In an earlier post on Search Engine Land called Giving Credit to Keywords Where Credit is due, I mentioned we’d be following up on this topic with some future analysis. The analysis we’ve done is in the form of a case study, below.

Most companies have difficulty justifying the purchase of general top-of-the-funnel keywords such as “slippers” because typically, these types of terms don’t seem to convert-at least when measured using the last click method.  Add this to the fact that general terms are usually more expensive than branded terms like “LL Bean” or “Victoria’s Secret” and you’ll see why some marketers avoid them altogether.

So when marketers look at these kinds of non-branded keywords in their standard analytics package, they can see that they get a lot of clicks and that they’re spending a lot of money.  But what they cannot measure is the value that these ads bring.  In most analytics programs, these keywords actually look like they’re having a negative impact on sales and driving the cost per conversion up.

General keywords show negative profit

Prior to working with us, our client simply couldn’t justify the expense of these types of terms.  The analytics package they were using employed the last click method of attribution, which only gives credit to the last ad clicked prior to conversion.  By using the last click method, their analytics showed that only branded keywords, specific product names, and model numbers were having a positive impact on the bottom line and that general keywords were showing a negative profit.

The first step was to have the client move from a last click attribution model to one that gives credit to all of the ads involved in the purchase path.  The purchase path is the chronological sequencing of all ad clicks, banner impressions, organic visits and direct visits that lead to conversion.  Once they moved to this model, the customer segmented their top 50 keywords into five different groups:

• Competitive terms
• Product names
• Brand terms
• Model numbers
• General terms

Suggesting custom attribution settings

After they grouped the ads based on the above categories, we showed them a report that attributes even credit to each ad in the purchase path and shows even attribution with exclusions.  Exclusions are when you exclude giving credit to certain ads even though they appear in the purchase path.  In our client’s case, they excluded giving credit to any of their brand terms that occurred at the end of the purchase path as these types of keywords are almost always used for navigation and have no impact on the sales process.

By doing this, they were able to see how profit levels for the general terms quickly moved from a negative to a positive return.   They were also able to identify the amount of additional profit they earned from general terms which they chose to reinvest to acquire more clicks and ultimately more profitable conversions.

In addition, we calculated the number of additional visitors and clicks they would be able to purchase with their newfound profit from general terms.  The report also outlined the average profit per order.

Armed with these reports, the client was convinced that, when valued correctly, the purchase of the general terms did yield a significant profit and they therefore invested more dollars into their general top-of-the-funnel terms.

Attribution model leads to increase in ad spend, visitors and profits

The results were impressive.  As a result of reallocation, the client was able to determine the real value that their general terms had. Where the client used to only invest in a few ad sources, they now advertise in many different sources, as their ability to measure an ad’s success wherever it occurs in the purchase path has been greatly increased.

They also enjoy competitive advantage by purchasing general terms that competitors simply find too expensive when evaluated under a last click model.

Most important, however, is the fact that by using attribution management, the company was able to increase their number of visitors by 222% and, most importantly, increase their net profit by 131% over the course of 24 months.

Your ads are often not selling the products that you designed them to sell. We have found that upwards of 48.38% percent of the time people end up buying a product different than the the one featured on ad they clicked on, and upwards of 11.28% of the time they bought a similar product, but not exactly the same product as the ad they clicked on.

When conducting this analysis, we excluded keywords that were brand terms and general category terms. For example, if we were analyzing a site that sold camping supplies and someone did a search for “camping supplies,” we did not include that type of keyword in this analysis because any product that was sold from the site was a camping supply. Instead, we focused on keywords that were product names, derivatives of product names or about a specific type of product.

For each keyword and product combination we looked at, we placed them in to one of three buckets: Exact, similar and unrelated. “Exact” would mean they purchased exactly the same product or product category of the ad they clicked on. “Similar” would be if they looked for XYZ sleeping bag, but purchased ABC sleeping bag, meaning they still purchased a sleeping bag, but it was a different brand or model from their search. “Unrelated” represented when the users search had no relation to what they ultimately purchased.

After conducting this analysis, we found the following information:

The benefits of this analysis are many, especially when it is available as part of your advertising analytics solution. First and foremost is the accuracy that comes with understanding which products are sold by each ad. This insight allows companies to apply real margins to each conversion vs. an average margin. This accuracy ensures that clients never overvalue or undervalue the contribution an ad delivers. Given that bid prices are set and budgets are allocated based on the metrics we see in our advertising, it is imperative that these decisions are made on accurate data to ensure we are getting the most out of our ad budgets.

In addition to the accuracy of performance metrics, this analysis can also be used to improve the performance of ads by identifying business actions that can be taken. For example, one of our clients had a set of keywords designed to sell product A, and we found that 90% of the time they sold something other than product A. This discovery was brought to our client”s attention. Our client did an analysis of their own business to see if they could understand why people looking for product A chose to buy a different product. They looked at the price of product A relative to the price the competition was selling it for. They also looked at their selection, shipping, return policy, etc., on that product relative to their competition and determined they had the most expensive price for that category of products relative to their competition, and that some of their competiton offered a much better return policy. Based on these findings, they brought their pricing in line with the competition and changed their return policy. Now, that keyword sells what it is intended to sell 40% of the time versus the 10% rate before they had this information.

A lot of times we blame our advertising for the lack of conversions or lack of profit it produces. An ad”s job is solely to bring the right prospective customers to your site. It is your site”s job to then convert those prospective customers. This analysis shows how a group of prospects that are looking for certain products may never end up buying them. This also forces the advertiser to ask the question “Why?” The “why” very often has to do with your price, selection, tax and shipping, site navigation, checkout, inventory, and perceived credibility. If you always blame your ads without looking at these other factors, you will waste a lot of money and time manipulating your ads only to end up with the same results.

A deeper dive into this type of analysis can help marketers identify potential up-sell and cross-sell opportunities. For example, if you discover that a lot of people who search for peanut butter and buy peanut butter also buy jelly, you could target any customer going forward that has purchased peanut butter or jelly with an offer to buy the other product. You may also rearrange your site so products that are often bought in tandem are shown together to increase the up-sell opportunities. Another tactic marketers can use is to send custom email messages to their customers that offer them products that are either related to what they actually purchased or are for the product they originally looked for, but chose not to buy.

This information may come as a surprise to many marketers. “How can someone that is searching for one product end up buying something very different?” The reason why people do this is interesting, but not as important as knowing that it does happen. If you have the ability to see which products are actually sold as a result of an ad, then you have a major opportunity to improve not only the performance of your ads, but the performance of your overall business.

We exhibited at the Internet Retailer show in Boston recently. The signage in our booth touted that we had the ability to do attribution management. For that reason, a lot of people came up to us and told us about a situation they are facing that is very common with paid search marketers: their branded keywords were the only ones were converting and their remaining keywords were just driving traffic.

For each person that told me this, I asked them, ‘How do you measure the success of a keyword?’ They would say if it was the last ad clicked and converted, and had a positive ROI, then it was a good keyword.  I then asked them, “If you were a basketball coach, what stats would you look at during halftime to decide who is going to play in the second half?” The answer I kept hearing were points scored, assists and rebounds.

I then asked them, “Why for your basketball team are you counting assists as a success metric, but when you evaluate your online advertising, you only value points scored?” The answer that I received across the board is that, “I don’t have the ability to see the assists, I can only see the points scored, so that’s what I have to focus on.”

And this is the problem that online marketers continue to be forced to deal with.

By utilizing a technology that allows marketers to see the team of ads, or purchase path, that leads to a conversion, along with implementing a simple attribution model that attributes profit and revenue evenly across the ads in a purchase path, marketers can now value ads that assist. With this type of system in place, marketers will quickly see that they have been giving far too much credit to their branded terms, which typically fall in a closing position (last click), and completely ignoring the value of their top of the funnel keywords, which assist in the conversion path.

To take this model one step further, we recommend implementing a purchase path with exclusions model.  In this model, we exclude giving credit to brand terms that occur at the very end of the purchase path. Many studies have shown, and our own research has found, that when people type brand terms into the search engine, they are doing so to navigate back to the site they decided to purchase from.  Under a last click model, the brand terms in this scenario receive all of the credit, while the actual ads that did all of the selling early in the purchase path receive no credit. By using the branded exclusions model, the true credit is being given to the ads that did the work needed to receive a conversion.

The reality with paid search is general terms do work and have real value, as do other forms of ads that do a lot of assisting, like display, comparison shopping engines, email marketing, etc., but you need the ability to measure their performance across a purchase path.

If you’d like more detail and insight into this topic, we’ll be publishing some further analysis in the coming weeks.

There have been numerous reports published over the years by Google, Yahoo and Microsoft that attempt to prove how the use of display advertising, when combined with search, can increase your overall campaign performance.

A recent report published by iProspect, Search Engine Marketing and Online Display Integration Study was featured in an article for MediaPost titled Study Confirms Display Ads, Paid Search Work in Concert. Robert Murray, iProspect’s CEO is quoted in the article, saying “Internet users are more likely to engage and/or eventually make a purchase from brands with which they are already familiar.” Display ads help breed familiarity. The article also reports findings that suggest that of the people that react to display ads,  31% of those people click on display ads, and 27% go to search engines to do a search related to the ad (brand, offer, product).

At ClearSaleing, we have been tracking the relationship between display and search for over 2 years. We assemble advertising into what we call a purchase path, which is the chronological sequencing of display impressions, ad clicks, organic visits and direct visits that lead to conversions and/or non-conversions or abandoned paths. When we analyze the purchase paths of our clients that use display and search advertising, and specifically look at the types of searches they do after being exposed to display, we learn something very interesting.

Based on our observations, the combination of display advertising and paid search can indeed reduce your cost per acquisition (CPA). Our analysis shows that the most common type of search by a user after being exposed to a display ad is a branded search. We have found that with some of our clients, upwards of 50% of the searches that occur after being exposed to a display ad are brand searches.  A brand search is a search for your company name, a misspelling of your company name, or a typo of your company name that leads a consumer to click on your sponsored listing (your PPC ad).  Brand ads are almost always the cheapest PPC ads that you can buy. Many companies have trademark restrictions on other companies bidding on their branded terms, and for those that don’t have trademark protection, they are still likely to pay a very low CPC when compared to their non-branded PPC ads.

For example, if you were in the auto insurance business and are not running display ads, the most common search for your products is likely “auto insurance,” which is a keyword with a per-click cost between $15 and $30+ historically. Our research shows purchase paths that are all search-based typically start with a general term, and often include a second search for a non-branded term, and then close with a branded term. They also go from a general term to a branded term and then convert. The CPA, when considering the entire purchase path in this case, typically runs in excess of $30 to $60+.

If this advertiser ran display ads in addition to search ads, the search the consumer most often does after seeing that display ad is for a brand term. When we measure just the cost of the advertising in these converting paths, the cost for a banner impression is just fractions of a penny, and the cost of a click on a branded term is typically no more than $.25. Therefore, the cost of these converting paths can be as low as $.25 versus a cost of $30 to $60 for the previously stated search-only converting Paths.

By adding display to your campaigns today, you will create new paths for your consumers to navigate, which are much cheaper than search-only paths, and this can allow for your CPA to decrease and searches for your cheaper brand terms to increase.

When I speak to lead generation marketers at conferences and shows, the first thing I tell them is to get beyond a Cost Per Lead (CPL)  metric and to tie their closed deals back to the lead and back to the ads that produced the lead. Once they’re able to do this, they can then judge the performance of their ads in terms of profit and loss versus deeming an ad to be successful just because it produced a cheap lead.

One question that often arises from lead gen marketers is, “How do I optimize ads when I have a long sales cycle?” This question comes from two types of companies:

1. Those with long sales cycles and few conversions over the course of a year, like an individual Realtor.

2. Those with long sales cycles and a higher volume of orders, like a large B2B company, such as SAP.

The challenge that affects the first group is that they don’t have enough conversions during the course of the year to have enough statistically sound data to judge ads solely on the basis of their ability to produce profit.

The challenge that the second group has is not as difficult as the first group. When launching new campaigns or ads, however, they still have to wait for weeks, months or longer for the sales cycle to be completed.  In the meantime, how should they determine if their new ads are having the desired impact or not when it is too soon to be able to tie conversions and profit back to the ads?

In the absence of having conversion data to make optimization decisions, the next best method which satisfies the two examples above is to develop a lead scoring system or lead scorecard. A lead scorecard is a method for assigning value to a lead to determine how qualified that lead is. It could be represented in terms of an actual score (10 being most qualified, 1 being least qualified) or it could be scored according to the stage of the sales process that the lead resides in (new, first call, completed demo, sent out proposal, closed won/lost).

An online ad is good if it produces a qualified lead. An online ad is bad if it produces an unqualified lead. A qualified lead is one that is from a company or person in your target market, and they have the means to afford the product/service you’re selling, they are looking to make a decision in an acceptable timeframe, and they are responsive to your sales team’s phone calls and correspondence.

With your lead scorecard in place, you can now begin to make optimization decisions based on the quality of a lead that is generated from your advertising versus simply looking at the cost per lead or waiting until enough sales have occurred to make a statistically sound decision. For example, a Realtor should determine that an ad is successful if the leads that came from it took their phone call, setup a face-to-face appointment to see houses, were approved for a loan, and were looking to buy in a reasonable timeframe.

When you are a lead gen marketer, each sale requires two things. First, a lead must be generated, and second, the sales person has to do an effective job of selling. If you find you are generating a ton of qualified leads, but are not converting enough to sales, you most likely have a sales problem and not an advertising problem. All lead gen marketers can ask is for their advertising to produce qualified opportunities. If we focus on CPL, we are ignoring quality and are solely focused on quantity. It is important to remember that leads do not pay the bills. Qualified leads, when combined with a skilled sales team do. Make sure you are generating qualified leads and have the right sales team in place, and you will be a successful lead gen marketer.

Increasingly, savvy search marketers believe that the traditional approach of giving conversion credit to the last ad click is a flawed attribution method. Therefore, many of you have investigated or invested in technology that allows you to track beyond the last ad that is clicked so that you can perform attribution over the team of ads that lead to a conversion. With this new tracking in place, you then need to determine the correct attribution models, and it is at this point that if you are like most marketers, you get stuck.

Why do we get stuck? We quickly realize that the argument could be made for many different attribution models. For example, consider the following purchase path: a user clicks on an ad for “running shoes,” then clicks on an ad for “Nike Shox,” followed by clicking on an ad for “Nike turbo 7″ and makes a purchase. One way to attribute in this scenario would be to give equal credit to all three ads. Another could be to give 20% of the credit to first ad, 30% to the second ad and the final 50% to the final ad. Or a third model could be the reverse, where 50% goes to the first ad then 30% to the second and 20% to the final ad. The point is that we can go on for a while with many different models. So which model is correct?

Marketers can find attribution very frustrating when observing the models above because there doesn’t seem to be a clear cut mathematical solution to what the right answer is. At this point, most marketers then begin to question if attribution is all that it’s cracked up to be given that there doesn’t seem to be any accurate way to solve the correct attribution models.

Attribution models take some very complicated mathematics to develop.  Recently, we conducted a webcast titled ‘Measuring the Immeasurable’ that featured our partner, Vetra Analytics. Vetra is a statistical consultancy made up of PHDs in statistics and mathematics that can utilize advanced mathematical modeling to create attribution models.

When solving for attribution, one needs to determine the “influence potential” of each ad click, impression and site visit. To determine this potential, one needs to consider many factors, including but not limited to the timing of the ad, decay rate of the ad, if a conversion was made, what products were sold, the amount spent, was it a first time buyer or repeat buyer, etc. To determine the influence potential, a model or models need to be built which will help to predict the consumer decisions as accurately as possible.

The statistical models account for one of the most difficult things for marketers to grasp: uncertainty. Uncertainty is all of the factors that may go into a buying decision that are not possible to measure, even with your advanced technology. For example, did a friend recommend this product to the consumer? Did they see a billboard or TV commercial? Or did a sales person in the store influence their purchase decision?

By accounting for the uncertainty and building a model that incorporates these factors, we can test the model on a go-forward basis. If the attribution in the model matches the actual results over a sufficient sample size and period, we then know the model is mathematically sound. If the reality does not match the model, we then know the model is not optimal and should be recalibrated.

There are solutions to attribution management that will change the ways in which you manage campaigns. If you are serious about solving attribution because you recognize that accurately attributing credit to your ads will allow you to make more effective media buys, which ultimately lead to greater profits for you and your clients, then you will need to utilize a tool set that implements advanced modeling.

Always remember that a technology is only as good as the people behind it.  When implementing an attribution solution, make sure you have the staff that understands how to calibrate it or that the vendor you choose offers services to assist you in building sound models.   Your success with attribution is solely dependent on your ability to employ accurate models.