Model performance: The power of measurement and transparency in affiliate marketing

The UK performance, or affiliate, marketing industry is currently worth over £1bn a year, with advocates claiming it generates one in 10 sales. However, despite being such a mature channel in terms of the sales figures it generates – £16bn plus a year – practitioners bemoan a widespread lack of quality attribution modeling in the sector .

Add to this some of the historic distrust of performance marketing, and it’s apparent that some marketers are failing to make the most of the opportunity, plus arguably wasting their marketing budgets as well as endangering their relationship with consumers, although some are working to raise awareness of how things can change.

Last-click
Sarah Treliving, head of digital at GroupM media agency MediaCom, explains one of the principal reasons behind the poor application of performance marketing techniques is using a one-dimensional attribution strategy; namely last-click.

This model rewards those in an affiliate network that have been deemed to have generated a sale using the last click before the conversion, regardless of earlier customer interactions upon the path to purchase (or further up the purchase funnel).

“Everyone knows last-click is flawed,” explains Treliving, adding that this had led to scepticism among some advertisers as to the incremental value posed by their affiliate marketing investment.

Clare O’Brien, the IAB’s senior programmes manager, says: “Our Performance Marketing Council also recognises that in some quarters of the marketing and advertising world, that there is a degree of legacy reputation existing.”

Cannibalisation
However, by its nature affiliate marketing employs a complex mix of channels (search, display, content, email, etc), and simply employing last-click attribution in such a complex sector incentivises some poor practice.

‘Last-click’ means all parties on a media plan are chasing the final click before conversion, this could lead to users being over -served with ads. This subsequently eats into an advertiser’s profit margin, according to Ed French, director of digital at ad tech company Ve Interactive. He explains that such a model also means that advertisers could end up retargeting users that would have purchased a product anyway.

The danger of myopic rewarding
Not only does this negatively impact upon profit margin, it could also potentially damage a relationship with a potential customer. Research published last year from InSkin Media and Rapp reveals that 55 per cent of consumers are put off buying an item they have previously expressed an interest in online, if they are retargeted with ads multiple times, after initially researching for it.

Therefore, it’s apparent that the lack of a holistic attribution strategy is not just poor marketing practice, but also potentially detrimental to sales.

Tom Rickey, Ve Interactive’s affiliate partnerships director, explains how such instances can simply be avoided by more responsible “rules set ting” for retargeting when advertisers agree upon campaigns with their affiliate network partners. This involves setting moderate time and recency settings, ie giving sensible directions as to the minimum amount of time to wait before serving a user with an ad after they have visited a website.

Multi-touch
However, a more fundamental tactic can be employed by advertisers. This involves completely overhauling how they reward their affiliate marketing partners by using multi-touch attribution, according to Rickey.

The IAB’s O’Brien echoes this sentiment, and explains how it further aims to improve advertisers’ knowledge of best practice in the performance marketing sector, and therefore improve transparency.

“This is one of those instances where better information and a more transparent lens onto the operations of the sector would reveal that in broad terms, affiliate marketing achieves complete and reported purchases,” she explains.

“CPA [cost per acquisition] represents one of the most risk-free ad spend strategies for advertisers. However, back to the expertise issue, it depends where in the purchase cycle the performance marketing part of the mix occurs,” she adds.

“There is no one-size fits all marketing technique and customer conversion varies hugely dependent of type of product and service and consumer purchase path,” according to O’Brien.

More advanced advertisers understand this, and have developed expertise that allows them to finely optimise their spend, and therefore results, she adds.

Focus on the upper funnel
Ve Interactive’s French explains that adopting a more holistic attribution model (one that focuses on the upper echelons of a customer’s purchase journey, as opposed to the lower end, or last-click) not only rewards media partners on a more equitable basis, but it can also help advertisers improve how they make use of their data.

For instance, advertisers can take their converted user data (i.e. the information on people that have bought items via an affiliate network) and then use this to work out the attributes of other web users that may also likely to purchase.

This can be achieved by cross-referencing it with second and third-party data sources, and is often referred to as ‘look alike modelling’, meaning advertisers can use their spend more efficiently. “The efficiency is gained by not just retargeting everyone that has visited your site, ” explains French.

By doing so, advertisers can work out the incremental value of their spend with affiliate networks, he adds. “Measurement has definitely driven better understanding of the ads and the platforms that they are on,” says Treliving, adding that MediaCom’s specialist direct response unit – MediaCom Response – has led efforts in this area. “You need a common source [of data] and measurement methodology for all those things,” she adds. “You have to look at all the different types of placement for the last click to understand how a rewards site might be contributing to an overall sale.”

New thinking emerges
Affiliate networks are taking note, and star ting to offer attribution payment models that rewards publishers further up the sales funnel. For instance, earlier this year, Affiliate Window introduced a ‘top-up’ commission feature if their content helps le ad to a sale further on another site. Previously the publisher which ‘won the last click’ would have been rewarded, by employing cross-device tracking.

Similarly, eBay Enterprise – the e-commerce giant’s affiliate sales network – more recently announced that it was to introduce a ‘Dynamic Commissioning model’ that lets advertisers pay publishers custom payouts based on new or returning consumers, the type of device used at check out, and product assortment.

eBay Enterprise claims this encourages advertisers to use metrics beyond simple conversion, and contributes to more complex marketing campaigns in or der to deliver incremental results.

Luke Griffiths, general manager, eBay Enterprise, marketing solutions, explains: “Our Dynamic Commissioning product spells the end for ‘one size fits all’ marketing. It means brands can have complete confidence that the customers they’re reaching are the right ones and allows for a greater degree of control over targeting.

“Previously, marketers were targeting indiscriminately and couldn’t provide robust proof that they were reaching the right customers. Now brands can adjust what they are prepared to pay for different audience segments.”

Griffiths says advertisers must analyse their data in order to effectively assess how their marketing spend is performing. He adds: “We still see brands that fail to use analytics and target according to old-fashioned demographic stereotypes, rather than real-time observed behaviour.”

All sources agree that intelligent mining of data in a coherent fashion is key to attaining transparency; failure to do so means less honest players in the system can continue to ‘ game the system’.

This feature was first published in the 16 September issue of The Drum.

Model performance: The power of measurement and transparency in affiliate marketing