Campaign Attribution - Why Do It, Value From Doing It, Who Can Do It?

Today, a large portion of any organisations marketing budget is typically used to drive visitors to corporate websites. Why? Well, according to a survey conducted by IBM in 2010, today, 70 percent of a consumer’s first interaction with a product or service takes place online. Understanding what those individual visitors did and which campaigns are driving the valuable visitors is challenging. 28% of Web site decision-makers dismiss the opportunity, because they believe attribution will never be precise, and therefore is not worth measuring (Forrester Research).

Results from IBM's 2010 CEO Study also in 2010 showed however, that 88 percent of CEOs will focus on getting closer to their customers in next five years and 82 percent of CEOs want to better understand customer needs.

With challenging economic times ahead and to support the stated CEO’s aim to get closer to customers and understand their needs, attribution is now receiving renewed interest. Accurate metrics derived from a sound attribution framework can deliver informed decisions on critical marketing expenditures.  Having this data empowers Marketers to demand more productive campaigns based on key business requirements.

The Challenge

Many believe Campaign Attribution to be purely based on measuring the performance of marketing communications by means of placing ‘tracking codes’ on inbound marketing campaigns, from which charts and graphs can be produced to show effectiveness and provide the return of investment (ROI) proof for partners, creative’s, placements, marketing programs etc. Without additional information, it is difficult for organisations to contest the value and ROI.

Indeed,  perhaps one of the ‘best kept secrets’ in the world of marketing (and particularly online marketing) is that most data models used by analytical platforms are in fact wholly inappropriate. 

An example is the widespread use of cookies as a tracking mechanism on the assumption that “one cookie = one visitor”. This could be misleading, for example:

• One cookie, many visitors - Many people share the same device such as a home or work PC

• One visitor many cookies - A unique visitor uses different devices such as a PC, a Smartphone and a tablet to access the same site

• One cookie , many accounts - The same person accesses different brands or accounts belonging to the same organisation  from a single device

Another common barrier to accurate attribution is a lack of complete and timely data.  The widespread use of traditional web analytics platforms means data may be:

• Aggregated / summarised, by its very nature, aggregating or summarising data means sacrificing individuality and detail. 

• Short lifecycle, possibly  only 30 days or less data is available meaning key events and  complete visitor history are not available 

• Extended latency – Data is processed by a 3rd party and can take over 24 hrs to be available.

• Non transparent – Data may be provided by 3rd parties and not provide the level of detail required

These make it very difficult to measure the effectiveness of online behaviour and thus correctly attribute value to campaigns. Add to this the growth of specialising digital marketing agencies, the increasing variety of analytical measurement platforms and the ever more complex Multi-Channel ‘conversational’ landscape, accurate attribution becomes a real challenge.

Why Change?

Today, agencies across the world are being used by companies of all sizes to manage their marketing communications, whether that is affiliate sites, partner sites, paid search, social networks, television, radio, print and more. The success of a particular marketing campaign is typically measured against key performance indicators (KPI’s) such as revenue, downloads or conversions. Yet no one really measures the success of the campaigns leading up to conversion. Instead decisions are based on first or last click methodology. In many cases, the campaign touch-points leading up to conversion are direct influencers and are significant in aiding and persuading a visitor to take action or make a decision whether or not to ‘convert’ with a brand, site or product.

For true value based Campaign Attribution, it is important and necessary to have a detailed dataset of visitor insight available and at an individual visit level. Lack of ‘business’ data (not click stream or raw data) is a primary reason many organisations are unable to perform such analysis today. 

Web Analytics platforms for example, are widely used to measure the effectiveness of online visit performance but are based on aggregated sets of data. This means the individual journey(s) made by a visitor over time i.e. the customer lifecycle, are lost as most systems summarise on visit level and use first or last visit/click campaign methodology. However, there can be many visits and referrals made by an individual leading up to conversion. Using traditional reporting platforms, only the first or last visit would be available, other potentially influential campaign clicks and the subsequent behavior and experiential data might be lost. This of course leads to an incomplete understanding of the true value of campaigns and visitors and therefore contributes to potentially incorrect and costly attribution.

A Value Based Approach

However, by taking an approach to maintain a consolidated record of every visit by every visitor (Non-aggregated Data), full campaign attribution at an individual visitor level, based on scoring and weighting of goals achieved in current and previous browser sessions, can be applied, calculated and quantified. This opens up the endless possibilities of measuring effectiveness of many things related to online marketing (both on and offsite). Enabling businesses to work out the effectiveness of their marketing campaigns on the lead up to conversion, as well as establish the quality of inbound visitors from various channels and campaigns. Now with complete and accurate data, it is relatively simple to provide summarised and trended ‘dashboard’ views of campaign performance based on visitor lifetime value. With the added benefit of drilldown to more detail.

How Do We Measure Value?

Performance weighting and scoring calculated on goals achieved whilst a visitor is on the web site can be used to gauge the effectiveness of marketing activities in attracting the most valuable visitors. 

For example:

A click-through from an affiliate site might be worth 5 points, arriving on a landing page might be worth 1 point, viewing a detail page might be worth 10 points and watching a video/picture might be worth 15 points.  So, on this occasion, the visitor has attributed 31 points but did not convert. However, the product set is such that it requires research prior to purchase. So, this visit was of value as it demonstrated a propensity to purchase and it’s highly likely the visitor will bookmark and purchase at a later date. Even beyond 30 days

Another visitor clicks through from a paid search link (which by the way, cost £3.20!) The visitor arrives on a landing page worth 1 point, views a product details page worth 10 points then leaves. This visitor generated 11 points but cost £3.20 to acquire.

If the above example were representative of a large amount of site traffic, you would very quickly be able to establish that the paid search marketing campaign has a high acquisition cost associated to it but very low quality of visitor based on their lack of engagement with site content. With this level of insight, businesses may well decide paid search is in fact yielding low value traffic which is depreciating the marketing spend. Therefore it might make sense to drop the particular paid search keywords and invest in the affiliate campaigns driving the ‘valuable’ visitors.

Another consideration when implementing attribution models based on less sophisticated  first and/or last click models is the possibility of significant duplication both in traffic reporting and payments to affiliates and agencies. It is not unusual to have in excess of 20% of budget wasted on duplicate payments for the same clicks/conversions. To prevent this and provide for fair attribution, a simple formula can be applied where no more than 100% of traffic and marketing budget will be allocated (de-duplication). There is a fixed percentage for the initiating click and the converting click, and the remainder is split over visits and weighted depending on the behaviour of the visit (for example, if someone hits just a landing page but then bounces, they will score differently to a visitor who hits the landing page, views target products, adds to basket and starts a checkout process).

How To Implement Such a Solution

For campaign attribution, weighting and scoring for events that occur during any and all marketing channel sessions need to be defined. Definitions for such events will vary greatly depending on the type of business concerned. For example, an Insurance business will have different weight scorings for hitting a landing page, viewing a product, beginning an application through to quote and then on to saving a quote/purchase. Compared to a travel company where a visitor may hit a landing page, search for a destination/route, start a checkout and go through the purchase process and so on. In retail there may be differences dependant on the value of products, higher value electronics will be researched more than say clothing, therefore comparing multiple products and reading reviews might be scored highly.

Once weight scores have been established, these are applied to non-aggregated visitor data. This is not something that can be retrospectively added to summarised/aggregated results typical of traditional Web Analytics platforms so the platforms capable of doing this today are very small indeed, but they do exist!

Thanks for reading.