I’ve been thinking a bit about targeting, and how we in the web analytics industry have just a ton of visitor or segment-level data that can be used for targeting ads or content, but most tools don’t let you use the data or easily feed it to other systems to do any targeting.  It’s rather odd, don’t you think?   Even Omniture Test and Target isn’t using, as far as I’ve learned, a single data model or the data collected from their behavioral tools, like HBX or SiteCatalyst, for targeting.  All their data models and thus, their data, are unique to the products in their platform.   So I decided to resussitate/revise a blogviation and offer it as food for thought on MediaPost.  When I reread this post, it’s more of an informational post for product managers on how I’d begin thinking about targeting with analytics data and what types of targeting are possible, so here it goes.   

Targeting refers to the process of delivering content or ads to segments or visitors based on their known attributes.  The goal of targeting is simple to understand: maximizing the performance of content or an ad by serving it to visitors at a time when they are most open to the receiving the message. 

For example, you may visit a site, and see some type of ad unit calling out at you to “meet singles in <insert_your_city>.”  When browsing a real estate site, you may see ad units for realtors and mortgage companies.  After entering a keyword such as “car insurance” and clicking through the search results, you may land on a site and see an ad for a car insurance company or land on a page that persuades you to begin the process for creating an insurance price quote.  That’s targeting in a nutshell.  It’s simple for a site owner to understand:

1. Visitor X has these attributes.  
2. We have content or an ad that we think will appeal to Visitor X’s attributes. 
3. Let’s show the relevant content or ad. 

In online media, targeting is associated with paid search campaigning, ad serving, and content optimization based on recognizing and responding to the following attributes:

• Category and sub-category.  Conceptual constructs like “categories” of topics on a media web site or products on an ecommerce site can be targeted to include certain types of ads or messages.   The idea is that if visitors are browsing your category for “hardware floors,” you could offer them an ad or content specific to “flooring installation services.” 
• Geography.  Country, region, city, state, DMA are all targetable constructs.  You may run a sports site and choose to target people surfing in from 02116 (Boston) an ad for Red Sox tickets or content about Manny Ramirez’s recent trade to the Dodgers.
• Browsing environment such as the connection speed, type of browser, operating system, user software, domain, and ISP.  An ad network could serve an ad for DSL to a modem-based surfer by detecting the visitor’s browsing environment.
• Time.  The idea of only showing content during specific periods of time is called “parting.”  Common types include day-parting and season-parting.  For example, a B2B site only choosing to show ads for a particular manufacturer’s product during business hours — the site’s busiest time of day — would be an example of day-parting.
• Keyword.  There are many different types of keyword targeting.  Search engines target ads based on keywords in queries.  Content Management Systems target content based on site search keywords or referring keywords.  “Keywords” may be associated as metadata with site sections or pages, similar to zone or category targeting on an ad server.  Once a page is associated with keyword metadata in an ad tag, you can tell your ad server to target ads to that keyword on whatever page or pages the tag was placed. 
• Language.  When a language can be detected or known in advance, you can target ads to visitors in their language.
• Demographics. If the ad server is aware of a segment’s demographics, such as age, gender, income, title, purchasing power, and so on, an ad can be targeted on that basis. 
• Context.  Think of AdSense and how it matches text ads based on the semantics in site content.  Or when, after adding a product to your cart, a site offers you “free shipping” if your total purchase exceeds a certain price.  This is content targeting based on context.
• Profile.  Targeting is possible based on conclusions drawn and rules created from attributes about an individual or segment (such as purchasing propensity or job title).
• Rules.  Serve an interstitial ad only to visitors who don’t have a cookie set for the site.
• Events.  Someone deposits a large sum of money into his bank account, so the online banking site offers him a CD product on his next login.

We’ve all heard, of course, about a very specific type of often-discussed targeting in online advertising: “behavioral targeting.”  Behavioral targeting refers to the technology and process in which an ad or content is shown to a visitor based on their past actions and reactions.

Behavioral targeting involves:

1. Collecting behavioral data about visitors.
2. Identifying when those visitors visit a site.
3. Determining the current context of visitors on the site.  
4. Detecting the visitor’s current behavior.
5. Serving relevant ads (or content) matched to the behavior.

The goal being to use past behavioral data to influence the customer buying cycle or marketing lifecycle, in order to more effectively and more quickly deliver on advertiser and site goals.

So where does Web analytics come in?  You would think Web analytics data from “Web analytics” technology would be used to enabling “targeting.”  After all the best Web analytics systems store detailed visitor level data about past behavior.  Web analytics data certainly can be used, but in most cases, targeting is a function provided by the ad server or network, perhaps the ISP, or another technology called the “behavioral targeting platform,” not from data collected by the Web analytics tool.

In order to make Web analytics data useful for targeting, you will need to use your data to:

1. Define segments to target or identify visitors to target.
2. Feed past behavioral data about segments or visitors to the targeting technology.
3. Analyze segment and visitor performance against site or advertiser goals after targeting.

Targeting has a proven ability and amazing potential to generate tremendous returns, especially when combined with the rich, detailed behavioral data available in Web analytics.  As a method for optimizing site content and advertising, targeting technologies that integrate with Web analytics data will only become more important and a necessary “must have” for innovative companies that want to maximize business opportunities on the Internet. 

From an article I wrote for MediaPost a few weeks ago:

Reach and frequency and the core concepts of traditional media planning and advertising.  For a given site, program, channel, radio station, billboard, newspaper section, a target audience (the reach) is exposed to a certain number of occurrences of the media (the frequency).  On the web, these concepts manifest themselves in metrics collected and reported from a number of recognizable services.  Audience measurement firms, like comScore and Nielsen, web analytics firms, like Omniture and Unica, to companies somewhere in between, like Quantcast and Google, all have reach and frequency data.  Many new media metrics can be used to proxy frequency- from time-based measures, espoused by audience measurement firms, to concepts like visitor retention or the repeat visitor rate cited by web analytics firms.  On the reach side, companies refer to concepts like “unique visitors.”

These data, of course, available in free tools or in for pay tools are certainly helpful for planning campaigns.  But reach measures can be dirty (cookies, unduplicated unique users, estimates from panels, coverage error).  Frequency measures can be just as dirty (problems recording time in single page visits or visits on the last page, do page views really matter with AJAX and rich media, cookies again, and so on).  We all are aware of the challenges.

Thus using basic reach and frequency measures for planning or evaluating a campaign does not suffice.   So advertisers and agencies target demographics, like gender, age, income, education, and job title.  It’s a given that advertising in the Robb Report reaches a different audience segment than advertising in Popular Mechanics. 

These brave new days we have “behavioral” tracking too.  By taking into account visitor activity across sessions, such as past actions taken on a site or a roster of previous purchases, we can attempt to deduce what a person or segment responds to or is interested in based on their behavior.

Even with reach, frequency, demographics, and behavioral data to help guide advertising and media buying, we are missing an important attribute for maximizing the potential success of our campaigns.  We do not have an available tool, whether free or paid, for advertising or buying media on or across sites according to measures of past performance.  Such measures include ad clickthrough rates, conversion rates, goal completion rates, delivered impressions, and perhaps even harder to quantify financial measures such as ROI, ROAS, and ROMI.

Sure, historic, tacit knowledge of campaign performance exists and is used by agencies or publishers.  However, there is no shared industry source that can help us answer “how has a site for display advertisement historically performed toward goals based on the reach, frequency, demographic and behavior of its audience segments?”  Interestingly, a company minting money right now, named Google, can masterfully demonstrate performance in paid search campaigning and help advertisers unify it with segmented reach, frequency, and demographics.

Outcomes based performance measurement unified with reach, frequency, demographics, and behavior is what is missing in audience measurement tools, not frequently reported externally by web analytics tools or ad serving tools, and not available in ad planning tools.  When advertisers can target display ads, or even video ads, to desired audience segments by reach, frequency, demographics, behavior in the context of known performance, media planning will be more effective.  

One of the many things on my mind in the online world these days is “deep packet inspection.” 

First, let me digress, packet sniffing isn’t new to web analytics.  From Accrue to Omniture (Visual Discover Sensor?) to AuriQ to Metronome Labs.  Packet sniffers are used to “do web analytics.”  It’s an uncommon method when compared to javascript page tags.

Web analytics packet sniffers are used to write logs for sessionization (and thus measure) the traffic on behalf of site owners (who don’t want to use tags or logs).  Once you’ve logged and sessionized you know what content people have looked at or downloaded on your site. 

“Deep packet inspection,” like WA sniffers looks at the entire payloadof packets in real-time across a huge number of simultaneous sessions.  Deep packet inspection, like regular packet sniffing, examines the files downloaded and the content of the pages viewed - the whole ball of wax. 

Deep packet inspection is being offered as a hardware/software technology by companies like FrontPorch and Sandvine (in the US) and Phorm(in the UK).  These companies are selling the technology to ISP’s (like Charter, Comcast, and Virgin Media) so that they can monitor the sites visited and the keywords used by customers, and then use the data collected for behavioral targeting.  The ISP’s want a slice of the juicy, lucrative online ad business.

What’s the difference?  Site owners collect data about what you do on ONE site (or a portfolio of their sites).  ISP’s collect data about what you do on EVERY site you visit.  As I understand it, some of these companies create an anonymous profile of your surfing activity by assigning a unique key to your browser.  Then they monitor the site’s visited by your browser, and use that data so that the ISP, or the companies to which they sell your data, can serve you what they conclude to be relevant, behaviorally targeted ads. 

Get it?  Packet sniffing by site owners = knowing about one site you visit.  Deep packet inspection by ISP’s = knowing about every site you visit.

Now to digress… In web analytics, we know that web analytics data is collected anonymously.  Unless there’s a login, you don’t know exactly who is coming from that IP address.  And in many cases, most companies data warehouses only contain purchase information, not the entire clickstream.  Once the data is collected, if you have the right architectures you can decode cookie values to people, and make that data non-anonymous (i.e PII).  Not difficult to do with some smart BI folks on your side.  

An ISP already knows who you are and can already identify the sites you visit.  Probably not that easily though on individual level.  They can dig through the logs, etc… 

So what’s the big deal and all the hoo-hah about  the “deep packet inspection” Phorm and FrontPorch are doing?   It’s the data they are collecting and the repository they are building containing data about every site you visit and all the content you view and download… Of course, these companies say that it’s all done anonymously and that your “privacy” is preserved “to the greatest extent possible.” 

Now let me quote Sir Tim Berners-Lee about the data collected from Phorm’s ISP tracking: “It’s mine - you can’t have it. If you want to use it for something, then you have to negotiate with me. I have to agree, I have to understand what I’m getting in return.”

And that’s the point of the blogviation, Tim is correct.  In web analytics, we do this - we try to operate within Tim’s constraints.  We enable opt-in with P3P statements and disclosures when you register/login.  Privacy policies disclose what we are doing with the data.  It’s just ethical and smart business practice to do so.

Thus, I think FrontPorch and Phorm and all the ISP’s who want a piece of online advertising should adhere to the following five rules for their services.

1. Move to an obvious “opt-in” model with full disclosure.  Tracking via “deep packet inspection” should be an all opt-in model.  If you want anonymous data from your browser collected so that you can be behaviorally targeted, then you should opt-in to be.  Right now, it’s seems to be all opt-out.  You probably don’t know if it’s being done to you.  It’s buried in fine print you’ve probably never read.  Is that your fault you didn’t read the fine print? Yeah, but the point is it shouldn’t be buried in the fine print…
2. Provide me with access to the data collected.  If I opt-in, I should be able to see the data collected from my browser.  It’s very simple.  I demand to see what you are collecting about my browser.  If you are building a profile, then I demand to see the data collected in the profile.  If it’s all anonymous, then explain how it is in detail, and then follow rule #1.
3. Enable me to edit or prevent the data from being collected.  If I opt-in, I want to be able to edit or prevent certain types of data from being collected.  If you’re tracking my browser, alert me before the data is transmitted, so I can decide if I want to share it.  If a profile is built, I want to be able to edit it!
4. Let me opt-out at any time EASILY. If I’ve opted in, and I’m unhappy with the service, allow me to opt-out simply.  Having to set an opt-out cookie on my browser is absolutely and completely absurd.  I want to be able to fully opt-out at the ISP level, just once forever, not at the browser level every time cookies are deleted.  Make it easy and permanent, not easily deletable.
5. Disclose who you sell my data too.  Like online list rentals, the next step in all this ISP profiling is selling the data to third-parties.  Let me know what you’re doing with my data-before you do it- so I can opt out or prevent it from being sold to parties to which I don’t want it being sold.

Consumers must be given a choice for preserving their privacy.  Anonymity to the “greatest extent possible” is not enough and neither are short-sighted opt-out cookies.  Companies like Phorm and Front Porch would be wise to apply these rules to regulate themselves.  Otherwise freedom-loving governments will almost certainly regulate them. 

And I haven’t even mentioned the issues with net neutrality and deep packet inspection (i.e. traffic shaping and access restrictions (called “throttling” as Clint points out in the comment), have I?

Targeting refers to the process of identifying characteristics of a segment so that relevant content may be matched to it and delivered at a time when the segment is most open to the message. The idea is the right content to the right visitor at the right time (optimally in real time). 

For example, you may visit a site, and see some type of ad unit calling out at you to “meet singles in <insert_your_city>.” When browsing real estate you may see ad units for realtors and mortgage companies.  After entering a keyword such as “car prices” and clickingthrough the SERP, you may see an ad for a local car dealer.   That’s targeting in a nutshell.  It’s simple: 

1. Visitor X has these attributes. 
2. We have content that we think will appeal to Vistor X’s attributes. 
3. Let’s show that content. 

While targeting has helped to increase ad clickthrough rates, it’s far from an ideal science.  Current methods for targeting have inefficiencies.  What if Visitor X just bought a new car after his recent marriage?  Unless the targeting engine is made aware of the visitor’s current state, the targeting may be off and not yield desired results. 

Even with limitations around “current awareness” targeting is perceived in the Internet industry as a crucial activity for maximizing the effectiveness of advertising and content.  Targeting is the next stage after A/B and multivariate testing.  Once you determine the preference of segments based on testing, you identify content to target. 

In new media, targeting is something associated with paid search campaigning, ad serving, and content optimization.  It’s not uncommon for targeting activities to be based on:

• Category and sub-category.  Conceptual constructs like “categories” of topics on a media web site or products on an ecommerce site can be targeted to include certain types of ads or messages.  The notion of a “zone” fits in here as well.  The idea is that if visitors are browsing in your category for “hardware floors” you could offer them an ad or content specific to “flooring installation services.” 
• Geography.  Country, region, city, state, DMA are all targetable constructs.  You may choose target people surfing from 02141 (Cambridge, MA) an ad for pre-sale Red Sox tix or content about Mike Lowell’s recent contract.
• Browsing environment such as the connection speed, type of browser, operating system, user software, domain, and ISP.  An ad network serves an ad for Verizon DSL to a modem-based surfer by detecting the visitor’s browsing environment.
• Time.  The idea of only showing content during specific periods of time is called “parting.”  Common types include day-parting and season-parting.  For example, a B2B site only choosing to show ads for a particular manufacturers product during business hours – the site’s busiest time of day – would be an example of day parting.
• Keyword.  There are many different types of keyword targeting.  Google does fantastic things with targeting ads based on the keywords in queries.  Content Management Systems can target content based on on-site search keywords or referring keywords.  “Keywords” may be associated as metadata with site sections or pages, similar to a zone or a category targeting on an ad server.  Once a page is associated with “keyword” metadata, you can tell your server to target that keyword (and all pages where it exists as metadata).  If two categories each with different content share a targetable keyword, I can target ads across both categories to pages tagged with that specific keyword.
• Language.  When a language is set, you can target ads to visitors with that setting. Think Google.  Keep in mind that when you target by language, the creative copy is not translated. 
• Demographics. If the ad server is aware of a segment’s demographics, such as age, gender, income, title, purchasing power, and so on, an ad can be targeted on that basis.  Sometimes this is called “profile targeting.”
• Context.  Think of Google AdSense and how it matches ads based on the semantics in site content.  Now you understand content targeting based on context.
• Profile.  Targeting is possible based on conclusions drawn and rules created from the known attributes (such as purchasing propensity) about and individual or segment.

Enter one of the holy grails of online advertising and new media: “behavioral targeting” – an advanced form of targeting. Behavioral targeting refers to the process in which content is shown to a visitor based on the web sites they visit (or have visited) and the actions they take on those sites.  

Behavioral targeting involves:

1. Knowing where a visitor “comes from” and what they’ve done in the past. 
2. Determining the context of the visitor on the site. 
3. Detecting the visitor’s current behavior.
4. Serving relevant content and/or ads matched to the behavior.

By understanding the visitor’s past history, current state, and most recent behavior the marketer can target content in order to influence some point in the customer buying cycle- often at the stages of awareness and consideration.

So where does web analytics come in?  You would think web analytics data from “web analytics” technology would provide the seed data for enabling “targeting.”  It can be but in most cases, targeting is a function provided by the ad server or network or another technology called the “behavioral targeting platform,” not the analytics tool… the data does not come directly from the web analytics tool.  I’d love to hear how well (or if at all) Omniture TouchClarity is integrated with Omniture Discover or other offerings. 

In order to make web analytics data useful for targeting (if you can at all), you will need to use your web analytics data to:

1. Define segments to target (hard to export from web analytics tools)
2. Feed those segments and associated behavioral data to another tool (achievable if you own your data and run a tool in-house.  Harder and more costly if not).
3. Report on segment performance after targeting (that requires employing the right people and enabling them with the right tools)..
4. Analyze segment performance after targeting (again employ the right people and enable them with the appropriate tools and resources).

While I’ve only covered a very little bit about “targeting” and even less about “behavioral targeting” in the context of web analytics, I hope that my simple description of current methods for targeting and some thinking about “what is BT” will help you understand the emerging ecosystem in which analytics tool are interoperating now and will interoperate in the future.