incredibly important, don’t overlook the impact of timing on programmatic media buying.

We looked at timing and recency in a recent study of more than 200 display campaigns conducted by Simpli.fi clients in the personal finance industry. Unsurprisingly, click-through rate (CTR) was highest in the moments immediately following the search. Meanwhile, cost-per-click (CPC) was lowest in the sweet spot between the aggressive bidding in the seconds immediately following the search and the period so much later that the ad was no longer relevant to the user.

Of course, you can’t tap into these CTR and CPC efficiencies unless your platform enables you to bid differently based on the recency of a data point (like a search query or site visit).  Knowing that your audience engaged in some undisclosed action in some unknown time window over the last thirty days is a dangerous way to go to market. This is, however, the reality of the legions of vendors buying prebuilt data segments off the shelf from the growing mass of data providers.

3. Where Exactly Is Each Ad Served? 

Any ad platform can tell you (most of) the sites upon which your ad was displayed. You just don’t know how many times it was displayed on each site. Let’s say you bought 250 impressions. Did you get 200 impressions on NYTimes.com and 50 on Jezebel.com? Or was it the other way around? Does it even matter? I think so.

Of course, in most cases, this information is not readily available. You may get a sample list of target sites, but you won’t get the whole list. Instead, your ads are blasted into a vortex with little or no visibility or accountability.

Even if you ask for a full list after the campaign, the answer you’ll often hear is no. How do you know that your ad impressions – and ad dollars – aren’t being sucked up by junk sites in the vortex?  This is not true of all vendors. Like the wise man once said, “Choose wisely.”

4. What’s The True Cost Of Each Ad?

In the real-time -bidding advertising market, costs fluctuate constantly based on supply and demand. Sometimes, the price climbs, and you pay more. Sometimes, the price dips and, well, you pay the same.

That’s right. Some technology vendors or DSPs charge fixed rates for the cost-per-impression based on a maximum bid, even though RTB prices are typically 30 percent to 40 percent below the maximum bid submitted to the auction. You, the brand, continue to pay the negotiated impression rate regardless of the true cost of the ad.

This impacts your campaign in two ways. First, you lose visibility into the true cost of the campaign in terms of the cost-per-impression, click, or acquisition. Second, you’re not gaining the financial advantage of real-time-bidding with your “fully-managed” campaign. What’s the point of investing more in “fully-managed” when the results aren’t accounted in key cost-per metrics, and the financial benefits are pocketed by the platform?

Demystifying The Box Of Chocolates

When marketers enter into an agreement with an advertising technology vendor, there’s an underlying expectation of transparency. And, yet, when you ask the questions above, many vendors have no good answers.

Marketers are more savvy that that, which means available data needs to be more granular. There’s no reason to be limited by segments and sample lists or to turn a blind eye to time and price. Whether we’re doing programmatic media buying for direct response or for a branding campaign, we have found that this level of insight and control is invaluable.

It’s time to start asking for more visibility. You shouldn’t be left guessing whether this particular chocolate will be filled with nougat or nuts.

Some of us may think a gaming site like IGN is premium; others may disagree. Overall, the basic premise is that premium inventory is the media found on a well-known and well-respected publisher.

Clean, Well-Lit Inventory

Personally, I prefer “clean, well-lit” inventory as an expression of what’s good rather than “premium.” The popularity of sites can wax and wane so rapidly that it’s impossible to keep up with the top 20 in every vertical – let alone the subsequent 50 or 100.

And, what if you are buying ads on an exchange? The reason we do that is so we can target individuals no matter what content they are looking at.

If you see that customer who’s about to convert and he’s on a safe site that has good media placements, isn’t that “premium”? It certainly seems a shame to call that sort of inventory “remnant” or “unsold” or even worse, “cheap.”

Inventory That Lifts Brand Metrics & ROI

Kellogg recently reported that more than half of their media now runs through programmatic buying with an increased focus on private exchanges. For Kellogg’s, premium inventory is simply defined as “inventory that lifts brand metrics and ROI.” Their representative even went on to share that “ironically, that could be the cheapest CPMs out there.”

We spoke with a couple of agency executives, and they shared their thoughts on how they define “premium” inventory.

Traditionalization Of Digital

According to Rob Griffin, EVP, Product Development at Havas, “Everyone wants to talk about the digitization of all media, but no one discusses the traditionalization of digital.”

Griffin adds that this is an important distinction when you look at the future of programmatic and media buying in general. Griffin’s perspective sheds light on the fact that all media will be programmatically bought, but that not all media will have an algorithm applied to it.

Griffin explains that premium inventory will still remain sold via upfronts, where the price is set with the publisher. However, the trading desks will handle the executions. The non-guaranteed media (RTB) will represent a digital spot buy, and then you have your performance agencies that represent a digital version of a DRTV shop.

In closing, Griffin states, “in this traditionalized model for digital, premium is defined advertiser-by-advertiser and publisher-by-publisher, and then purchased accordingly.”

Limited Inventory In Short Supply

Joel Nierman, Marketing and Media Director at Critical Mass takes a slightly different approach to defining “premium.”

“In digital, premium inventory denotes severely limited supply; limited in such small quantities as to actually drive the price up, given the size of the digital ad opportunity.

With that definition, this leads digital premium to be ads that are supply-limited in at least one of the four following ways: 1) context, 2) user groups, 3) performance, 4) ad format.

The common bond here is limited supply – there are limited opportunities to be next to the best content, to reach truly high ROI user groups and tightly-focused groups, and limited opportunities for immersive, innovative creative executions.”

However, just like Griffin, Nierman goes back to the notion that advertisers define value, stating, “Advertisers are willing to pay more for something that performs well over the normal performance band.”

Inventory Defined By Value

It seems like the overall takeaway here is that inventory is defined by its value and comes down to what an advertiser is prepared to pay. Data has emerged as the key contributor to the change in buying and inventory channels.

The days where there is a clear-cut line between premium and remnant inventory sources is gone. Simply put, it’s subjective and means different things to different people.

To some advertisers, premium can still be defined as the major publishers with direct sales channels; while on the flip side, some exchange-based inventory can be defined as premium by marketers’ standards. Let’s delete the pejorative “premium” and “remnant” and go with “clean, well-lit.” If it isn’t clean and well-lit, we don’t want to buy it.

Campaign Code Filter

Coming Up With Code Names In The AdWords UI & AdWords Editor

Codes may be your secret power against time. There are so many options now for setting up your Google Display campaigns that you really need a way to stay organized when you are managing your display campaigns and reporting for them. Code names are not just for display. Since they apply to paid search also, you will see several tips sprinkled in for PPC, too.

To get started, you will want to use codes in the names for your AdWords campaigns and for your AdGroups. The campaign codes are the most important. Coding AdGroups are helpful, but it can be fairly time consuming to go back and code/rename all your AdGroups.

Ideas For Code Names

• Network = GS (Google Search), GP (Google Search plus Partners), DP (Display), OV (Online Video)
• Device = Comp (Computer), Mob (Mobile), Tab (Tablet)
• Geographical = US (USA), CAN (Canada), UK (United Kingdom), AUS (Australia)
• Languages = ENG (English), FR (French), SP (Spanish)
• Ad Scheduling = AS (Ad Scheduling) or MF85 (Monday through Friday 8 to 5)
• Company or individual who created the campaign = CP (Cardinal Path) or SE (Shelley Ellis)
• Types of display campaigns = KW (Keyword), TP (Topic), IN (Interest), RM (Remarketing), YT (YouTube), MP (Managed Placements)
• Bidding Strategies = CPM, CPC, CO or CPA (Conversion Optimizer), DCO (Display Campaign Optimizer)
• Combinations = KW+TP (Keywords + Topics), etc.
• Ad testing = AT (easily filter campaigns using the AdWords interface for ad testing)
• Other ideas = Codes for categories of products, sales and promotions, date triggers

Tips For Coding

• Use a hash or period between your codes to make it easier to filter. Otherwise, you end up filtering parts of the actual campaign name, too.
  • Example #1: GS.MO.US.CPA.KW+TP.SE.ThisCampaignName
  • Example #2: GS-MO-US-CPA-KW+TP-SE-ThisCampaignName
  • Be consistent. There are times when you need to filter all Google Search campaigns in the US and it will be much easier to do that with one filter rather than custom filters.

Using Filters In The AdWords UI, AdWords Editor & Analytics

• In the AdWords UI, start by using your codes to filter your Campaign view.
• In the AdWords UI, use filtering with the AdGroup tab, the Keyword tab or other tabs; you can use a custom filter to see only the data for a specific set of campaigns. For example: under the keyword tab, I could filter for –RM- to see all remarketing campaigns.
  

  Keyword Filter

• In AdWords and AdWords Editor, save custom filters for the search results you need to see daily or weekly.
• In Adwords Editor, codes can make it easy to find campaign sets through the regular filter or through the advanced filter.
  

  AdWords Editor Filter

• Analytics offers a similar opportunity for you to quickly use filtering to narrow your view of data and reporting.

Using Excel Filters With Placement Performance Reports (Full URL)

Domains and URL’s often use a variety of keywords in their formatting. Filtering allows for you to create very targeted managed placement campaigns and AdGroups with keyword targeting pulled from the domain or URL name.

1. I start by pulling a placement performance report for ALL URLs (it can be across one campaign or multiple campaigns).
2. The first tab in your Excel worksheet will be for ALL keywords.
3. Using a custom or keyword filter allows me to start breaking my placement URL list down into very targeted groups of managed placements with each new tab representing a new filter. If I were an online gift store, I might do filters for words like birthday, graduation, anniversary or new baby, and then create AdGroups and new ad copy to target those placements specifically.
   

   Excel Filter

4. Use filtering for impressions, clicks and conversions to find new audiences. I never cease to be amazed as I review placement reports and discover audiences that no one in my company or my client’s company had considered before.

What types of codes do you use in your AdWords campaigns and AdGroups? Do you combine AdWords labels with codes or just use one or the other? What lessons have you learned or tips do you have for using codes?

To those with experience in the RTB realm, it’s well understood by now that there is an element of unpredictability and fluctuation in the availability and consistency of volume from specific publishers or sources of ad inventory. This can often lead to frustration, especially when trying to optimize and automate your campaigns.

To appreciate why this happens, it helps to understand how ad inventory becomes available in the RTB ecosystem, how direct sales orders (aka insertion orders) affect the amount of inventory available, and the interplay between the two sales channels for website publishers.

The Publisher’s Dilemma

Website publishers are in a tough position. They have a strong desire to increase their advertising revenue with as little headache as possible. For publishers needing a quick revenue solution, outsourcing the whole process to Google AdSense has typically been the easiest way to go. The problems with AdSense, however, are that it’s restricted to buyers on the AdWords platform, the revenue model lacks transparency, and the revenue yield to the publisher is low.

More savvy publishers — often with larger amounts of traffic and higher revenue goals — either turn to selling their inventory directly to advertisers through direct ad sales, or by outsourcing it to an ad network that handles it on their behalf. Direct ad sales typically yield the highest RPM (Revenue Per Mille — or revenue per thousand pageviews) for publishers, especially if their inventory is highly sought-after. Direct sales usually require more technology, like an ad server, as well as people to handle ad operations (like setting up campaigns, updating creatives and managing clients).

The downside to this process is that it’s very inefficient for both publishers and advertisers: it requires numerous manual tasks, such as negotiating and transmitting contracts, trafficking campaigns, billing, and so on. Despite these challenges, publishers continue to prefer direct ad sales for the simple fact that this yields the highest revenues.

Yet, most publishers are unable to sell off their inventory directly. So, what to do with this unsold (or, in industry-parlance, remnant) inventory? The answer, traditionally, was Google AdSense (which makes this unsold inventory available to AdWords advertisers). But, now it is possible for publishers to expose their unsold inventory to the RTB ecosystem to auction it off indirectly to the highest bidder, getting the highest RPM possible.

Revisiting Session Depth

In my last article, which discussed the mechanics of real-time bidding, I touched on the subject of session depth and how the first impressions seen by the RTB channel are merely the first impressions after direct sales orders have been fulfilled.

As we can see in the illustration above, the lowest possible session depths are always served through direct ad sales along with the true first impressions. Moreover, in the fictional example above, if the average visitor accounts for less than 10 pageviews, then the chance of them being available to the RTB auction becomes very low. In practice, it is entirely dependent on the order sizes and campaign rules of direct buy advertisers.

Flux In The RTB Ecosystem

This natural dynamic between direct and indirect (RTB) ad sales creates an inevitable fluctuation in inventory volume for the RTB ecosystem.

For example, during peak advertising seasons like Christmas, the influx of advertising dollars, especially in the direct sales channels, creates lower volumes of inventory in the RTB channel. In some cases, publishers effectively “disappear” from the RTB ecosystem completely if their entire inventory gets pre-sold from direct deals. After all, display ad space is a finite and ephemeral commodity online.

During the last holiday season, there was a drastic spike in the cost of RTB inventory following Black Friday. In that case, the increase of advertising dollars went beyond the direct sales channel and spilled over into RTB, causing the increase of eCPMs across the board.

That combination of lower inventory volumes with higher inventory prices forced marketers to drastically increase their bids in order to compete for what was available, making it challenging for many performance-based marketers to maintain their volumes and ROI.

Continuing The Discussion

I hope this provides some clarity as to the relationship between these two different sales channels, and the economic motivation that drives publishers to segment their inventory in such ways. In my next article, I will continue to discuss the differences between these two approaches to display advertising and how these differences impact marketing strategies.

You see, data is the lowest level of abstraction. On its own, Data carries no real meaning. Data viewed in context then produces information. Information tested and interpreted then results in knowledge.

Experimental Data

In online advertising, there is a potential treasure trove of audience data. The lowest level of known variables that are then mixed and matched determine one of two things:

Individual Intent:  The specific purpose for why an individual would perform a certain act or series or acts. A prediction of what result or end an individual is seeking. Common examples, among hundreds, include keywords searched, websites visited and content read.

Individual Interest:  A predication of how thoughts, beliefs, feelings and perceptions influence how people buy and relate to goods and services. In marketing, common variables, among thousands used to predict interest,can include gender, geography, income, race, age and much more.

When viewed in this context, audience data can be properly defined as “Experimental Data.” Experimental data is data generated within the context of a scientific investigation by observation and recording. This is the fuel that drives Programmatic or algorithmic auto optimization models popular among demand-side platforms and many alternative display channels.

Group Data Segments = Information Loss

I have noted in prior articles, and it is common knowledge among online advertising professionals, that the vast majority of display platforms in practice today are not in the data collection business. They are in the data consumption business. They purchase pre-packaged interest or intent “Audience Segments.” This is done for a fee, of course, and the bigger the audience in the segment, the more the data brokers can profit. Are you concerned yet?

A common segment, for example, would be “Cruise Intenders.” This may be an audience segment of 3 million unique individuals currently open to or actively seeking to go on a cruise.  The probable data in this group segment could include a spectrum of data elements from individuals who have searched on cruise specific keywords or visited travel websites, to individuals of a certain age and income. These individual consumers are hardly equal in the level of their interest or intent.

Enter the display platforms with machine learning. “Machine learning” and “Patent Pending Algorithms” are phrases we are all force-fed. In reality, most of these companies have only automated the dated practice of ignoring the individual audience variables and optimizing the traditional group segments such as: Which sites should be blacklisted from this group? At what times of day are this group converting best? What ad frequency is ideal for this group?

There is a significant loss of valuable information when data is packaged into these groups. What individual behaviors are behind the highest converting and engaged consumers that are on the receiving end of your campaign? Imagine the knowledge and power that could be harnessed from programmatic media buying if you could see and control every data variable as if it were its very own segment!

The Most Valuable Data Variable In Advertising

In an effort to see if we can reach the same conclusion, I will begin with a short quiz for your testing pleasure:

Which cake is likely to taste best?

A)  A cake baked just minutes ago

B)  A cake baked 8 days ago

C)   A cake baked 3 weeks ago

D)  A cake whose baked date is a complete mystery

You are a car dealer with the world’s greatest sales force. A family of four is in desperate need of a new vehicle, and they walk onto your lot. Your odds of securing their business are best when:

A)  The family is currently on your lot

B)  The family left your lot 24 hours ago

C)   The family left your lot 2 weeks ago

D)  You have no idea when they visited your lot

You are a single location restaurant specializing in pizza. You want new customers to try your pizza. You select an advertising vendor who can serve ads to anyone who has searched on “pizza coupon” in your city. Which person is most likely to be influenced by your ad?

A)  A consumer who performed the search 5 minutes ago

B)  A consumer who performed the search 5 days ago

C)   A consumer who performed the search 2 weeks ago.

D)  A consumer who performed search in unknown 30-day window.

By now, you know the obvious truth is that the variable of “Time” can dramatically change the meaning behind any data element. In 90%+ scenarios it is the most recent data variable that contains the most predictable truth.

There are going to be dramatic differences between messaging to someone actively searching in the last 24 hours when compared to weeks ago. Alas, major brands and the agencies who support them continue to pour dollars into companies who can neither track nor bid differently for individual impressions within their group segment based on this information. All data within the segment are subjected to the same bid and same traditional optimization techniques.

Unstructured Data = Recency Insight & Control

At my company, Simpl.fi, we have found it critical to gather each data element behind every impression — What keyword was searched?  What website was visited?  What subject matter read? — rather than being limited by segments.

Alone, this is good information. Now, introduce the knowledge of when the search was performed and when the subject matter was consumed, combined with the ability to bid differently based on the recency of data. We have found that this level of insight and control is invaluable across direct response and branding campaigns leveraging programmatic media buying.

Here are some findings regarding “Data Recency” taken from more than 200 recent display campaigns for clients in the personal finance industry:

• We have found, for instance, that Click Through Rates (CTR) and conversion rates drop significantly once the data element used to opt the consumer into the campaign ages beyond 24 hours.

• The same data when examined at the lowest attribute level, in this case “keyword,” showed that optimum click and conversion rates vary dramatically based on if the search occurred within minutes vs. the last 24 hours:

• Another interesting finding is that when attempting to maximize conversions while achieving the best possible Costs Per Click (CPC), it is vital to know the recency of the data and to have a platform that can programmatically adjust bidding on highest converting elements and the age of the data. In the chart below, you can see how a sweet spot developed for achieving the best CPC between the aggressive bids tied to “Instant” recency and the diminishing clicks and conversions as the data signal ages

Summary

“Time is what we want most, but what we use worst.”  ― William Penn

Time is precious, and never has this been truer than when attempting to string together data elements to predict the interest and intent of the audience you seeking to influence. We are moving into a new era of big data, when old rules around processing and grouping people into generalizations will reach their limitations quickly.

Unstructured data, vendor transparency and algorithm-driven platforms that can control the time variable at the smallest level will mean deeper insight and better performance.