Internet sales typically account for 5%-20% of a company’s revenue (depending on the industry), but influence on average 37% of sales, according to the J.C. Williams Group “Start Sampling Multi-channel Report” (2006). Therefore, measuring the bidding process in relation only online transactions significantly under reports SEM’s overall contribution to company revenue. Also, remember that paid search conversions are influenced by more than just the last click; many times a conversion happens a few hours, days or weeks after a series of clicks and searches. So how do you measure and analyze paid search’s contribution to offline conversions that may account for equal—if not more—revenue than online sales?

The key is understanding the full sales cycle by incorporating data from advertising events that happen further up the funnel, but which still play an important role in leading to that final sale. Start with upstream, top-of-funnel influencers such as newsletter subscriptions and store locator searches. The cycle should then be followed all the way through, ending with downstream, bottom-of-funnel influencers such as call center conversions and returns.

To better illustrate this fundamental shift in revenue attribution and tracking, let’s take a closer look at Extra Space Storage Inc., a leader in the self storage rental industry.

Extra Space knew many of its shoppers were looking online to rent storage space, but would ultimately finish the transaction offline by calling the 800-number or driving to their local storage facility. Extra Space quickly realized that under-tracking revenue resulting from online actions like “facility address look-ups” drove down both bidding and volume across their paid search program, and ultimately reduced sales and revenue at their retail locations. They knew they had to find a way to more accurately measure revenue and they found a method using activity attribution.

Extra Space identified website activities that influenced sales. These activities happened early in the sales cycle, and were either good predictors of future revenue, had intrinsic value, or both.  The activities were:

• Product Detail/ Storage Unit Locator Page: Prospect visited the page of a certain facility to get the address.
• Hold: Prospect entered name and address to reserve a unit at a facility.
• Reservation: Prospect provided a deposit on a unit.
• Revenue generating conversion: Actual rental of storage space.

Extra Space surveyed their customers, analyzed the CPM market, and lined up sales data in context of the above activities.  With this analysis complete, they applied discrete values to the four activities, independent of an actual conversion.  Using the activity attribution method enabled them to:

• Correlate keyword assists to discrete web activities for smarter bidding.
• Smooth out sparse sale data by relying on more frequent activities that occur earlier in the sales process.
• Gain earlier insight into the sales cycle and changing market conditions.
• Understand which keywords contribute to a sale, even though they may not be the last click.

Avinash Kaushik, Analytics Evangelist for Google, depicts activity attribution as mapping portfolio of keywords to different success metrics (this is based on John Dewey’s seminal “How We Think” (1910). The following diagram, proposes a similar mapping, using consumers behavior lifecycle for the left column ².

For more on this concept see Avinash Kaushik’s post, Paid Search Analytics: Measuring Value of “Upper Funnel” Keywords.

By leveraging keywords and web site activity that occurs earlier in the sales cycle, Extra Space provides a more reliable stream of conversion data to its SEM team. This consistent conversion stream creates a more accurate picture of future revenue, and conversion rates allowing for smarter keyword bidding.

In addition, the earlier insight into changes in the auction environment allowed the bidding algorithm to react during the first few days of the sales cycle, eliminating inefficient spend while maintaining conversion volume.  Most importantly, Extra Space took activity on their website and created actionable data for reporting and bidding purposes. As a pioneer in activity attribution, Extra Space’s approach drives increased ROI from their search campaigns.

Next time, I’ll delve into the specifics of how to value each activity, how to compare to revenue attribution and show the results, so please stay tuned.

Let’s begin with keyword categorization. One way to categorize your keywords is to relate them to the number of conversions they can generate. You may categorize them into groupings called “head terms,” “fat-middle terms,” and “tail terms.” Head terms generate most of the traffic to your website, but generally need to operate above your CPA goal. Fat-middle terms often meet your CPA goal, and they can be bid on using a straight rules basis. Tail terms can generally come in under your CPA goal, but cannot produce the massive revenue generated by head terms. In order to have enough money to run your head terms over your CPA goal, you need to find places in your PPC campaign to run keywords below your CPA goals.

In general, you can keep the “tail” and “fat-middle” out of the top positions, since many times the ROI on those tops spots is tough to achieve. By shielding yourself from some of the irrational dynamics at the top position in the auction, you can then apply the money saved to head terms that require a higher bid. The graph below shows the categories of terms and how bidding some terms below your CPA goal and others above your CPA goal can help you achieve your overall CPA goal.

Marketers that reallocate spend in this manner generally increase impressions, clicks, and conversions at a lower overall cost than buying keywords in top spots. For those who like to understand the reallocation math, here is an example:

Assumptions for the example:

• CPA goal = $10
• Total conversions for March = 1,000
• Current Overall CPA for March = $9.20
• Average position of all keywords = 5

In order to create savings so you’ll have spend to reallocate, you do not increase bids on keywords above position 3.5, even if they are operating below your CPA goal. These keywords for the most part will be tail terms, usually either product-centric or geo-centric. To get an estimate of the savings created by this strategy, we will find the total number of conversions in the last month that occurred for keywords with average position above—or “north” of— position 3.5 (I’ll call these maxed keywords).

• Conversions occurring north of position 3.5 = 400 or 40% of total conversions
• CPA of keywords north of position 3.5: Northern CPA = $8
• Conversions occurring south of position 3.5 = 60%
• CPA of keywords south of position 3.5: Southern keywords = $10

To achieve the overall CPA goal of $10, southern keywords need a CPA target higher than the overall goal. The formula to calculate that goal follows:

Southern CPA Goal=((CPA Goal-Nothern CPA)*Northern Conv.)/(Southern Conv.)+CPA Goal

$11.33= (($10-$8)*400)/600+10

The formulae above estimates that southern keywords operating at the overall CPA goal of a $10 CPA need their bid boosted to $11.33 to allow the portfolio of keywords to reach an overall goal of $10.00. The formulae for calculating the required boost percentage follows:

Boost=(Southern CPA Goal)/((Northern CPA Goal))

Boost=$11.33/$8=41%

Conclusion: Bidding 41% higher on the keywords south of position 3.5 allows you to hit your CPA goal of $10. Also, by allocating spend from the top of auction to positions further south, you can increase overall volume. Some advertisers act irrationally, overspending just to be in top position, or because they are not accurately measuring the performance of their paid search campaigns on a keyword basis. Since competition is heaviest at the top of the auction, moving a keyword from position 2 to position 1 can easily cost 5 times as much as moving a keyword from position 6 to position 3. This attribute of the auction landscape can be seen indirectly by examining the delta between your bid and CPC for keywords at different auction positions. This delta is called headroom (headroom = MaxCPC-AvgCPC). Headroom does not show the exact cost to move from one position to the next, but given the opaque auction environment, the measure indicates the competitiveness of different parts of the auction. In nearly all paid search auctions, positions further north have larger headroom.

The graph below shows the headroom in an auction by position. Circles indicate keywords, size of the circle indicates number of clicks.

In the graph above, the cost of moving from position 2 to position 1 can mean a massive CPC increase of $0.50 – $1.00. In this same auction environment, a marketer could usually spend just a $0.25 to move a keyword from 3 to position 2. Going from position 4 to position 3 is even cheaper. If the headroom between position 1 and 2 is that much larger than the headroom from position 2 and 3, most likely there are participants in the auction willing to lose money to take the top positions. As a savvy marketer, your best move might be to avoid that part of the auction. Be sure to automatically boost keywords selectively based on their average position from previous days.

Manually executing headroom calculations and spend reallocation strategies like the one above, as well as managing other vital mathematic functions over a large keyword set on a daily basis, is at best daunting and, at worst, impossible. The latest tools allow you to calculate the best bids on your tail terms, fat-middle terms, and head terms using a variety of algorithms. The best algorithms will at a minimum consider conversion rate, average order value, recency, volume and average position when calculating the best bid.

I hope these strategies I’ve provided help you generate the highest return out of your search dollars as we continue to weather these current economic conditions. Do you have a sure-fire search strategy you’d like to share? Please add it in the comment section below.

Don’t scrap your worst-performing creatives altogether; update them with ideas from best-performing creatives. To find your top creatives, you can rely on traditional tools such as AdWords Editor, as well as the search engine ad centers. Focus on the creatives with enough clicks to ensure the data set is statistically significant. As a rule of thumb, if your conversion rate is 5%, you should consider a minimum of 50 clicks as the baseline. If your conversion rate is less than 2%, the relevance threshold curve below shows the minimum number of clicks you should capture.

Now that you have screened the creatives for statistical relevance, you need to pass them through one more filter. Separate your ad-groups into categories and analyze them together; for example, brand and non-brand ads may perform differently. Now that you have logical groups of creatives, you can use different metrics to evaluate creative performance such as click-through rate (CTR%), revenue and profit dollars.

Let’s take a quick look at the pros and cons of the approaches in the context of the example in the table below.

Click-through rate. Creative A is the winner for click-through. Search engines try to maximize CTR. Since they get paid by the click, maximizing clicks (cost-per-click being equal) will maximize their revenues. Google, for example, will prioritize position for an ad just because it has a higher CTR (even if the bids are the same).

Optimizing creatives for CTR maximizes exposure of landing pages and may improve revenue, like it does in this example. However, it rarely maximizes conversions or profit dollars. For example, if you ran the keyword “asics running shoes” you may find the best copy for optimizing CTR is “cheapest price anywhere.” Unfortunately, you may also notice that the bargain hunters clicking on these ads rarely convert and you’ll quickly burn through your SEM budget.

Revenue. Creative B is the revenue-maximizing creative, and for certain companies at certain times, this can be the best creative. However, most companies these days are more concerned about the bottom line: profit dollars.

Profit dollars. Creative C maximizes profit dollars and profits per impression (PPI). These two metrics are closely linked in the above table, as we hold CPC and impressions static to simplify our example. In a world of Quality Score, different creatives on the same keyword can see variance on impressions and CPC. With those variables in play, PPI should be maximized. PPI encompasses all variables influencing profitability and a customer’s purchase/conversion decision. These variables include:

• Inventory on SERP
• Quality Score
• Conversion value chain – Creative, landing page & conversion

Now that you’ve identified your top-performing creatives, what are some common elements you might pull from those best performers?

• Offer type: best price, unique value proposition, call-to-action
• Display URL semantics
• Using dynamic insertion parameters
• Not using dynamic insertions parameters and handcrafting each creative

Find the common elements of your best-performing creatives, and apply those to your worst-performing creatives to see if it makes a difference. Here’s an example of two specific recommendations regarding offer types:

Recommendation 1: Use keyword insertion where possible. Our data found that copy with {KeyWord: __} delivered a 25% lift in profit dollars with a 67% statistical confidence.

Recommendation 2: Test the results of different terms that you think say the same thing. For example, our data showed that the use of “Photos” in copy provided a 72% lift in profit dollars, while using “Pics” reduced profit dollars by 25%. Therefore, simply replacing “Pics” with “Photos” may improve your profit significantly.

By identifying top performing creatives using CTR, revenue or profit as your metric, and then applying the findings from those creatives to your worst-performing creatives, you will maximize the overall performance of your campaign.

In Installment 3 of this series, I will discuss how to leverage rules and portfolio based bidding approaches to maximize towards profit and/or revenue.

Note: Part 1 of this article is Making Your Search Campaigns Work Harder During The Economic Downturn.

But as paid search becomes ever more crucial, it’s also becoming more complex—and the stakes to wring every last dollar out of your search marketing spend are higher. As a serious search marketer, you need the right tools and applications to run your campaigns efficiently and effectively. Just as you wouldn’t set out to landscape your garden with a plastic shovel, you shouldn’t try to manage your crucial, revenue-driving search campaigns with a hodgepodge of spreadsheets and online tools. To help you get the most out of your search programs—especially in these challenging economic times—I’ve put together some meaty tactical tips that serious search marketers can use to help generate the highest returns out of their search dollars.

Using day of week and seasonality data to increase campaign performance

Check to see if certain days of the week have better conversion rates than others. If you are generating a significant number of conversions per day, take advantage of “day-of-week” bidding. The prime example of day-of-week bidding is the marketer selling leads to the auto industry. He knows that on Thursday, Friday, and Saturday his conversion rates go up, and his cost-per-conversion goes down. He is willing to bid more those days because he can stay within his CPL goal due to the increase in leads generated. When engaging in day-of-week bidding, for every 10% increase in conversion rate, you can increase your bids by 10% on those days.

Similarly, if you are headed into a period of higher overall conversion rates, such as the holidays, you should increase your bids now in anticipation of those higher rates. Below is a chart from a leading retailer with the recommended bid increases they follow during the holidays. On the X-axis you can see date as we run through the holiday season. On the Y-axis you can see indexed conversion rate, with 1 being the conversion rate baseline during non-holiday periods. As we move through the holiday season, keywords experience increased conversion rates and increased revenue per conversion. Increases in these metrics increase your revenue-per-click, allowing you to spend more on each click and still hit your margin goals.

Source: Hitwise 2008

For the details in the matrix below, we just focus on revenue-per-click. The increase in revenue-per-click could come from improving conversion rates, improving order sizes, or most likely from both metrics acting together.

Each row on the matrix above matches one of the four holiday periods highlighted in red in the graph. As we watch our revenue-per-click number move up, we can increase our bids by the same percentage. The only thing not reflected in this strategy is the changing competitive dynamics in the auction. In general, auctions get more competitive as the holiday season progresses. Keeping the same average position will be more expensive in December than in September. This means that, all other things being equal, to maintain your place in the auction you will need to sacrifice margin in the holiday season. The art and science of SEM meets with the marketer’s determination to change competitiveness in the auction and blend that data with the hard science of revenue per click.

I hope this strategy serves you well during the competitive holiday season. This year, especially, marketers need to use every weapon possible to increase sales. Next time I’ll examine New Year’s Resolutions for ad copy creation, including ways to save you time & money by recycling top performing creative elements into your lower performing creatives.

Chris Wine is Director, Product Marketing for Marin Software, responsible for driving market positioning and strategy.