Marketing is like a football team. The success of the team is not only dependent on the performance of the striker, who scores goals, but on the performance of the entire team, from the goalkeeper to the defenders, midfielders, and forwards. At the IX SGH Forum, I had the pleasure of discussing the challenges in online marketing related to how to attribute revenue generated by our website to the traffic sources bringing users to the site.

What challenges do ecommerce businesses face?
Marketing’s perspective on revenue
How to determine channel effectiveness based on generated transactions?
Models recording user data on the website
Attribution modeling
Summary

What challenges do ecommerce businesses face?

I wanted to present the challenges related to revenue from a business perspective, particularly concerning e-commerce. Speaking as someone responsible for marketing data analysis, I’d like to shed light on the analysis of revenue and the challenges e-commerce faces from this perspective.

Most finance professionals look at revenue through the basic lens of an income statement. From this point of view, examining the structure of a typical online store reveals several positions that will be considered from a financial reporting perspective, focusing on the income and expenses disclosed in a profit and loss statement.

This analysis often follows a more product-centric approach, meaning net sales revenue is broken down into product categories, possibly even subcategories, ultimately down to a detailed product level.

Marketing’s perspective on revenue

In e-commerce, the marketing department is responsible for bringing users to the online store and bears the costs of the marketing mix to attract these users.
The challenge is how to attribute generated revenue to specific marketing channels. Marketers view revenue through what is known as the revenue model.

The outcome of how much revenue e-commerce generates depends on how many users it brings to the online store, the site’s conversion rate, and the average order value.
The conversion rate is simply a measure showing what percentage of visitors to our online store convert into transactions.

This model can be perfectly represented by a funnel diagram. Those familiar with marketing, particularly online marketing, know this model well. It illustrates our website, especially the online store, as a place where a large number of users enter the top of the funnel. Ultimately, a small percentage of these users, usually a few percent, result in transactions.

Users come to the site from various places on the internet through marketing activities known as the marketing mix. Depending on the industry we operate in and our marketing strategies, this marketing mix will generate different user acquisition channels.

These include Google search, display ads on portals, social media, price comparison sites, or our own resources like an email database and newsletters.

How to determine channel effectiveness based on generated transactions?

The key challenge in marketing is determining the effectiveness of individual marketing channels based on generated transactions.

In reality, every user follows a specific purchasing process, which depends on the product being bought, whether it’s comparable, complex, and most notably, its price.

Although it might seem straightforward, imagine a situation where you’re reading an online portal and see a Media Expert ad for a PlayStation console. Clicking on it, you see that it costs 500 euros.

If you click the ad and buy the product, it’s easy to attribute this revenue to the display ad. However, if you don’t buy it immediately and instead discuss it later with friends, eventually searching Google with the query “PlayStation or Xbox,” you might find Media Expert again via organic search.

Still undecided, further discussions might convince you to buy a PlayStation. Next, you search “which PlayStation to buy” on Google, click on a sponsored link from Media Expert comparing different models, and finally, enter a price comparison site to find the best deal and make your purchase for 250 euros.

Summarizing this scenario, you encountered the display ad first, then organic search, followed by a paid search ad, and finally a price comparison site, resulting in a 250-euro purchase.

In terms of cost, marketing departments know the expense for each marketing channel. Assume these are the cost distributions across the channels.

Thus, to generate 250 euros in revenue, we spent 7.50 euros to attract this user to the site.
Assume a 10% gross margin on this product, giving us 25 euros on the 250-euro sale. The problem arises when attempting to split this revenue across individual marketing channels to calculate the return on investment (ROI) for each channel.

In practice, with hundreds of thousands of users, these conversion paths will vary widely—some shorter, some longer, with varying channels. Especially for complex products like booking international vacations online, there can be up to an average of 30 site visits before purchase, each potentially from different traffic sources.

Models recording user data on the website

Tools that record data on user navigation and origin typically assign 100% of the revenue, meaning in marketing, 100% credit for the conversion, to the last source. This is known as the last-click model.
This model is unfair to initial traffic sources since ROI appears higher for comparison sites and lower for initial traffic sources. Cutting off initial sources might prevent users from discovering our site, missing out entirely even through comparison sites.

There are other models, such as first-click attribution, but it also has flaws. Linear models, which are also not ideal, tend to distribute credit equally – akin to everyone at an Olympics receiving the same reward regardless of their performance.

Two other models pre-assign a distribution of revenue percentages across channels, which can be problematic because of the diverse paths and sources contributing to the revenue in the marketing mix.

Attribution modeling

In Conversion, we treat the Marketing Mix like a football team. We can’t solely credit Lewandowski for the team’s victories. Marketing Mix, like any team, functions cohesively. Evidence of an effective Marketing Mix is akin to shifts in public discussions like, “Robert Lewandowski hasn’t scored a goal in 670 minutes on the field,” underscoring the collective effort.

Therefore, in marketing, we use data from analytical systems for attribution modeling. Attribution modeling employs more statistical mechanisms, integrating science into business. The results of attribution modeling are known as data-driven models.

Data-driven models accurately state the percentage of the final conversion attributed to aggregated paths due to the multiple paths leading to revenue. This allows us to precisely calculate ROI, enabling financial controllers to decide which channels to invest in.

The effect of these models is decisions on reallocating the marketing mix budget to generate a higher impact with the same cost or achieve the same effect while saving costs. Marketing aims to be efficient, and these models serve this purpose.

Summary

Although the explanation might seem straightforward, a significant challenge lies in the quality of input data. The principle of GIGO (Garbage In, Garbage Out) states that using these mechanisms with low-quality data renders data-driven model calculations ineffective.

Nevertheless, proper attribution itself remains a considerable challenge.

The post Challenges in online marketing first appeared on Conversion.