Razorfish Search Shots

Currently viewing Department of Searchology posts.

Google Display Network Targeting

Monday, October 24th, 2011

Background

The Google Display Network (GDN, formerly referred to as the Google Content Network) has an extremely large inventory pool of sites across the internet. GDN was initially launched on October 23, 2000, and in more than a decade has grown to one of the largest online advertising properties in the world.  It is estimated that this network reaches 89% of the internet in the U.S., with over 1 million publishers and 211 million unique users per week (comScore Networks machine-based panel). Paid search and display media ads can be served across this network, and audiences can be targeted in several different ways. Ads are served alongside content specified by the advertiser. This brief will take a deep dive into the targeting capabilities of the GDN, and the benefits of the GDN for a paid search advertiser.

GDN and Digital Advertising

Paid search advertising and display media advertising both have the opportunity to advertise within the GDN. The main difference between these two mediums is cost structure. Display media is usually bought on a cost-per-thousand (CPM) basis, meaning the advertiser pays each time 1,000 impressions are served. Thus, each advertiser’s display media impression must be a valuable placement.

Paid search advertising is usually purchased on a cost-per-click (CPC) basis. This means that the advertiser only pays when their ad is clicked on. With this cost structure, there is more flexibility in what sites these ads are placed. If the site is not compatible with the ad, then the ad will not get served and no cost is incurred. Paid search advertising using the GDN is an excellent way for an advertiser to reach a greater audience and still maintain efficiencies.

While targeting is critical for both types of digital advertising to reach the right audience at the right time, the implications of highly specific targeting are usually more essential for display media to ensure that impressions are not lost on an extraneous audience. However, all targeting options in the GDN are available for purchase on a CPM or CPC basis for both text and display ads, depending on the advertiser’s goals.

Types of GDN Targeting

Contextual Targeting - selecting specific keywords and/or topics where the advertiser would like an ad to appear. Contextual targeting is done on the page level, not the site level for maximum relevancy.

Keyword Contextual Targeting – advertisers select certain keywords that are relevant to them, and bid to appear alongside this content. This ad may appear on any site across the GDN where there are those keywords on the page. The scale of this method of advertising could be very large, depending on the keywords that are being targeted. It is usually recommended to layer this type of targeting with another method to increase relevancy and minimize waste.

Topic Contextual Targeting – advertisers select certain topics that are relevant to them, and bid for their ad to appear on pages of these sites.  This ad may appear on sites across the GDN that are categorized under that topic. This method is very broad-reaching as well, and is usually recommended in combination with another targeting method for an advertiser interested in reaching a specific audience.

Placement Targeting – advertisers select certain sites and/or sections of sites that are relevant to them, and bid for their ad to appear on pages of these sites. These sites can be selected by the advertiser using Google Tools such as Ad Planner, which uses Nielson data to index sites in the GDN based on:

  • Demographics (Household Income, Age, Gender, Education)
  • Online Activity (Other Sites Your Audience Visits, Keywords Your Audience Searches For)
  • Interest Categories (i.e. Cooking & Recipes, Women’s Interests, Weddings)

In practice, Razorfish usually finds this method to be the most successful approach to the GDN, because sites/sections that index highly against a target market can be cherry-picked for extremely relevant targeting.

Behavior Targeting – Advertisers select certain topics that are relevant to them, and bid for their ad to appear across the GDN to users who match those interests. This method can be used to reach a large audience as well as a more targeted, niche audience. A user’s interests are either declared interests (through the Ad Preferences Manager), or are inferred based on their browsing behavior, specifically their recent and frequent site visits. This method of targeting is usually used for broad-reaching awareness campaigns or advertisers that seek site visitors that abandoned part of an intent funnel.

Inferred Demographic Targeting – advertisers bid on an audience where Google has inferred their demographic based on their GDN history. A user’s demographic is determined by a number of sources, including user registration data, 3rd Party data and site composition. The registration data that is used in Inferred Demographic Targeting may come from YouTube registration, or other undisclosed sites in the GDN that capture registration information. Specific targeting sources cannot be cherry-picked, and Google takes all into account when inferring a demographic. The composition index of a site determines the inferred demographic. For example, if a person visits a fashion site, and then visits a parenting site, then Google may infer the demographic as a Female 25-54. If an advertiser uses Inferred Demographic Targeting for this target, then Google will serve an ad in the GDN network to that person. Google is constantly improving the dataset used to determine user Demographics and will incorporate new data sources as they become available. Additionally, this feature is currently in Beta and advertisers must request to be whitelisted by Google to participate.

This newer method of targeting may have benefits for an advertiser that has an extremely specific audience they wish to target, and should be tested alongside other GDN methods. However, because demo-inferred targeting is still in beta and takes very little user self-identification into account, it should not be used exclusively as a preferred GDN targeting method without the support of testing and analysis alongside the other methods.

Below is an example of how an advertiser can use a Google tool, Ad Preferences Manager, to target select audiences.

Remarketing – advertisers bid on an audience that visited their site (or a site which will allow them to implement a pixel) and their ad is displayed across the GDN. This method is often used for CRM marketing, or if the visitor was in the middle of a conversion funnel and then left. In the example given below, Special K can remarket those that visited the Special K Challenge Registration page but did not fulfill registration.

Hybrid GDN Targeting - GDN Targeting products can be combined together to reach a very specific, desired audience. As targeting layers increase, an advertiser will be reaching a more specific audience and a smaller percentage of the total population.

GDN Benefits for Paid Search

With the targeting capabilities and mass reach of GDN, there is a greater opportunity to expand paid search marketing campaigns. Expanding an advertiser’s paid search marketing presence will lead to impactful benefits such as:

1. Efficiencies – It has been established with prior campaign history that paid search is one of the most efficient means of advertising. However, paid search on sponsored search (i.e. Google.com) can sometimes be expensive if an advertiser is bidding on keywords where there are many other competitors, which may increase CPC and overall cost significantly. Including GDN and network targeting can be essential for an advertiser with many competitors who is concerned with efficiencies such as CPC and CPA (cost-per-action).

2. Reach – there are a finite number of searches for a set of keywords, and search trends must increase if an advertiser wants to expand their paid search marketing efforts. GDN offers an opportunity to expand reach beyond basic sponsored search results.

3. Relevancies – GDN is an opportunity for an advertiser to appear alongside relevant content or a target audience efficiently.

4. Testing - GDN is a way test sites and targets with minimal cost commitment (budgets can be set as low as $1 a day for only one day)

5. Turn-Key Implementation – ads are the same format as traditional paid search ads (130 characters) and can be created quicker than other advertising creative.

The Facebook “Like” Cycle

Tuesday, October 11th, 2011

The Foundational Approach to Facebook CPC Ads

Executing for a Single Interaction

When developing a foundational Facebook media plan, you need to incorporate the needs of the client request and a desired goal into the strategy.  A project brief and planning session can help create the framework of the campaign and answer questions, such as:

  • Target audience (demographics:  M (25-49), educated, in GEO, with certain behaviors)
  • Product information (price, seasonality, use cases)
  • Goal (build a fan base, drive conversions on a website)

 

With this information you devise a marketing plan as usual, while utilizing Facebook’s vast amounts of targeting options to create detailed audience segments and match the right messaging to these segments.  By segmenting your audiences clearly in Facebook (or via a management tool like Marin’s Facebook Management Solution), you can efficiently manage and optimize your segments easily and:

  • Measure performance by audience including location, age, likes & interests
  • Filter new ads by targeting criteria
  • Create automated bidding strategies by segment to desired goal

The execution of a foundational Facebook CPC ads campaign is to support a relevant set of placements for a SINGLE interaction. (Search Marketing minded approach).

Executing on Paid & Social Synergies

Executing for a Series of Interactions

To effectively target different segments that may have different goals (engagement vs. conversion) at different stages of a purchase cycle (or seasonality), you can build a more complex plan that feeds off and builds from its own momentum.    Unlike the foundational Facebook campaign that drives to a single desired action, this approach incorporates a series of consumer interactions.

It all starts with a Facebook ad with the goal of driving Fans and “Likes”.  This is where Paid & Social Media collide and can do powerful things together.  The Facebook CPC ads in a more intricate model take the consumer mindset and social interaction stage into account to drive a multi-stage cycle.

 

Introducing the Facebook “Like” Cycle

 


Facebook ads have a lot of opportunity to reach, communicate and interact with audiences within its own social environment, or cycle.  Break out of the mold of thinking of simply the foundation but applying a more complex and dynamic campaign execution that speaks to consumers at different stages of their brand engagement.

Let’s break down each segment further and discuss the possible consumer mindset within each:

1. AWARENESS:  Build a Fan Base

This might be the most obvious but could also be shortsighted and capped of its extended benefits.  Serving Facebook CPC ads (Sponsored Stories Ads, CPC ads or Video ads) to increase traffic, are valid ad types to make consumers aware of your Brand.  The capability of directly “Like”ing your brand fuels your Fan base that is not only a pool of loyal customers that can develop and share positive sentiments of your brand, but those positive sentiments can then be turned into your Sponsored Stories ads in your campaign.

Additionally, adding “Fans” to your brand’s Facebook page has more value than just counting the #s of likes.  Driving Facebook Likes can also benefit your SEO program as “Likes are the new Links.”  Further value are mentioned in the next segments of the “Like” Cycle.


2.  dWOM:  Facebook Interaction and Engagement

As the Fan base to your Facebook page grows the fans will inevitably voice their opinions and interact with the fan page.  The positive sentiments and stories (even simple “Likes”) can become Sponsored Stories ads that will show to a Fans’ Facebook Friends which acts like Digital Word of Mouth (dWOM).  This ad type brings a new power to online advertising as it integrates dWOM into its copy that it isn’t surprising to hear Facebook claims Sponsored Stories ads “perform twice as well in engaging users” compared to standard CPC ads on the network.

Facebook recently announced at AdWeek its belief in Facebook WOM and belief “that word-of-mouth conversations among friends are the most influential for getting a brand’s message across.”  They went on to cite comScore research “showing that fans and friends-of-fans of a Page are more likely to visit a store, website, and even purchase a product or service.”  For instance, “Fans and friends-of-fans of Starbucks spend 8% more in stores than the average Starbucks customer and transact 11% more frequently”.  (Source: Fast Company: Oct 3, 2011)

3. ACTION:   Convert Your Audiences

Now that you have an established (and growing) Facebook Fan Base and built out audience segments to target with Facebook you can run a Facebook CPC ads campaign with a direct response message that leads the consumer to a desired action.  This campaign can build off the momentum of the first 2 cycle segments that drove Fans and inspired Engagement with the brand to build demographic targeting segments.

 

4. CRM:  Consumer Retention

As you continue to build the fan base and loyal customers, you need to retain them and continue the conversation and build further loyalty.  From here you have access to the Fans through creating a Facebook email strategy or implement contests and promotions on Facebook.  Additionally, Sponsored Stories ads can be run throughout the year to not only drive new Fans or “Likes” but also to leverage positive sentiments and push those stories and wall posts to Friends of Friends as a Digital Word of Mouth (DWOM) campaign.

Conclusion:

Executing for a Series of Interactions:

The PPC Ads Facebook “Like” Cycle has the ability to continuously evolve, grow and drive results for your Brand.  The “Like” Cycle requires executing placements for a SERIES of interactions that aren’t easily identified in advance, but if planned for strategically can result in more sales.

It is a complex ecosystem that takes the basics of targeting, buying and optimizing to a new level of campaigning that is dynamic, complex and potentially more rewarding.  With Facebook’s movement of viewing the act of online Sharing as an indication of Value, it is a testament of the “Like” Cycle and how marketers should adapt to this new model.

Think through your own Brand’s campaign and how the “Like” Cycle fits your Brand’s needs.

Caveats:

* People can enter the cycle at any moment, not necessarily at #1.

**This cycle has many “legs” to it and doesn’t fully represent Social Media’s engagement and interaction experiences and benefits

***This cycle is a depiction of the number of ways to reach, engage and retain customers through Facebook ads and interactions.  We realize this cycle is not perfect or one size fits all.

Google Analytics Premium, Attribution Modeling, and Right Now

Friday, September 30th, 2011

Google Analytics Premium: An enterprise platform

Google has officially announced Google Analytics Premium! Google’s first enterprise analytics product will be available directly from Google and many Google Analytics Premium Certified Resellers. They apply a more simplified pricing approach compared to the confusing contract pricing which is based on server calls and options that the rest of the enterprise industry has established. The especially interesting thing is this flat fee could result in saving money for larger clients switching from Omniture or Webtrends; however, this could also end up being more expensive for medium-sized clients looking to move to Google Analytics Premium. Regardless, it’s important to weigh the Pros and Cons of each product’s features and perform an analysis of whether the investment is worth the price tag. Here are the key features of your Premium investment:


In addition to requiring no code change to upgrade from Standard to Premium, one great aspect of Google Analytics Premium is that the interface look and feel hasn’t changed at all. The only change is the addition of the Unsampled Report export button:

Attribution Modeling (PREMIUM ONLY): Advanced analysis, simplified

Although it’s limited to only Google Analytics Premium customers, this advanced analysis tool provides powerful multi-channel attribution that actually seems easy to use! If your client is a DART For Advertisers customer then this tool will prove even more powerful since media impressions can be weighted in the model as well. There are several different types of attribution models that a user can select or compare based on their needs. Here are the ones that were presented:

  • Last Interaction (only the last touchpoint matters)
  • Linear (every touchpoint gets equal credit)
  • First Click (only the first touchpoint matters)
  • Time Decay (more recent touchpoints get more credit)
  • U Curve (value early and recent touchpoints)
  • Engagement-based (value the touchpoints based on the amount of time on site they drove)
  • Custom (you name it, you got it)

This powerful feature may truly put Attribution Modeling on the map, in terms of digital analytics, in the same way that Google Analytics initially put Bounce Rate on the map. Unfortunately, only Premium customers will get to feel the power. The interface looks relatively simple, but will require a truly analytical eye to glean actionable takeaways. One note worth mentioning is that Google demands that privacy policies be updated to include mention that attribution modeling is taking place, yet no PII is being collected.

Google Analytics Right Now: Real-time site-side reporting

Google has a completely new product feature that nobody else in the enterprise site-side analytics space has created: real-time reporting. According to Google and tests we’ve performed, this new report set has a delay of only 1-2 seconds between the tag being called and the report populating. When you view the reports, you can see a live stream of statistics and learn exactly how your website is being used at that very moment. Google has actually decreased the session timeout to just 5 minutes for Right Now. These new reports provide the following report data:

  • Medium / Source
  • Geo-Location (IP-Based)
  • Pages
  • Pageviews
  • Visitors
  • New vs. Returning %

Right Now seems to be one of those features that sounds (and looks) really cool, but probably won’t get used every day. While chatting with colleagues at Razorfish, it sounds like the best use for this data could be during events: new site launches, campaign launches, marketing events, and social media events.

If you’d like to gain early access to Google Analytics Real Time, you can sign up here: https://services.google.com/fb/forms/realtimeanalytics/

In Conclusion: A lot to get excited about

These great features are certainly worth getting excited about! Now that Google Analytics Premium has been officially announced, it will be very interesting to see how the enterprise analytics landscape changes.

What are some reasons clients will prefer to stick with Omniture or Webtrends?

  • Improved Pathing Capabilities
  • Visitor-based Segmentation Capabilities
  • External Data Sources
  • Site Optimization and Display Ad Targeting Integration
  • Social Data Integration
  • Search Management Integration
  • Genesis Integration
  • Classification Systems
  • Familiarity with the System
  • Built-in Hierarchy Tracking
  • Built-in Video Tracking
  • XML Data Insertion API
  • Export Capabilities to PDF, HTML Email, Scheduled Reports, and Alerts
  • Improved Shopping Cart Tracking
  • Option to keep data in-house (Webtrends OnPremise)

Webtrends has already released their opinion on Google Analytics Premium: http://blogs.webtrends.com/blog/2011/09/30/why-enterprise-marketers-should-be-wary-of-google-analytics/

 

The Medium is the Message

Thursday, September 15th, 2011

As a continuation of The “Always There” Brand, let’s focus on a key behavior of the modern brand:

If brands are open and transparent, let’s have more authentic dialogue with our consumers.

The consumer journey has more touch-points than ever before. Broadcasting the same message at each one of these touchpoints is like having the same conversation with another person over and over again. Like friendship, conversation matures over time and it becomes more meaningful and impactful. Our conversations adapt to the environment around us.

Shouldn’t modern brand communications strive for the same effect?

Shouldn’t we tailor our messages according to where our consumer is within their journey instead of shoving one message or idea throughout each channel?

Brand marketers still insist that messages from Television be “integrated” across Print, OOH, Email, Display, Search, etc. It’s one thing to intend to increase a consumer’s receptivity, or CTR, for an advertisement by featuring messaging to which they may have previously been exposed. But, this approach also fails to build an authentic relationship.

The narrow focus of one benefit or one communication idea fails to identify the benefit of your brand’s presence within the consumer’s chosen medium. The medium in which a message is communicated fundamentally alters the consumer’s perception of that message.

Speak to consumers in a manner that corresponds to their current stage of the journey.

As stated by Razorfish’s VP of Search, Adam Heimlich: If web technology makes for a substantially different way of delivering brand messages, search’s function of making it easily, instantly, flexibly accessible is also substantially different. Remember that the goal of branding is to simulate human relationships, and consider the relative importance, in human relationships, of first impressions vs. verification of integrity over time.

If your consumer is searching for you, they are granting you permission to speak to them. They chose the medium of search because it provides utility, or some valued benefits. As search marketers, we should strive to better understand this benefit and augment it with our relevant message and consumer-centric branding.

Paid and Organic Search: Why the Marriage of Both Is Important

Thursday, August 11th, 2011

The industry continuously talks about how Paid and Organic Search work better when in market together.  It’s no secret, when you have 2 listings you own more real estate or “shelf space”, but what is the degree of this impact and is it significant enough to require investment in one, when you have a reasonable presence in the other?

Google published its own recent research and presented interesting results across a variety of large brands and industries, including the finding that when both are present on a results page:  conversions increase and revenue per visitor is higher.

Additionally, Paid Search is the guaranteed way for a brand to emerge (with a reasonably competitive rank) for generic searches when Organic Search typically cannot, due to lack of meaningful content.  Google’s research (found in Google AgencyLand) shows the importance of non-brand throughout the search process as a large part (~50%) of the research phase happens on non-brand terms throughout the search cycle.

Google’s claim that “more searches + more ads = more traffic”, sums it up.  

Mantra for the marketer: Own as much real estate to capture consumer interest across the funnel.  This is precisely what Razorfish saw in our own research.

 

Razorfish Paid and Organic Synergies Research

Razorfish wanted to prove the relationship of paid and organic search visits and their impact on client revenue as well as to explore additional attributes that contributed to this revenue.  To do this we built statistical models on a year’s worth of client data of a large retailer.

The goal:  to identify key factors that impacted Organic/Paid Search revenue and quantify the synergistic or cannibalistic (if any) impacts of Paid and Organic Search channels.  The results were clear: the chemistry between the two channels worked great together!  Here’s how:

Before a consumer clicks a Paid Search ad, the probability that the consumer already visited the site’s homepage through Organic Search is very high.  Our research showed at least half (53%) of conversions and revenue happening through Paid Search are preceded by Organic Search visits within the previous 7-days.

It gets even more interesting!  For Branded Keywords, Organic Search visits impacted the Paid Search visits by 81% indicating the cyclical switching between Paid and Organic listings. The Razorfish models show strong synergy between paid/organic links, especially for Branded keywords. Our hypothesis  ( that was supported by the data), indicated that visitors use Organic Search links to navigate easily to a site’s homepage for research, before converting through Paid Search ads.

In spite of this ‘friends with benefits’ relationship between paid and organic: why does the paid link get the edge? As consumers research across both non-brand and brand keywords, Promotional messaging on Paid Search ads trigger conversions by reducing research time-span. The paid link triggers the conversion and is just the better deal since it offers you more ‘benefits’.

Playing defense against your competition is important.

What else did we learn?  Organic Search links and Premium Paid Search ranking (top ranked ads above Organic listings – not right rail) drives greater coverage on the Search Engine Results Page (SERP) and prevents diversion of traffic to competitor sites.  So when someone asks “is it important to buy keywords if you already show up in Organic Search, even Position 1?”  The answer for the most part is YES!  Otherwise you set yourself up to lose that traffic and potential revenue to your competitors.  Our research quantified this impact: for every unit increase in Competitor coverage (a unit was defined as premium listing, vs. top 3 right rail, etc.) revenue declines by ~12%.  But more interestingly for every increase in Paid Search ranking (that resulted in a click), revenue increased 10%.  This tells us your revenue declines at a faster rate when your ranking slips.

This data tells us to protect our ranking, but it should be our standard practice to prevent competitors, negative ads or misleading messages from exploiting the SERP space for our brands.  It’s a crowded marketplace so advertisers must blend Art and Science when managing a Paid Search campaign to ensure our campaigns are successful in hitting our client’s goals (profit/sales/leads) with a mix of Awareness and strategic placements that keep the purchase funnel full and push competitors down.

In Summary:  Organic Search plays an important navigational role in the consumer behavioral patterns while Paid Search is known to close the deal to a conversion as promotional messaging trigger the close.  Again, just mere investment in Paid Search is not enough, but aggressive ranking in both channels is key to positive impact on client revenue.

 

Why Paid and Organic Work Well Together

  • Consumers convert  after multiple types of searches and clicks, in their ‘research’ phase
  • Organic Search ranks well for Brand terms(read:  mostly navigational), but Paid Search can fill the gap on Non-Brand coverage (read:  awareness, deal-breaker offers)
  • Paid Search messaging can be managed, tested and optimized.  And promotional language helps to close the consumer to the desired action.
  • You can ensure an optimal experience by driving consumers deep into a designated landing page that relates to the intent of the search query through Paid Search ads.
  • The more coverage you have, the less room available for competitors to steal traffic and revenue
  • 1+1 = >2 (Friends with benefits can end up having a family!)