Over $155 billion worth of consumer goods was purchased online in the U.S. in 2009, yet a far larger portion of offline sales were influenced by online research, according to a March report from Forrester Research. Forrester estimates that $917 billion worth of retail sales last year were “web-influenced,” with online and web-influenced offline sales combined accounting for 42% of total retail sales. That percentage will grow to 53% by 2014, when the web will influence $1.4 billion worth of in-store sales.

The above findings—as well as general observation—prove that consumers today are constantly shifting between shopping channels, and often using search as the starting point in their purchase process. Multi-channel marketers looking for an edge in the paid search game should look to measure the impact their programs, not only on online conversion, but on offline conversion as well. When search marketers simply measure online clicks and conversion, they aren’t getting an accurate picture of the value of their keywords or campaigns. A keyword may seem to have a low conversion rate, but what if a large number of people who click on that keyword end up purchasing in store or over the phone? Or what if a campaign has a high click-through rate and a very low conversion rate; does that mean it’s a dud, or does that mean most of those clicks later converted offline?

The truth is, if you’re not attempting to measure the influence of search on offline sales, you are probably underestimating the value of your paid search marketing efforts. To fully understand the ROI of search programs, marketers need to leverage clever ways to integrate offline data into their measurements. By analyzing the impact that online searches have on offline sales, search marketers can dramatically improve bidding, decision making and analysis across their entire SEM program.

Make no mistake, measuring offline conversion is not easy, and it’s rarely an exact science. But there are several concrete ways to measure the impact of paid search marketing on offline conversion. Here are a few strategies to get you started:

Use unique landing pages

One of the easiest ways to begin assessing the impact of paid search on offline conversions is to measure and assign value to key landing pages on your site. For example, for an online retailer, when a site visitor lands on a store locator page, that’s a good hint you are dealing with an offline purchaser. Contact us pages, completion of a contact form, or even views of products that can only be bought in a store also provide good signals that visitors are interested in making an offline purchase.

The key is attempting to quantify this value. With a store locator page view, for example, you need to understand how many offline sales you are driving which are enabled through these types of visits. To get to a number here you might consider providing visitors to your store locator page with a coupon or offer which could only be redeemed in the store. This would allow you to calculate an “offline conversion rate” from your website and then apply that estimated value back to store locator page views.

Because you are dealing with an online event, using this approach, you can easily measure the estimated conversion value down to the keyword level, and use this measurement to make better bidding decisions. Since this calculation is dynamic, consider using a bidding technology that will allow you to overwrite conversion values as the offline sales rates change.

Use dynamic toll-free numbers

For many companies, the ability to accept inbound calls to a contact center is a critical enabler of sales. However, when website visitors interact with salespeople over the phone, marketers typically lose visibility into which online initiative drove the contact. Increasingly, sophisticated search marketers are looking to remedy this gap through use of dynamic toll-free phone numbers. By assigning a toll-free phone number to a web page based on the search marketing channel (Google, Yahoo!, Bing)—or even the campaign that an individual clicked on to reach the page—marketers can tie call-center sales back to the appropriate acquisition channel.

While this may be difficult to do at a keyword level, just measuring the overall impact that paid search programs have on call-center sales can have a dramatic impact on search budgeting. Identifying, for example, that call-center sales from paid search are 2-3x the size of online sales, can quickly change how you view the importance of search on your overall marketing success. You can also use the data to get actionable insights into how to allocate your marketing spend. For example, you may find that the ad groups that drive offline sales are different from the ones that drive online sales. This could necessitate a shift in budget allocations within search, and even inform how you write copy and market your product in the paid search channel.

Use in-store or phone surveys

If all else fails, surveys can be an effective mechanism for estimating the value of online marketing. Check with your customers when they buy—whether in the store or over the phone—and ask “How did you hear about us?” While surveys can suffer from sample bias as well as data errors, they can provide valuable insight into online marketing effectiveness when used correctly. One way to avoid over-complicating surveys is to simply focus on finding out which online channels drove a sale. This information alone is extremely valuable for search budgeting and prioritization. It’s important however, to remember that just because a customer says they “found you on Google” doesn’t mean they did so via paid search. Organic traffic still accounts for the largest percentage of online referrals from search. Check to see what the breakdown of paid and organic referrals is for your site, and use this information to infer how many of your survey respondents are likely to have interacted with your keyword advertisements.

Measuring offline conversions driven by clicks on paid search ads is hardly an exact science. But given the seismic shift taking place in the way people shop today—jumping between online and offline channels many times during a single purchase process—multi-channel marketers need to begin taking steps to integrate online and offline data. Knowing whether your paid search campaigns are working to drive sales—no matter where those sales take place—can be the difference between a good paid search program and an exceptional one.

“Media-mix modeling is the key to marketing success!”

“Without an accurate attribution model you’re throwing away your marketing dollars!!”

At SMX West last week the halls were echoing with passionate cries about attribution analysis. It seemed as if all topics (other than the Yahoo-Microsoft search deal) had taken a back seat for a moment, and suddenly the most important thing to consider was attribution analysis, specifically whether or not you are giving too much credit to SEM and not enough to other media.

Believe me, I share people’s enthusiasm on the topic. It’s clearly one of the next big online marketing problems to solve. And despite the fact that moving away from last-click attribution toward a more elegant and accurate attribution model can really only serve to divert budget away from SEM and toward other channels, I do think it’s the right thing to do. But after talking to as many people as I could, gathering my own data and soliciting opinions, I’m convinced that we are a still a long way from being any good at this at all.

Currently there appear to be two basic types of approaches, both of which seem to me to be fatally flawed.

One approach I see in the market, offered by various otherwise credible services, has the advertiser entering percentages into boxes on a screen, assigning portions of the conversion value to different marketing channels—25% for SEM, 35% for display, 15% for email, and so on. As a large advertiser myself, I can safely say that this approach gives someone like me entirely too much credit as a sophisticated marketer. I don’t know anyone who has a good enough grasp on their business and the implications of attribution analysis to make an intelligent decision in this type of situation. No knock on my fellow advertisers, but seriously, this is way out of our league. Even so, a Google representative stated during a panel I was moderating, that they intend only to provide attribution-related data, placing the burden of analysis on the advertiser.

The other approach I see emerging is a black-box math-based approach. This is more likely to be done in-house by large advertisers, using statistical and predictive modeling to simulate different attribution models, and mapping their outcomes to business metrics like profit, revenue or ROI. While I do think there is significant value in doing the hard math and understanding these problems from a statistical point of view, this methodology tends to be short-sighted. I don’t believe there is a one-size-fits-all approach to attribution analysis where you simply dump your marketing data in, and out magically pops an attribution model that maximizes profit, for example. It’s just not that generic of a problem.

It’s easy for me to sit back and criticize the status quo—so why not offer some solutions, you say? Well, here goes: I envision a three-phased approach that takes some elements of the existing practices, then combines and expands upon them to provide a more complete, appropriate solution for each advertiser.

The first phase involves smart people talking to each other. Revolutionary, no? We need an attribution specialist to lead off this effort by conducting a fairly exhaustive analysis of the advertisers’ business and online marketing programs. Starting with business goals and product adoption cycle, to conversion window analysis, on to a channel-by-channel audit of both online and offline marketing. The purpose of this consulting and analysis is to provide the proper inputs into phase two.

Phase two is the super-math modeling I describe above. With the proper inputs as they relate to an advertiser’s business and its metrics, statistical modeling is needed to predict all possible outcomes and understand which model will best support the advertiser’s business goals.

Finally, phase three makes all of this actionable. We need a way to pluck the wisdom out of phase two and apply it directly to actual media channels the advertiser is running. Ideally we’ll find a way to automate this or at least automate the recommendations, which can then by easily implemented into the media buys themselves.

But before any of us sprint into the world of attribution analysis and media mix modeling, let’s step back and take a long look in the mirror: I don’t know of a way to realistically pull any of this off if an advertiser doesn’t have a common tracking/analytics system for all marketing channels. So before we start hiring expensive analysts, consultants and statisticians, let’s be sure to clean our own houses and get our own data in order. Standardize your analytics and measurement on a single platform so you can compare “apples to apples.” Then you can start to focus on the fun stuff.

In many large enterprises, when SEO is first introduced, it happens because some internal person is championing its cause. If you are that person, chances are really good that you are facing some very interesting challenges related to the ignorance of others in the company. They probably think you are speaking Swahili to them.

You can recreate this phenomenon very easily. Try explaining what SEO is to a group of your neighbors. Undoubtedly, you will get a lot of blank stares, and quite possibly some “what are you talking about?” type comments. Based on their frame of reference they can’t relate. This can—and does—also happen in the senior management team of a company.

One of the problems is that SEO is simply not something that is taught at any level (except at SEO conferences, and by SEO firms). There are no college courses on SEO that I know of, and SEO is not generally discussed in marketing or web development degree programs. As a result, SEO is something just about everyone learns on the job. People in an organization who don’t have to do SEO generally have no understanding of why SEO is important or what it means. Overcoming this ignorance is critical to the success of your organization’s web site.

Start with the business case

You have to tackle the challenge of persuading decision makers first. Here is how you do it:

Develop an estimate of the quantity of related search queries done every day. Start by writing down what search queries you think people might use when they are looking for products and services like yours. If you sell computers, you may end up writing down terms like “computers,” “notebooks,” “laptops,” “Dell,” “HP,” etc. Make sure your list of the terms you come up with is relatively thorough, but do not include your own brand terms in the list.

Once you have the list, use your favorite keyword research tool, such as WordTracker, Keyword Discovery, Google Trends, or whatever tool you prefer, to get the volumes of the individual search terms. Add them up, and multiply that result by 10. For more information on why we do that, check out my recent article onThe Decisive Advantage Of Optimizing For The Long Tail.

The rest is easy. Calculate the “total available market” by multiplying your calculated search volume times your site’s historical conversion rates times the average order size on your site. The calculation might look like this:

Of course you will not get 100% of the search traffic, but at this point we are just sizing the total available market.

Do a competitive analysis of the total available market

So who is actually getting that traffic, and all those orders? You may be getting some, which is great, but your competition is getting the rest of them. This is a critical point to emphasize. Someone is getting those orders. You can actually break this down a bit by doing another calculation. Start by taking the list of search terms and getting yourself a list of rankings of your competitors on each of them.

Then use the data released by AOL a few years back to see what each position in the SERPs will get in traffic (on average):

Run a set of calculations across all of the rankings for your competitors, times the click through rate, times 10. This will give you a crude estimate of the search traffic they are getting. Then apply your conversion rate and average order size data and you have an estimate of how much money your competitors are making from search. Here is what the calculation might look like:

This should start to get the juices flowing. Business that you don’t get is business that your competitors get, and this helps them become stronger. Make sure that everyone understands this crucial point before trying to explain the role of SEO and what it means, because there is no point in going further until they do.

While the importance of search is obvious to the readers of this column, it is not obvious to most people. Don’t assume that your management team buys in to this concept, because you will face roadblock after roadblock until they do. Sell the business case up front.

Today, the convention of using indexed page counts and backlinks to benchmark and report on site performance is facing a similar demise.

Google has always been careful at hiding data from prying eyes, including its supplemental index results (which Google claims do not exist anyway) and backlinks. The link: command contains purposefully obfuscated data. Even for sites that have been validated in Webmaster Tools, the link information we’re given has not evolved much. We get a very simple result set of sites linking to us, but we don’t get to slice, dice or order the data much further than that. It’s pretty much the same tool today that it was when it launched in October of 2007.

Index page counts returned using site: commands have been very inconsistent since at least November 2009, in my experience. Depending on how you search, or when you search, or where you search from, you’re likely to see very different results—sometimes varying by tens or even hundreds of thousands!

Yahoo gave us Site Explorer, the de facto backlink research tool in the SEO industry. Thanks, Yahoo, for that. But what’s happening with that tool? It’s been giving us very inconsistent, downright misleading data for the last few months. It’s no longer reliable.

Google is doing too good a job hiding the kinds of info we need as SEOs (supplemental index, indexed pages, backlinks) while Yahoo has given up completely and isn’t putting any resources into search. What we’re left with is a wasteland.

And the worst part? We keep right on doing what we’ve been doing. We really have no choice—we need data points like this. We’re basically stuck in a catch-22 of needing these metrics, while realizing they’re inconsistent and unreliable. That’s not a good place to be.

So just what is the enterprising enterprise SEO practitioner to do? There are options, but unless and until the search engines themselves give us valid and accurate data, these are workarounds and conduits rather than outright solutions.

What tools to use

To be fair, even with accurate data from the engines we’d still use our own tools and perform our own analyses, so really we’re back where we started from. There are some excellent tools we can use to track backlinks. Indexed pages, not so much, but there are workarounds for that, too.

There are three primary tools that every SEO worth her salt should be using (and probably already is). These are Linkscape, MajesticSEO and Open Site Explorer (OSE).

Although SEOmoz has gotten a lot of flak for the technology behind its tool Linkscape, it’s actually an awesome piece of kit. (I wrote “kit” so I could sound like a cool Brit. Yes, I failed.) There are many different ways one can use Linkscape to analyze, dissect, reverse engineer and just basically own the link information of competing websites.

Majestic is another great tool and it really shines once a site is authenticated (for sites you don’t own, you can pay Majestic a fee to get the same data). However, I’m anxious to see how their data ages and the accuracy of its trending information.

Both of these tools do a great job analyzing backlink profiles and showing important metrics such as the domain and URL authority, the number of unique domains linking to a page or website, anchor text information and much more. They also make use of proprietary scoring methods: “ACRank” for Majestic and “mozRank” for Linkscape.

OSE is the newcomer, also from SEOmoz and it’s designed to replace the already dying Yahoo Site Explorer (thank god). It’s another good tool and is based on Linkscape.

While we can’t accurately track indexed pages outside of the search engines, we can use a tool like SEMrush to analyze overall keyword visibility in Google. No, this isn’t going to give us N number of URLs in Google’s index, but what it will do is surface a selection (the data can’t be definitive, nor does it need to in order to be actionable) of terms a domain is ranking for. Pretty darn useful, if you ask me.

The best part about all of these tools, including Google’s own Webmaster Tools, is that the data can be exported into Excel for further analysis. That’s where the magic happens.

What metrics to track

The primary metrics we track are the percentage of overall site traffic from free search, the percentage share of each engine, free search traffic at the keyword level (using clustering for related terms) and the delta between branded and non-branded free search traffic. Then, you can slice and dice as deeply as you need to go, looking at bounce rates, conversion data and more. Often, we do specific analyses at the category or product page level, too.

These are really the primary metrics we should track. You can go further, too, to track things like landing page yield and keyword reach, as illustrated by Brian Klais in the Analyze This column (which I highly recommend adding to your feed reader subscription list).

I really want to stop tracking indexed pages and backlinks using Google and Yahoo’s data. I need to stop. But until there’s a better way, until a reliable and accurate set of metrics and tools exists, this is the standard that we have. This is the best we can do. Sure, we can use our own tools like the excellent ones outlined here and perform our own analyses and we can ensure that cutting-edge and competitive SEO strategies are being formulated.

But C-level executives will continue to request things like ranking reports, indexed page counts, backlink counts (from the engines, not from a third-party tool) and even toolbar PageRank. The problem with these “metrics,” if you can even call them that, is they’re only useful in documenting what the engines are saying. They aren’t useful in documenting a site’s health, potential or competitive SEO stature.

Let’s start with the first question—should I have a strategy for the long tail? The answer is maybe. For some marketers having a small keyword set is the right approach. If you have a limited number of products, operate in a limited number of markets, or your primary campaign goals are focused on branding, there may not be a true “long tail” to the set of keywords that users associate with your offerings.

Other companies will find that classic long tail marketing is an ideal strategy. Retail companies with catalogs that have millions of SKUs for example, may find it useful to purchase inventory for the various keywords associated with their products. Similarly, companies with content that is exceptionally seasonal or dynamic may need to apply strategies for updating ads for thousands of products that are constantly changing. Evaluate your own business model before considering how important the long tail is to your campaigns.

Assuming a long tail strategy makes sense for your business, the next question is can it be implemented cost effectively? To answer this question, first you need to understand the time and effort required to implement a long tail strategy. Expanding your keywords to millions of terms doesn’t happen without cost. To begin estimating the costs, first evaluate the data you have available in-house which can be applied to a long-tail campaign. Do you have a product catalog? If so, how clean is the data? How many attributes do your products have which could factor into a keyword campaign? Being able to harvest the long tail at scale requires having a large and clean data set for building out new campaigns and keywords. If your company needs to undergo a sizeable data cleanup effort before embarking on a long tail strategy, you will need to factor this into your costs.

Once you have a clean data set in place, the key to cost-effectively implementing a tail strategy is automation. Tackling the long tail of search without software is at best challenging, and at worst impractical depending on the size of your product catalog. Packaged software solutions are capable of delivering automation of campaign creation and management through data feeds. We also see many search marketers building in-house solutions for campaign automation. Regardless of the route you choose here, the costs of software need to be factored into the cost of your long tail program. This includes not only the cost of uploading, formatting and publishing feed based ads, but also the costs of managing more accounts, building larger reports and performing analysis at scale on an ongoing basis.

Once you fully assess the benefits and costs of a long tail strategy, compare the ROI to other possible initiatives and prioritize them. Assuming implementing a long tail strategy is right for you, here are some best practices to keep in mind while you’re building out your campaign.

Data quality is critical for campaign success

By definition, when you scale a campaign to millions of keywords, it’s not possible to review your ads using human editors. As a result, it’s important to focus on data quality up front. Make sure that data feeds are well structured and that the data is from a reliable source. This will prevent inappropriate words and landing pages from finding their way into campaigns. When you go to build out your keywords and creative, make sure to perform checks to avoid inserting terms that break character limits imposed by the engines. Also watch out for special characters that will result in your ads being rejected.

Test everything at scale

Because long tail terms receive very little traffic on an individual basis, it’s tough to effectively test keywords, creative and landing pages one at a time. As a result, you need to scale your tests to thousands of terms. For example, if you are testing a creative, try multiple versions of the same ad where you dynamically insert either a price or color across hundreds of campaigns and ad groups. Assign segments for color and price to each of these creatives in your analytics tool so you can perform cross-campaign analysis of performance.

Tail terms demand a different approach to bidding

Sparse data in the tail makes traditional rules-based bidding ineffective. Because individual terms have little click and conversion data history, it’s not possible to accurately estimate a bid based on past performance of the keyword alone. Applying a portfolio approach to bidding allows you to leverage data across related keywords to properly estimate bids. Using a bid tool or methodology that supports portfolio based bidding is critical to success with long-tail strategies. When applying a portfolio approach, watch out for tools that require a “learning” period. Using “test bids” to zero in on the right spend levels can be costly, especially when you are talking about millions of keywords.

Retailers should factor in product margin and inventory data

Because many long tail campaigns are based on product catalogs, you can often get more return on your investment by making a few small tweaks. Clearly when a product is out of stock, it’s probably not appropriate to be advertising for it on Google. Make sure you are able to automatically pause or bid down keywords based on the availability of the products that were used to generate the terms. Additionally, if you are able to incorporate product cost or net margin data into conversion statistics, it will result in a more accurate picture of keyword value, improving your bids and helping you to hit overall ROI targets much easier.

While applying a long-tail strategy may not be for everyone, if done correctly it can be an important driver of growth in keywords under management, clicks and conversions. More importantly, because there is often less competition for long tail terms, long tail campaigns can deliver a lower cost-per-action than ordinary search marketing campaigns.

In the past I’ve talked about the fact that at Yahoo, we have many different properties (think Yahoo Personals, Yahoo Sports, Yahoo Shopping, etc.), and each of these properties has a business that’s unique in it’s own way. Some properties are supported primarily by display media (Sports, Finance, News, Etc.). In others, users can subscribe to services (web hosting, merchant solutions). Yet other properties list products sold by other vendors e.g. Shopping and Travel. Since each of these businesses has a different way of making money, for each property we need a unique way of defining the value that a user can bring to the company. I went into much deeper detail tn this topic of customer valuation a while back.

But what about SEM campaign management? How do you manage different SEM programs for a diverse set of businesses? It turns out that each type of business requires a unique approach to SEM program management, particularly as it relates to keywords, targets and bidding. Let’s look at the following 3 examples I pulled straight out of the trenches.

Keeping it simple

For our branding campaign, as I wrote a few months ago, our goal was to deepen engagement with the Yahoo brand and products. We were able to track and measure engagement through a proxy of web events to which we assigned points to define relative value. In executing against this goal, we (with the help of our agency) managed to a cost per point or cost per value model on a keyword-by-keyword basis. We did this on a fairly manual basis, as the keyword set was finite and manageable, and performance was fairly stable. This is the simplest of examples I can provide from purely a bid management perspective.

The tail wagging the head

In our orders-based businesses, where we’re driving users toward a common conversion point such as a subscription or sale, we (again, with agency assistance) use a somewhat more complex management strategy that is now gaining traction in the industry, though we’ve been employing it for some time. We try to break up our keyword portfolio into the smallest possible discrete datapoints that we can, while maintaining a sufficient quantity performance data for bidding purposes. For example, a high-volume keyword such as “dating” gets broken down into bits that look like “dating on exact match in Chicago during nighttime hours with a value-focused call-to-action ad,” and so on. Bids can then be managed on these micro-units so the campaign can be optimized to a level of efficiency not possible when managing solely at the keyword level.

Kicking it up a notch

For our listings properties, where users come in, research products and services then leave, the model is completely different. For one thing, our revenue is mostly earned in-session (as opposed to a 30-day conversion window, for example. To make matters more complex, we have keywords of every type that number in the millions. High volume, seasonal, long tail—you name it.

To get the most out of these programs, we must deploy multiple management strategies within a single program. For head or high-volume keywords, we can employ automated bid management algorithms to optimize to a desired business metric—revenue, profit, ROI, etc. For other keywords we need to take a more rules-based approach. For example, we may want to take all the keywords that fit a certain profile—ranks on the second page, has above 100% ROI, etc. and perform calculated bid management techniques—bid a percentage of revenue or profit, increase bids by 20%, bid to higher position, etc.

As if that weren’t enough to worry about, for seasonal and other reasons, at any given time we have a large number of keywords that aren’t getting any impressions at all. In cases like this we periodically “re-activate” keywords and try to bid them back to profitability once again (it’s assumed that we bid them down to inactivity at some point because they weren’t profitable). This takes a measured, rule-based approach to qualify and bid systematically to ensure the best chance at regaining profitability.

As you can see, as the landscape across which you’re managing SEM campaigns becomes increasingly varied, the more complex and custom your approach to search marketing will need to be. It’s only when you can match your businesses one-to-one with uniquely suitable approaches that you’ll be able to bring your ad spend to a truly optimized level.

Notice in the left menu the neat categorization of the shoes category of products on the left. First we see a breakdown into women’s and men’s shoes, and from there it breaks down further into various type of shoes. In addition, across the top you see an alphabetic menu for drilling down by manufacturer name. This provides the site visitor with multiple ways to navigate to the product they want to buy.

Now if we click on sneakers under men’s shoes we can see what this looks like a layer deeper:

In this shot we see that Zappos offers many levels of refined navigation. You can pick your shoes based on theme, size, width, color and brand. You also start to see individual products listed on the page. Overall, this provides a great experience because shoppers can use the approach that works best for them. However, if you step back and think about it you can begin to see the basic complexity involved. There must be tens of thousands, or hundreds of thousands of shoe products on this web site.

On many ecommerce sites of similar complexity (whether the product is shoes or not) you will see one of two things:

Hundreds of thousands of pages with an image of the product, drop downs for selecting specific product features and a few bullets about product features (which were supplied by the manufacturer) and little else. The problem is that these pages will likely be seen as low quality pages because they have so little search engine parsable content. As a result, these pages may not even be indexed.

The same scenario as above, except with the addition of a manufacturer supplied product description. The problem with this scenario is that the manufacturer is providing that same text description to everyone else selling their product, resulting in duplicate content. Whenever the search engine sees duplicate content they will choose only one of those pages to show in the search results, and the page they choose may not be yours.

Unfortunately, this is not the only set of problems. Let’s now consider all the navigation pages. What would be the material difference between a page offering blue men’s sneakers and a page offering brown men’s sneakers? Not much at all, so those pages are likely to be seen as duplicates of one another as well.

Solutions

Create a killer link profile. If you have a strong enough link profile, Google will resolve many of the duplicate content choices it needs to make by picking your site instead of another. In addition, it will much more likely be forgiving to include what would otherwise be seen as low quality pages in the index. What kind of link profile is good enough? The best way to think about this is that the objective is to have a better profile than the competition.

One of the reasons for Zappos’ success is the strength of its link profile. Zappos was an early entrant in the market, offered a superior shopping experience, and built a strong brand that consumers trust. All of these things have helped them build a strong link profile and continue to enhance it over time.

Build out unique content. One obvious solution is to put unique content on every page. If you have hundreds of thousands of pages this sounds like a big mountain to climb. In addition, how would you write a different description for a blue, size 6, nike air, men’s running sneaker and a blue, size 6.5, nike air, men’s running sneaker. Definitely challenging.

Amazon has addressed this problem using user contributed content in the form of customer reviews, as you can see in this example:

This gives them great unique content for each page. However, this will not work for sites that do not have a critical mass of traffic. For sites without an authoritative link profile, or without enough traffic to generate a lot of customer commentary, it is probably a good idea for them to generate their own unique content.

The way to do this is in layers. Probably 20% of the pages of any site are the most important ones, at least from a search engine perspective. Of course, on larger sites even 20% may represent tens of thousands of pages. In that case, the best thing to do is to go yet another layer further up in the hierarchy and perhaps address the top 2%. Once you have identified the targets figure out how to write 100 to 250 words of unique content for each page (more for the top 100 pages or so). It does not take a lot to increase the perceived quality of the pages for users and search engines, so this effort will be well worth it.

You can make this more affordable by contracting with overseas-based content writers. We do this in our efforts, and the writing quality is not bad. The information is usually quite accurate, though the English is sometimes not so good. But, then we use local (US based) resources to do a quick edit of the content to fix the language errors and to verify accuracy. Using this process we can get these short articles for under $10 each.

How do you prioritize the pages that should get this new content? Focus on pages close to the home page of your site, as well as pages that appear to be drawing the most traffic. You can get this latter metric from the “most requested pages” or “top landing pages” report in your analytics package.

Engage in PageRank sculpting. No, this concept is not dead, just changing. For example, I consider use of the canonical tag to tell the search engines that one page is essentially the same as another one as an exercise in sculpting. So if you have a page for size 6 men’s blue Nike xx27gh running sneakers (not a real product), and one for size 6.5 men’s blue Nike xx27gh running sneakers, you can use the canonical tag to tell the search engines that these pages are both really the same content as the men’s Nike xx27gh running sneakers (the base page prior to picking a size or a color).

This has the effect of telling the search engine how to deal with pages that it will likely perceive to be low quality or duplicates. The link equity will be consolidated on the most important version of the page, and over time, the search engine is likely to crawl much less frequently pages that have a canonical tag on them.

There are other ways to do sculpting as well. A variety of techniques exist for implementing links to pages that make the page invisible to the search engines. For example, a link to a page can be implemented in an encrypted form of Javascript, or within an iFrame to make it invisible. If you use one of these methods you prevent the search engines from spending any time crawling pages you want them to ignore, but you have to worry that this may change as algorithms get tweaked over time, causing one of these types of techniques to stop working. I would avoid this route unless you are very savvy and able to respond quickly when such algorithm changes occur.

All of these techniques work together to provide an ecommerce site a strong growth path. Sculpting helps the search engine focus its attention on the most differentiated pages of your site. Creating great unique content directly on product pages provides a better user experience and gives the search engine spider something to chew on. It also makes the site more attractive to link to, which along with the sculpting will help drive increases in search engine traffic. This will provide cash flow that enables funding the development of more content. It is a great virtuous circle to build your business steadily over time!

I’m lucky to work with Darrin and the rest of the Zappos team through AudetteMedia in my role as one of their lead SEO consultants.

One of 2009’s iMedia 25, Darrin is the consummate professional. As the senior manager of search marketing for Zappos, he oversees all of the paid search programs, comparison shopping engines (including Google Product Search initiatives) and paid inclusion for Zappos. He also works closely on SEO initiatives and is involved in many other areas of the company, from affiliate marketing, to analytics, to display advertising.

In this industry there are very few people that impart the professionalism, analytical mindset, and insightfulness you get from speaking with Darrin. He has real passion for his job, and it shows. And when it comes to squeezing ROI out of massive and complex PPC campaigns, Darrin has proven he can run with the best.

I asked Darrin a few questions about his experience in the industry, his work at Zappos and what areas of search marketing he’s focused on in 2010. You can follow Darrin on Twitter at wakemono.

Can you give SEL readers a brief background on your history in the industry, and your role at Zappos.com?

I started selling display ads in 2000 as a way to pay for college. It was exciting time though as many online companies started to go under I felt it was time to head in a different direction. So I went on to co-found a computer networking company. After 3 years of 70+ hour weeks and little income, I decided that for the health of my recent marriage and sanity, it was time for another switch. This time I figured I should explore a career that involved my major in college: marketing.

I joined Zappos in 2004 as the 3rd member of the marketing department. On account of our small team, I was fortunate enough to become involved with everything from print buys to paid search. From the very beginning, our team was very focused on direct response, I loved it. I was able to combine my passion for consumer behavior with the technical intricacies of paid search. Today, I’m lucky enough to be involved with a talented team whose passion and focus continue to be around evolving the channel.

As the director of paid search at Zappos, what does a typical day look like for you?

Each day is different so I can’t say that I have a set routine though I do like to start the day by reading industry reports. It’s always healthy to get another point of view from industry experts.

Aside from this, the day quickly turns into a barrage of emails, meetings, reports and nerf football.

What areas are you particularly interested in right now? Where do you see the most potential in your PPC programs?

• Switch to contribution margin bidding
• Stronger emphasis on local markets
• Placing greater value on terms with a higher propensity to drive new customers.

Is the future of PPC automated, or will there always be a human element involved?

We have attempted to automate many of our functions. This approach has helped us to become much more effective in reaching deeper into our campaigns with the time we have. That said and as it stands today, many functions such as campaign sculpting, ad copy and keyword discovery around trends are best left to those in the know.

What ramifications do Google’s personalization, blended results and localization have on your PPC strategies?

These are some of the most exciting parts of the industry. The game keeps changing so it’s a race to keep up and become increasingly more efficient. Greater understanding of the Google ecosphere and the various pieces that make it up keep us on our toes. All of these elements weigh heavily into our ongoing strategy.

What does the acquisition of Zappos by Amazon mean for your PPC strategy and workflow?

Up to this point, this new association has been nothing but complimentary to our internal efforts. We have had some great conversations with their teams yet all decisions regarding strategic direction have been left to our teams to devise.

Last, what do you feel are the most important skills a PPC specialist can cultivate? What do you look for when hiring?

The industry is becoming more data-driven than ever before. Today we look into clickstream behavior as opposed to last click to make decisions. We need people that can chase this data and sort it into meaningful clusters from which to make decisions.

Aside from our ten core values that focus on character, we look for people with a strong analytical backgrounds and a rich understanding of Excel.

Advertisers often start with the wrong approach to the content network altogether. As a result, they find the return on their investment to be sub-par. The most typical mistake we see is when search marketers simply run their search network campaigns on the content network. While this approach may save some setup time, it more or less ensures your content network campaigns will be poor performers—because the two networks are inherently different, so your ad campaigns for both must be, too.

However, even when advertisers follow best practices and organize their content and search campaigns separately, they still find the content network difficult to measure and manage. With your ads being distributed across thousands of sites, it’s hard to ensure they are being displayed on websites that deliver high quality and relevant audiences. At the heart of this problem lie two challenges: measurement and scale.

You can’t manage what you can’t measure and many advertisers don’t have visibility into how their content network campaigns perform at a granular level, such as by site or placement. Addressing this problem requires implementing a robust analytics package that manages the process of integrating click and cost data from Google with conversion data from your site—so you can understand the performance of your placements on the basis of revenues, profit-per-impression and return on investment.

Once you have the tools you need to perform accurate measurement, you also need to find a way to manage campaigns across thousands of potential placements. Technology can be part of the solution here. Using an analytical system that is flexible enough to allow for management by exception—enabling you to search, filter and sort in an iterative fashion—is crucial. But even more critical, is how you go about managing your overall content network programs. By applying some simple best practices to your marketing efforts, you may be surprised to find that your ROI on the content network can often be higher than your ROI on the search network.

Here are four best practices to keep in mind:

Create themes, not lists

Advertising excellence on the search network requires building exhaustive lists of keywords, using phrase and exact matches and attempting to account for every variation, plural or misspelling of a term that a user might type into Google. The content network is different. Google seeks to match the keywords you choose with thematic content on a web page.

As a result, your goals for creating keywords on the content network should take this into account. Organize your ad groups into themes and put together a small but closely related set of terms that target that theme. Make your themes specific and try to include the most important word for your theme in every keyword term. For example, if your theme is Florida real estate, choose a set of keywords such as “sunny florida real estate,” “beautiful florida real estate,” “luxury florida real estate,” etc. By anchoring your terms around a specific word, you will help Google zero in on what you are after. Likewise, by modifying your anchor term with a variety of adjectives, you will create keyword terms that improve your quality scores for potential variants on that anchor term.

Add placements as negatives to prune traffic

Once your campaigns have launched, you can begin to measure the results and try to exclude irrelevant traffic. On the search network, this process involves refining your match types and adding negative keywords. On the content network, however, you have another, more powerful option. By reviewing the performance of the sites where your ads are being clicked on using a profit-per-impression metric, you can quickly add negative placements to your campaigns. This not only reduces your costs, but increases the click-through rates on your campaigns, driving your resulting quality scores higher.

This type of refinement can be more powerful than using negative keywords, because negative placements can easily be applied across all of your content network campaigns. If you find that a particular site doesn’t resonate with your target audience for one product, it’s likely that it won’t resonate for others as well. Because there are no limits to the number of negative placements you can use, pruning over time allows you to increase ROI—not only for your existing campaigns, but also for campaigns that you have yet to launch!

Create separate bids for winning placements

When you find a winning keyword on the search network, your best option is to simply bid it up. With the content network, however, you can do one better. Start by identifying which placements are most effective for your keyword campaigns. Then create a separate campaign targeting these high performing sites and placements only. You can leave your keyword-targeted campaign in place, but bid more on your placement-targeted campaign. Over time, you will be able to cherry pick the placements that are most important to you and bid them aggressively, while at the same time maintaining your existing keyword campaigns to identify new high-value opportunities for placement-specific targeting.

Test, measure, repeat

As with any search marketing methodology, you should treat this as a process, not a single change or tactic. Investment in the process, however, will pay dividends over time. As you build a robust list of both negative and positive placements, you build an asset for targeting your high-value customers and understanding where they spend time on the internet. As you launch new keyword campaigns on the content network for new product lines or offerings, your placement list asset will allow you to rapidly generate ROI with a lower investment of time and effort. And with an understanding of where your customers are on the web, you can begin to target them through other channels such as social media, display and sponsorships on these sites.

I’d love to hear more about your experiences on the content network. Leave a comment below about what has and hasn’t worked for you.

For search marketers like me (and probably you), the question of how to balance the paid/organic dynamic has been around for years. So why is there such an amazing dearth of good information on this topic? Why isn’t there any kind of industry-accepted framework with which to address the age-old question?

I believe that the reason is that the conversation around the interaction between paid and organic search has historically been sorely lacking any good data. As a result, we get stuck talking about opinions and assumptions, and we typically don’t come to any meaningful conclusions. I am grateful that at this point in my career, I am surrounded by savvy marketers who understand how search results pages (SERPs) work. They understand that the SERP is a complex landscape, that each link has its own clickthrough rate (CTR), and that any link’s CTR is affected by the other links with which it shares the SERP. This is the path to meaningful dialogue on the subject, so I encourage everyone to get intimately familiar with the data around the paid/organic dynamic.

So how do we look at the data in a way that can help us understand this phenomenon? First let’s get a few ground rules straight:

• What keywords are we really talking about? Those that match exactly with your brand name or branded product name, where there is generally no competition. So if you are the Acme Widget Company we would be talking about keywords like “acme” and potentially “acme widget.”
• What are we actually trying to compare? Ultimately we want to compare two different conditions: a) a SERP where the organic link for brand keyword is ranked #1 with no PPC ad (and no competitors’ ads) present and b) a SERP where the organic link for a brand keyword is ranked #1 with a #1 rank PPC ad (and no competitors’ ads).
• What phenomena are we trying to measure? In the above cases there are two things that normally happen. I call them cannibalization and lift. “Lift” is the net amount of traffic that is added to the mix by virtue of the PPC ad. Cannibalization is the portion of PPC ad traffic that comes at the expense of the organic link. If you can quantify cannibalization and lift in any situation, you can then begin to think intelligently about what to do.

One thing we need to also acknowledge is the fact that the many variables affecting paid and organic search traffic—search volume, page layout, keyword bids and rankings—prevent us from doing any rigorous scientific testing around the paid/organic dynamic. It’s simply impossible to isolate all the variables necessary to completely understand what’s going on. However, there are some terrific ways that you can at least gather some meaningful data that can be interpreted and analyzed, and from which we can actually draw very useful and actionable conclusions.

Next, let’s agree on a few basic principles:

• Internet (and search) traffic patterns move in weekly cycles.
• Search volume is affected by seasonality, media and other factors.
• You’ll want to “test” in a period of minimum volatility (avoid holidays and seasonal peaks and dips if possible).

Now, consider the following approaches to gather the data required to quantify cannibalization and lift:

• On/off weekly: Pause your paid ads for one week and then resume. This is the simplest approach and takes the least amount of time. If you have more time, try alternating weeks as long as you need.
• On/off daily: For a two week period, alternate pausing and activating your paid ads on consecutive days. Why two weeks? This is the minimum duration required to get both “on” and “off” data for each day of the week.

These are just examples. Use your imagination to design something more elegant if you have more time or budget.

Now What?

Now, you need to gather your data and estimate your lift and cannibalization. The incredibly tricky (and potentially inaccurate) part of this is trying to establish a baseline for organic traffic. Naturally, you will want to use the organic traffic during “off” periods as a baseline, but what about the “on” periods? What would the organic traffic have been without the paid ads present? For this you will need a third data point. Either use averages of organic traffic during “off” periods that bookend an “on” period, or if you have access to data like search volume for a given keyword, you can use this trend to estimate what your organic baseline should be.

The key here is to come up with an approximation for cannibalization and lift. It doesn’t have to be perfect, because you’re going to use this data to determine, based on your business goals, what you should be willing to pay for a click on your PPC ad. Here’s an example:

Let’s use a day’s worth of data, and suppose we determine that our organic baseline traffic is 100 clicks. When we add a PPC ad, that ad provides us with 100 clicks, but when we do so, our total organic+paid total is only 180 clicks. That means that of the 100 PPC clicks we bought, 80 were “lift”, and the other 20 were “cannibalization.” Then, all other things being equal, you should discount your maximum allowable CPC on your brand keyword by 20% to account for the cannibalization, and adjust your bids accordingly. Make sense?

Finally let’s look at the extremes. If your PPC traffic is 100% lift, then you can confidently say that buying your brand terms is absolutely justified, and you have the data to prove it. What, then, if all your PPC traffic is cannibalized organic traffic? If that’s the case, then you had better have an incredibly good reason for paying for the PPC ads. One reason might be that you want to put a differentiated message in front of people, a message that’s not reflected in your organic link. Possible reasons for this might be a brand re-launch or a strategic event like an important product launch or corporate milestone.

This may sound complicated, but I can assure you it’s both do-able and worthwhile. I just completed a study for one of our keywords and I can tell you that I am ecstatic about the results. I can’t wait to share them around the company! Good luck!

]]> http://searchengineland.com/paid-vs-organic-search-understanding-the-dynamics-33155/feed 5