The success of online distribution and digital shopping forums mark a shift in the consumer funnel, especially for traditional brick-and-mortar stores. Despite this market trend, in-store purchases (on average) continue to account for 90% of online-and-offline retailers’ total sales—which raises the question: how much is digital spend impacting offline purchases?

The Problem

As many other online-to-offline (O2O) capable companies have experienced, a large-scale mattress retailer was experiencing difficulty attributing its digital marketing spend to offline purchases in brick-and-mortar locations. The company wanted to connect the dots between platforms, which would help justify its digital spend. Previous attempts using conventional advertising platforms and traditional tracking methodologies offered inexact results and unreliable attribution capabilities. However, Clearlink’s ability to create a tailored-to-fit prediction machine tendered unique, more accurate insight than anything currently available in the market.

The Methodology

To combat this issue, Clearlink created a one-of-a-kind prediction machine. Clearlink knew that this machine could accurately estimate the in-store sales volume driven by preceding online activity. Not only would this approach provide more accurate sales projections, but it would also clarify for the retailer the value and impact of each digital advertising channel—thus enabling the company to understand how to better optimize each investment.

Clearlink established a baseline correlation between channel-specific site engagement and in-store purchases as a starting point. Then, to determine whether digital spend and resulting online behaviors were affecting in-store purchases, we utilized a regression model coupled with machine learning and applied it to organic and various paid channels. Ultimately, this strategy enabled us to determine correlations between specific digital marketing spend to in-store purchases with an extremely high level of accuracy—specifically more than 90%. 

Step 1: Establish Correlation Hypothesis

First, we needed to determine whether or not there was a strong correlation between various digital marketing channels and in-store purchases. To do so, we analyzed a battery of digital engagements throughout our clickstream data and compared them to in-store sales. Once we established the connection and affirmed our hypothesis, we fed the data through a regression model and were able to pinpoint the degree of variability in the parallels between site engagement and in-store sales.

Step 2: Employ Machine Learning

To increase correlation accuracy and ROI by marketing channel, Clearlink developed and utilized a proprietary machine learning model. We then hyper-localized the correlations to promote maximum stability and flexibility within the machine. This ensured that we could strategically identify and locate where in the process the marketing budget would be best spent, and therefore help our partner optimize their digital investments.

Step 3: Compile Granular Data for Slice and Dice

Our team of data experts bundled all relevant data together for additional, more granular correlation assessment. This left us with around 260 regression models, which were filtered through a machine learning environment. This process kept the models up to date with the latest digital consumer behaviors and in-store sales trends. The data was then compiled into a custom dashboard, giving the client’s marketing teams an easy visualization of where budget and spend should be allocated or strategically repositioned.