Paid Search Marketing: Can It Learn From Database Marketing?

Paid Search Engine Marketing (SEM) is ripe for what a number of top selling books identify as the “the analytical edge” in marketing.  Indeed, SEM and technology companies regularly and steadily generate copious amounts of data and reports.  Thus, by its very nature, SEM is “data centric.”  But the struggle, as often seen at the intersection of data and business, is the identification of data patterns that significantly impact decision marketing regarding marketing investments.  Exactly how much should I budget for my campaigns (each day)?  How much should I bid on my increasingly large portfolio of Search keywords?  How is my SEM performance being impacted by creative copy, landing pages, check-out processes, and the overall marketing mix?  Common questions for SEM — with answers “in the data.”  But how can we dig the answers out of the data?  And how can SEM avoid becoming report-heavy yet also analysis-light? 

Some commentators highlight the abilities of database marketers and analysts (direct mail, in and outbound telemarketing, email, loyalty programs) to synthesize, sift, and understand data, and then produce actionable (specific, timely, impactful) recommendations.  And those database marketing abilities translate directly over to the data rich environment of Paid Search Marketing.  Is that true?  What are the similarities between database marketing (DM) and Paid Search Engine Marketing (SEM)?  What are the key differences and key pitfalls?  And what are the opportunities for “cross learning” from DM and SEM?  (The same might be said of online marketing, in general.  But I will confine myself to Paid Search for now.) 

The similarities and differences between DM and SEM exist in three primary dimensions – the amount and character of data, the sophistication of the analytics, and message delivery system (velocity, intermediaries). 

DM and SEM are most similar when it comes to transactional data volume, less similar in terms of analytical sophistication, and least similar in terms of marketing velocity.  More specifically, the key similarities are:

1.    Heavy transactional data volume
2.    Understanding the customer
3.    Concern with message timing
4.    Prevalence of quantitative methods and expertise
5.    Focus on testing
6.    Heavy reliance on performance feedback
7.    Dependence on marketing mix

While the key differences (and dangers) are:

1.    Heavy data volume in combination with marketing velocity
2.    Proximity of message timing
3.    Peculiarities of message delivery system
4.    Tracking and privacy complexity

Let’s dive into each similarity and difference in more detail.

1.    Transactional data volume.  The sheer amount of Search data requires substantial computer resources (storage, processing), data management (“highly operational” databases with automated ETL), and analytical sophistication (predictive modeling).  Excel spreadsheets, much less the human mind, cannot grasp the volume of data. 

In the database marketing world, oftentimes the number of potential responders ranges in the many millions with potentially many hundreds of characteristics (response history, demographics, etc.)  In many cases, the database marketer even has transaction data (purchases, products, marketing touches, loyalty, etc.) which potentially transforms the database into many millions of records (even billions for very large organizations).  The emphasis on data integrity and continuous ETL (extraction-transformation-loading) forms a critical component in database marketing.  Likewise, Search Engine Marketers deal with many thousands or millions of keywords with performance data that are reported every day (if not hourly from “real time” tracking systems).  But on top of that, the search engine “touchpoints” offer a tremendous depth of marketing experimentation – different positions on the search engine result page, different ad copy, different landing pages, different geographies, different budget levels for groups of keywords, etc.  This latter layer of Search engine complexity feeds into a “pressure cooker” of marketing management where database marketers seldom tread.  Test designs and analytics in database marketing are rarely updated and acted upon in real time.  Thus, the velocity of Paid Search Marketing produces a unique data intensive environment not found in database marketing.

2.    Understanding the customer.  Once the marketer understands their customer (“gets inside their head”), they know how best to get their attention, position their product or service, and achieve the desired response (purchase).

Understanding customers, of course, requires knowing your customer – their state of mind, tastes and preferences, unmet needs, circumstances, etc.  Knowing the customer or prospect is, of course, the meat and potatoes of database marketing.  The depth of information about the prospective message recipient exceeds anything available to the Search Engine Marketer (not withstanding some of the recent services that profile searchers on gender, age, and income).   This represents perhaps one of the greatest opportunities in the area of Search Engine Marketing, because Search is so often handled in as an acquisition channel, as opposed to an opportunity for retention marketing.  Search marketers should ask themselves this question: What percent of responses are attributable to existing customers?  Database marketers know the answer to this basic question.  Most (if not all?) search engine markets do not.  Go to any book on database marketing, and examine the differences in marketing approach to new versus existing customers.  Ask: Could Search Engine Marketing be done differently in light of basic customer differences?  Even with real-time tracking of website visitors (a very TACTICAL version of “know thy customer”), the on-the-fly segmentation of online marketing pales in comparison to the rich, STRATEGIC focus in database marketing.  (Behavioral targeting – or as Google coyly refers to it, “interest-based targeting” – offers a major step in the right direction of Search marketers.  For more on this tactical understanding of customers, see below.)

3.    Message timing. As all marketers know, “timing is everything.”  Marketers are obsessed with timing their messages to prospects or customers at “critical points” in their lives or business relationship, because it means the difference between success (acquisition, retention, upsell) and failure (indifference, disengagement, defection). 

When is the right time to deliver a marketing message?  Carefully studying prospect or customer data points allows the database marketer to deliver the right message at the right time to the right person.  In some cases, timing decisions result from “alerts” being generated by transactional systems (usage has declined by X%, prospect purchased product Y, etc.) that presage defection or buying for other goods or service.  In other cases, appropriate timing is inferred from general life stage or circumstances (demographics, age).  In the latter case, message timing tends to be somewhat blurry – anyone in the segment “qualifies” for marketing for the next several years.  But, in general, predictive models are very good at blending multiple customer characteristics into a “response likelihood score,” which permits precise rank-ordering of customers in terms of “most likely to respond at this time” (a kind of message timing assessment).

4.    Quantitative methods.  The use of powerful statistical algorithms lends themselves to prediction and goal-oriented optimization that incorporates myriads of data influences on the “response.”  In doing so, marketing budgets are allocated to maximize return/response.

In order to exploit the vast amount of data being generated by transactional data systems by DM and SEM, it is understandable that an entire sub-discipline in marketing focuses on reporting, data mining, forecasting and prediction, and optimization.  SAS, the premier data management and statistical analysis platform, organizes analytical sophistication into eight different levels:

a.    Standard reporting – what happened?
b.    Ad hoc reporting (filtering, dimensions/attributes, metrics) – how many, how often, where?
c.    Query-like drill down – where exactly is the problem?
d.    Alerts – what actions are needed?
e.    Statistical analysis – why is this happening (taking into multiple explanations)?
f.    Forecasting – what if these trends continue?
g.    Predictive modeling – what will happen next given different conditions?
h.    Optimization – what’s the best that can happen?

As quantitative marketing progresses from standard reporting to optimization, its degree of intelligence and competitive advantages increases.  In database marketing, it’s quite common to operate at the highest levels … predictive modeling and optimization.  SEM aspires to an equally high level of sophistication.  (The fundamental basis of IMPAQT’s Adaptive Bidding and budget optimization tools is forecasting, predictive modeling and optimization.)  But more often than not, it is often a struggle to get the data organized enough for robust reporting, much less intensive study.  The volume and complexity of data in online marketing (due to competing measurement platforms and “views of reality”) makes analytical SEM quite challenging indeed.

5.    Testing.  Champion-challenger, control-versus-treatment, statistically significant lift over controls, full factorial and partial factorial experimental designs, etc. are the parlance of database and online marketers. 

To understand your customers, the appropriate timing of message, and the effect of different marketing levers, it is of critical importance to use testing.  Why?  First, in almost any situation, including those with tons of data, it is difficult (or impossible) to eliminate alternative explanations for why X message seems to be better than Y.  How many times do shifts in marketing tactics (price, product positioning, etc.) occur at exactly the same time (as part of the “new campaign”)?  Consequently, it is difficult to answer the question – “why did we succeed/fail?”  And more importantly, “what should we do next?”  (Recall the goal of predictive modeling – the prediction of effects given different conditions.) 

Both database marketing and SEM are quite familiar with the principles of testing, although testing velocity is much greater in SEM.  Additionally, SEM and online marketing in general oftentimes can support very complex tests because of high data volumes.  In any case, testing remains one of the areas of fruitful “transfer of knowledge” between the marketing sub-discplines.

6.    Performance feedback.  Both database and online marketing explicitly incorporate marketing performance feedback (“hard numbers”) into strategic and tactical marketing actions. 
                                              
The incorporation of performance feedback – response rate, cost per lead/sale, ROI, etc. – follows naturally from the data-centric, quantitative, and testing culture of DM and SEM.  Marketing programs are consequently fine-tuned to capture and act upon those learnings, in a process of continual enhancement.  (How often have we seen the “feedback loop” graph as general descriptions of data-centric marketing?)

While feedback is important to SEM, database marketing is far more sophisticated in terms of “feedback” relevant to non-acquisition marketing activities.  As noted earlier, SEM is hyper-focused on acquisition … as opposed to the richer examination of how marketing impacts retention of customers, up-sell, and lifetime value.  That additional layer of performance feedback associated with customer data, therefore, represents a major opportunity for database marketers to help SEM.

Additionally, one of the greatest flaws in the performance feedback aspects of database marketing SEM is the struggle to “get outside the silo.”  Both database marketing and SEM oftentimes share a flaw – the incorporation and study of feedback in a very close system of data, testing, and optimization.  Efforts to get outside of their respective silos bring us to our next topic.

7.    Marketing mix.  While there is a tendency in both DM and SEM to operate in a silo, each channel depends on the marketing mix or environment in significant ways.
                                        
All savvy marketers recognize that the performance of one marketing channel depends on the activity occurring in another channel.  The general “marketing environment” also plays a role in response and performance.  Pricing and promotions, external “events” (news, economy), and media mix (TV,etc.) all critically impact the response rate to database marketing efforts, as well as those in Search Engine Marketing.  (IMPAQT has documented the effects of media mix on SEM time and time again, and incorporates these external influences and “shocks” into its budget and bidding platforms.)
                                                                                                 
It is easy to build in marketing mix effects into DM and SEM efforts?  Of course not.  The use of control groups (to eliminate “external influences” as an explanation for testing differences) helps to control for marketing mix effects.  But control groups do not directly enable forecasting when the marketing mix changes (and we know it will).  Tests are time-bound, in that respect.

What’s needed in addition to “slice in time” testing is a complete picture of all the marketing channels – marketing (prices, promos, etc.), media and conversions.  In other words, data for all critical “upstream” and “downstream” channels.  And then a proper allocation of success (conversions, sales) across those channels in proportion to their influence.  This allocation permits proper marketing allocations to SEM (and database marketing, etc.)  for the purpose of both strategic as well as tactical planning.  In doing so, SEM and database marketing can be managed in such as a way to drive the most value to the business as a whole. 

How can marketing mix analysis with proper channel influences and conversion attribution be done?  In general, the expertise is not commonly found in either SEM or database marketing organizations, but rather specialized marketing (or media) mix analytics firms.  This expertise centers on sophisticated econometric analysis (predictive/forecasting models that look at data over long periods of time) AND understand the complexities inherent in diverse marketing channels and customer segments.  Here at IMPAQT, we marry the marketing mix focus to the fast moving arena of SEM where marketing decisions are made in real-time, auction-based technologies.  As the online channel grows, and its targeting ability increases (in a direction similar to database marketing), marketing mix analysis will become imperative to a business’ competitive edge.

Clearly, the parallels between database marketing analytics and paid Search analytics are indeed quite striking.  In fact, it would seem entirely reasonable that anyone specializing in Search Engine Marketing should be conversant in database marketing.  But, it is simplistic and misguided to assume that database marketing skills will transfer over to SEM without major bumps, misunderstandings, and missteps.  Why?  Because there are also major differences between DM and SEM.

Differences/Dangers

1.    Data volume and marketing velocity.  In many cases (esp. direct mail), DM tends to be very deliberative with long performance feedback cycles.  Whereas SEM data refreshes, scoring, testing, and performance feedback occur very rapidly.

In database marketing, predictive models are rarely updated more than often than monthly, if that.  They are built in a careful, methodical way – looking for new data, better predictors, and squeezing out that last few percentage points of lift.  Some might point out that feedback is rapid in certain database marketing applications, especially email and telemarketing (inbound and outbound).  Feedback data (“response”) is used to ascertain the performance of campaigns.  In certain cases (telemarketing, for example), campaigns can be adjusted with minimal delay.  ((Typically, the direct mail or email piece is “over the horizon” already.)  Thus, the marketing velocity of Search Engine Marketing greatly exceeds that of database marketing. 

2.    Proximity of message timing.  In many respects, Search represents the ultimate insight to “what’s on a customer’s mind.”  That being said, the common lack of long-term tracking for a specific searcher often limits the ability of SEM to the “here and now” as opposed to the “life cycle” of the customer.

Database marketers, in fact, seldom deal with explicit expressions of interest except in the most narrow of circumstances like customer service inquiries.  (Pre-qualified prospects are like gold.)  In fact, sometimes they’re more in the mode of generating interest than capitalizing on it.  As such, they are often looking for “leading indicators” of (potential) interest, like contract status, life events or stages, latent interest as defined by prior consumption patterns, etc. 

In Search Engine Marketing, the delivery of the message at the “right time” takes on an entirely different quality, because, in many cases, the message recipient (by searching for information on a keyword) is “raising their hand.”  In other words, it’s time to deliver a message NOW.  Granted, that expression of interest could be very distant from a monetary transaction.  A potential customer might just be learning about a general interest, and be far away from searching for alternative purchase options.  But it almost goes without saying that Search marketers recognize and capitalize on the “pre-qualifying” nature of Search, and the precision of appropriate message timing whether it is a “leading indicator” of a future purchase or an immediate opportunity for a transaction.

But does that mean that Search Engine Marketers have an advantage over DM?  No.  In general, SEM lacks good “leading indicators of interest” because of the absence or complexity of “customer tracking.”  This topic of course is quite complex, and touches on privacy, technology (e.g., cookies), and behavioral targeting (tracking online interactions across sites, time, etc.).  But it does not detract from the point that SEM tends to market “here and now interests” as opposed to “potential interests.”  And there are benefits to marketing to both of those customer mindsets.

3.    Message delivery system.  In SEM, the search engine is the ultimate arbiter of message delivery.  In real time, it determines the relevance of the PROPOSED message, assigns the message value, and – depending on the bid amount and auction dynamics – serves the message to the searcher.  There is no parallel to this independent 3rd-party arbiter function in DM – no Post Office, telephone exchange, etc. which looks at, auctions, prioritizes, and controls the delivery of the message.  (At most, spam filtering of email represents a 3rd party arbiter, but without the added layer of “is the message relevant to what the customer “wants”?”.) 

Imagine the US Post Office determining that mail boxes only can accommodate 10 pieces of mail per week, and they would judge what actually got delivered based on relevance of the mail and the postage bid.  Furthermore, in order to complicate things just a bit more, the message delivery system (Google, US Post Office, etc.) can change its criteria for delivery when it deems fit.  And finally, to cap it all off, it only has to discuss those relevance factors (and how they combine with bidding) in the most general way.  They don’t publish their mathematical formula (or “secret sauce”).  In other words, as the marketer, you’ve got to figure out the black box in order to truly optimize your spending decisions.  (Email marketers are probably more sensitive to this line of thinking, given spam filtering technology by Internet service providers and email software.) 

Additionally, paid search depends critically on a “bidding methodology” for visibility.  Unlike database marketing, there is no “real time auction” (via bidding) for a customer’s attention.  Granted, different mailing formats, etc., result in a different price-per-piece.  And delivery options (1st class, etc.) result in different expenditures per message.  But those values are not changed in real-time.

Database marketers might simply point out that bidding models are similar to response models, in that one need merely calculate the likelihood of a conversion (at a particular cost), and act (prioritize) accordingly.  But let’s circle back to the feedback loop.  The cost is dynamic (almost in real time), and therefore the “return on investment (bid, cost)” changes constantly as new competitors move in and out of the market.  Therefore, the Search Engine Marketer is COMPETING for visibility (not just the end response).  And they have a finite range of visibility to operate with (the position on the search engine result page).

4.    Tracking and Privacy.  Comparing and contrasting SEM and database marketing in terms of tracking and privacy is a very complex topic.  To a large extent, the discussion depends on what type of database marketing we’re talking about – email, telemarketing, or direct mail.  Therefore, most generalities about the topic are just that – general, and with major caveats.  In addition, online tracking continues to be an area of ongoing development and controversy, as compared to traditional database marketing.  Also, the integration of online data with other forms of offline data remains under-developed, (again) as compared to database marketing.

As discussed earlier, both SEM and database marketers are comfortable with “lots of data” (quantity).  But that data differs quite dramatically in terms of “quality”/meaning, especially when it comes to persistent customer characteristics and behaviors.  Database marketers tend to know quite a bit about their prospects and customers – specific location, actual or likely psychographics (attitudes, purchasing behavior), media consumption habits, etc.  SEM, however, is PERHAPS tracking someone’s digital footprints on their site, and POSSIBLY across sites.  In rare, survey-like situations, additional attributes are also being captured onsite. 

Limitations in online data quality are the result of at least three barriers to measurement.  First, the sheer amount of data generated via online activities is quite overwhelming, and subject to considerable “interpretation.”  In fact, operational (measurement) definitions of even the most fundamental concepts — impressions and clicks – prompted the issuance of guidelines from the Interactive Advertising Bureau (IAB).  Yet, these guidelines do not step into the complex definition of business rules for counting onsite navigation (session cookies, page refreshes, page revisits, etc.) and conversion attribution.  Granted, many database marketers are familiar with “channel attribution rules.”  But they might be shocked to know that most web tracking packages automatically credit the “last touch” (or click) as receiving all the credit for a sale.  (To make matters worse, sales are often counted independently from the “associated click,” thereby producing paradoxical reports with sales on dates where no clicks occurred.)  These topics are just the tip of the iceberg when it comes to the “hidden machine” behind online tracking.

The second limitation on online data quality pertains to the use of tracking cookies.  These persistent cookies (not session-level) gather information every time you permit data loading (which is quite frequent; opting out is difficult) and visit a cooperating site (which form into multi-site ad networks) (e.g., Yahoo, Google, Tribal Fusion, Tacoda, ValueClick, BlueLithium).  Therefore, the amount of tracking available is proportional to the scale of the network, or reach.  (Some ad networks claim 75%+ reach; but their definition involves % of internet audience visiting their network at least once as opposed to the % of an audience member’s visits tracked by the network). 

What about Internet Service Providers (ISPs)?  It is true that ISPs, such as Verizon and Comcast, have tremendous amounts of cross-site behavioral tracking (down to your computer processor ID).  Indeed, they do not depend on cookies like the ad networks, but simply “sniff” the data traveling on their transmission network to/from computers.  Technically, nothing online is beyond the reach of the ISPs; hence, concerns regarding privacy and security are moving front and center.

The third limitation on tracking measurement centers on privacy and security.  Interestingly enough, the amount of information about customers available to database marketers is substantial – location, birth dates, number of children, likely income, ethnicity, and purchases (from a wide variety of “coop databases”, etc.).  Only the most sacrosanct information – Social Security Number, credit score, health conditions (from medical providers), financial accounts – remains unavailable (to law abiding or unauthorized viewers).   Anything else goes.  No one asks whether your data re: your magazine subscription can be shared with other marketers.  (It is.)  Yet, the accumulation of online data in the hands of ISPs (and ad networks) has drawn considerable scrutiny from privacy advocates, security experts (where is the information stored?), and the FTC.  For better or worse, this creates a major stumbling block for database marketers entering the world of online marketing and SEM.  The tracking data either does not exist, or remains unavailable (not shared / private). 

In summary, database marketing and Search Engine Marketing do share many characteristics – heavy transactional data volume, focus on the customer, concern with message timing, quantitative methods and expertise, testing, heavy reliance on performance feedback, and dependence on marketing mix.   But the similarities are offset by an underlying set of differences, which impede any direct translation from one discipline to another –heavy data volume in combination with marketing velocity, immediate proximity of message timing, peculiarities of message delivery system, and tracking and privacy complexities. 

The mix of the similarities and differences also represent the strengths and weaknesses of database marketing and SEM as shown below.

The further toward the upper right, the potential for transition and thought-transfer from DM to SEM is quite high.  The lower left represents areas where considerable works needs to be done in both areas.  The “off diagonals” represent areas where cross-learnings are quite limited, mostly due to data limitations and technological challenges.  In particular, the segment and customer-value focus on database marketing seems of particular value to the myopic, acquisition-mode of SEM.  The use of retention and life-time value modeling are unheard of in SEM.  How often are paid Search campaigns used to guide customer service or build value for existing customers? Not very often.  In a similar vein, the focus of database marketing on marketing to “latent interests” (not just those self-identified by customers) represents a major area for development in SEM. 

In summary, the importation of database marketing skill into SEM represents a mixed bag of opportunities and potential pitfalls.  By clearly recognizing those situations, companies will continue to capitalize on the growing importance of SEM.  Furthermore, they will also be positioned to achieve a competitive advantage as traditional advertising channels, such as TV/rich media, converge online in the future.
 

credit-n.ru
http://credit-n.ru/zaymyi-next.html
http://credit-n.ru/zaymyi-next.html
http://credit-n.ru/zaymyi-next.html
http://credit-n.ru/zaymyi-next.html
http://credit-n.ru/zaymyi-next.html

www.otc-certified-store.com https://zp-pdl.com/online-payday-loans-in-america.php http://www.otc-certified-store.com/cardiovascular-diseases-medicine-europe.html https://zp-pdl.com/online-payday-loans-in-america.php https://zp-pdl.com/best-payday-loans.php https://zp-pdl.com/emergency-payday-loans.php