In Part I, we looked at two ways in which paid search may not be receiving its full credit – when lifetime customer value is not measured properly and when online sales tracking is not properly implemented in order to capture all PPC driven clicks.

In Part II, the focus becomes more complex because we dive into the dynamic elements that the advertiser needs to understand in order to correctly attribute the PPC click to a sale.

3. Your Cookie Window is Too Long or Too Short: Now this depends on your business model, the type of product that you sell, whether it is B2B or B2C, what the cost is etc etc. The important element to consider in allocating the appropriate cookie window is the length of your sales cycle – if you cookie window is too short, as in, it is set at 30 days but people take on average 50 days to make the purchase decision then not enough sales will be attributed to paid search and it will appear that your campaign is underperforming. On the other hand, if your cookie window is too long, you may then be attributing too much revenue to the PPC click, particularly if your selling cycle is short.

4. How is Multi-Channel Attribution Handled? Imagine someone who has never visited your website, clicks on a PPC driven keyword and lands on it. Then that person fulfils the desired conversion on the website. Thus, the keyword that was clicked on can be directly attributed to the conversion. That’s the simple version.

Most frequently though, searchers visit your website multiple times, via multiple channels – both offline and online. If this is the case and somebody visited your website through three different channels on three separate occasions, making a purchase on the fourth visit, which of the channels and which of the clicks should receive the revenue attribution?

• The Last Click – This way you are in a sense disregarding the first three clicks and their marketing channels, which means that they are not receiving some percentage of due credit. This has to date been the most common form of attribution within paid search, wherein, no matter what the interaction of the searcher was prior to the purchase / conversion, it is the keyword directly prior to action that is given 100% credit for the conversion.

• The First Click – Vice versa to last click, this model turns things around and ultimately says that no matter what happened after the first click, it is that first interaction with the website that is ultimately the most important. So, 100% attribution is allocated to the first click within the conversion window.

• Linear Allocation – This method credits each touchpoint that led to the final conversion. If four visits were required to drive the conversion, each one is given equal credit for the conversion. Even though this may not be fundamentally the case, such equal attribution, at least this method recognises each touchpoint rather than dismissing them like last and first click do.

• Weighted Allocation – This is the most complex of the models because it is in itself dynamic and unique to each business. This model takes into account all your marketing touchpoints and weights each touchpoint according to a statistically developed model to attribute revenue for the sale.

Most businesses will either be doing no attribution modelling at all, or they will focus on the first three, given their relative mathematical simplicity. In a perfect world, we would all be using some variation of the weighted model.

Gone are the days when PPC was new, sexy and could do no wrong – in the wake of a crippling recession that continues to hold us in its grip in 2010, even online marketing programs are under greater scrutiny.

Clients are asking more tricky questions about their paid search programs in order to evaluate their performance. They are tinkering with sales data and trying to make the right attribution calls. The client is becoming more sophisticated and agencies need the intellectual property and arsenal to be able to provide the answer even before the question is posed.

But, are clients asking the right questions and drawing the right conclusions? Or, are they beating to death an excellent marketing medium by over-analysing data incorrectly and making bad business calls as a result, such as dropping PPC spend or pulling out altogether?

There are a number of factors, if not measured / addressed properly, which will significantly skew paid search results. We will explore four of these factors in  this two-part post:

1. Measuring Full Customer Value – When you get a paid search driven purchase, are you fully measuring the value of that purchase or are you selling PPC short? Beyond the immediate short-term benefit of a sale, there are long- term benefits that should not be overlooked:

• Customer Lifetime Value: If you make monthly recurring revenue from a client, this should be factored into the sale and attributed to paid search. Certainly, recurring revenue will be a factor of attrition rate but perhaps estimate your average attrition rate and calculate the real lifetime value of a sale rather than just the initial value of the purchase. This will provide you with a much more accurate picture of sales margin.

Furthermore, depending on the type of business in which you operate, you may accrue the value of repeat purchases. If you run a solid business and happy customers return, the long term value of the paid search driven sale may far exceed the marketing cost to capture that initial sale.

• Customer Referral Value: WOM (word-of-mouth) and direct referrals are incredibly valuable given their credibility. This is ‘free’ marketing that should not be overlooked!

2. Is Sales Tracking Fully Functional? Ironic, that with paid search the costs are completely transparent – we see all the costs incurred, but we don’t always see all the sales generated by PPC. This is for a plethora of reasons, some of which are simpler to solve than others:

Tracking code is incorrectly set up or the javascript is located in the footer so if a content heavy, image laden page takes a long time to load, the searcher may already have jumped to the next page before the tracking parameters have kicked in to record the searcher on that page. Searchers disable cookies making cookie tracking redundant or they use one machine for the initial search, only to make the purchase from another machine, at work, with a completely different IP address and a new cookie to track. Or if the cookie window is not long enough, the searcher comes back, via an organic click so the sale is not correctly attributed to PPC. The reasons are many, they are diverse and sometimes complicated – but it is important to know about them and try to fix them in order to minimise that margin of error.

In the next post we’ll dive a little bit more deeply into understanding cookie windows and multi-channel attribution.