Problem

The Challenge of Business Owners

Small and Medium businesses (SMBs) face an extraordinary amount of hurdles to building sophisticated marketing operations. Especially with the pandemic and subsequent SMB financial crisis underway, SMBs have been forced to accelerate investment in online marketing spend as their primary channel to drive revenue.

SMBs spend $100K - $120K on average each year on online marketing. Yet unlike their large company counterparts, who have SEO experts, AB testers and data scientists on staff to inform what campaigns to run, SMBs often are making educated guesses on how much to spend online, and on what.

In recent years, via tools like google analytics and shopify, online marketing has become more data driven, even for SMBs. These tools provide SMBs access to AB testing platforms to try out various marketing campaigns and optimize clickthrough rates. Facebook ad services displays graphs to advertisers on ad performance.

While these tools have led to more insight into individual campaigns by providing information on clickthrough rates and impressions, they do not help SMB advertisers develop an understanding of a multi-touch customer experience.

For instance, a customer may first see an Ad. Several days later, they may see the ad once more and click this time, but do not purchase anything on the website. Days later, they search Google for the website, click on a paid link, and decide to purchase. Intuitively as consumers we know that our experience with buying a product usually isn’t a straightforward, one-click ad to purchase pipe. To fully understand which campaigns were worth it, a business must not just track clickthrough rates on individual campaigns but understand the journey a customer takes through their marketing funnel, into an eventual conversion.

The Customer Funnel and Allocating Budget

Companies and businesses usually have a combination of different marketing approaches. These can be sponsored search, display ads, or emails to name a few. Each of these channels includes multiple activities and ad campaigns with varying costs. In addition, one customer can see multiple, duplicate ads or receive multiple marketing messages. The effect of different interactions and touch points at different stages in their journey to purchase aggregate and intertwine. As a result, multi-touch attribution with nuanced representations of customer journey states is critical to spotting high quality user acquisition campaigns.

Without fully understanding which channel drives customer growth or conversion, increasing budget does not necessarily result in proportional increases in conversions. Having the right method to track marketing attribution can help SMBs to reduce the chance of having an ineffective marketing mix. Our team at BECASU explored different machine learning methods and developed a technique to analyze sequences of customers’ journeys and use the insights generated by our model to suggest an optimal marketing spend for each SMB. Our mission is to help SMBs have the right campaigns, spend the right amount of marketing dollars, and the right allocation to attract most customers per dollar spent.

Do SMBs Have This Data?

We only require a persistent user-id that tracks over campaigns, the campaign seen, the timestamp, and whether the user converted. The proliferation of website cookies via google analytics, the Facebook pixel and other user tracking tools have made it easier for businesses to track repeat visitors to their site. Any SMB that does google analytics, for instance, receives this information as a G-ID.

One particular challenge is tracking potential customers who see an ad but do not click or interact with it. For example, a customer seeing a billboard on their drive to work might inform their likelihood of purchasing, but since we aren’t able to track that they saw the billboard, it won’t be represented in the marketing funnel or our analysis. Some services provide information on whether a user was simply served the ad, but this is less common information. We are only able to provide recommendations based on what data we are presented with.