By Avinash Pareek, regional head of product innovation & partnerships – media, content and technology
Purchase journey frameworks have been in existence for more than 100 years now. They have generally been mapped in four stages. The journey begins with priming the brand, building saliency around its usage and distinction. Then the prospect is triggered to explore various brands. She makes an active purchase evaluation and purchase, and finally closes the loop with post-purchase behaviour.
In this journey, media and messaging play in parallel to create memory structures (associations that help build a brand’s noticeability). The challenge has always been how to attribute data to this framework.
Studies suggest that multi-channel attribution modelling done well can bring up to 40 per cent more efficiency to marketing budgets. With the evolution of e-commerce, digital attribution of data has seen a tremendous amount of work. It is a well-established piece of measurement to understand how different touchpoints lead to the eventual transaction. However, the purchase journey from a customer’s behavioural lens doesn’t really differ if it is an online purchase or offline purchase. Yet the attribution has been under-explored for brands with offline, physical sales. Is it impossible to attribute data to the offline purchase journey? Not really.
The starting point is to assign an objective to each asset for the physical transaction. For instance, if the overall ambition is to achieve sales for an automobile brand, then objectives need to be assigned to each asset, as each one of them plays a significant role in the purchase journey. A TV spot may be critical to creating saliency; an Instagram carousel may trigger consumers to explore further using search; a feature-comparison native ad may lead to active evaluation; and a Facebook post could drive recruitment to book a test drive.
While there are plenty of determinants affecting consumer actions on a purchase journey, at roughly 40 per cent of total advertising expenditure, the impact of TV cannot be overlooked. Historically, TV has been a strong priming driver but because of its limited ability to integrate with digital data, its attribution to the overall purchase journey has been unexplored. In the absence of anything else, a good way to attribute TV audience data would be to layer it on consumer household panel data. This can loosely help in planning according to purchase behaviour linked to viewership behaviour. This may not be accurate, but is extremely dimensional and it is worth putting a few extra dollars towards making the shopper data talk media language.
Layering on this, digital attribution can be measured in multiple ways including first click, last click, linear, U-shaped, time decay and others. Each model has its merits and demerits. Both first click and last click have plenty of shortcomings, as they assign the conversion to either first interaction or last interaction respectively. They don’t take into account the impact of every touchpoint in between. The last-non-direct-click model assigns weights to a touchpoint that has been strategically placed as a transaction driver.
Converse to this model is the one where equal weightage is given to all the touchpoints. This diminishes the vitality of certain touchpoints that play a more critical role in driving the transaction. A better way is the U-shaped model, which upweights the first and the last touchpoint as they follow through priming to purchase, and all other touchpoints are weighted equally.
In a situation where the category needs a higher evaluation period than impulse, a more sophisticated method of time decay can come into play. This is useful for campaigns with a multi-channel plan, as it can take into account the amount of time taken between decisions, therefore covering the overall purchase journey through
Not studying attribution can cause damage. In a depressed market where every dollar spent on media is questioned, failing to attribute data to the purchase journey can drive an irrational cut-down of the media mix. In a world where everything operates in connected scenarios, the classical theory of optimising chosen media in a mix should be looked at differently. At the same time, modelling helps in building efficiencies over the right mix, since everything operates through ROI-based planning. It also helps in building your own learnings, which may be far different from classical notions prevalent in the market.