How to build a data-driven setup for Ecommerce - DesignYourBike
How to read this marketing setup?
It all starts with the yellow block, highlighting platforms with user interaction.
- In the middle, the website and e-commerce shop (Wordpress and Woocommerce)
- On the left, the ad platforms that drive traffic to our website: Google Ads, Meta Ads and TikTok Ads.
- On the right, email campaigns that also drive traffic to our website from our customer database.
Server-side Tracking
If we follow the arrow down from the website, we go through the Google Tag Manager diamond (in my graph, diamond means "data connector") and connect to 2 destinations:
- the ad platforms
- and our Web Analytics solution: our beloved Google Analytics 4
That we track our website data to GA4 via Google Tag Manager is not really a surprise. Using Google Tag Manager instead of plugins directly in Wordpress and Woocommerce offers you more flexibility to track more events, debug, ...
I set up server-side tracking. So with Google Tag Manager we can also control and transform the data we send to Google Analytics 4.
And not only to GA4 but also to our Ad Platforms, such as Google Ads, Meta Ads and TikTok Ads.
We can ensure no PII (Personal Identifiable Information) is sent to them.
Via Google Tag Manager, we feed Google, Meta and TikTok algorithms with website events. The goal is that these algorithms get as much data from us as possible. They learn what works best in our campaigns to send us better traffic according to our campaign goals.
As selling custom bikes online is a low-volume business model, we need to rely on more events than just purchase to feed our ad platform algorithms. We therefore send a number of micro events (such as "bike_configuration_saved") to qualify audiences that don't necessarily buy.
"We need reports. We don't know where we stand with marketing or sales."
That's what one of the founders told me when we first met. They had a pretty ineffective visualisation of their marketing activity and sales.
Based on that, I knew I needed to build data pipelines to bring together:
- Orders
- Marketing expenses
- Website analytics
After some research, I decided to use Fivetran to connect my different sources to my data warehouse.
Fivetran has a generous free package that covers DesignYourBike's volume.
It is super easy to set up.
It connects our ad platforms and Woocommerce to BigQuery, where we store all the data.
You noticed that Google Analytics is missing as we don't need to send it via Fivetran. Google Analytics 4 has a free integration with BigQuery.
Step 1: Done.
Step 2 is creating dashboards based on this data.
I did that in Looker Studio.
I created many dashboards: some for me to check our Performance Marketing activity, some for management to have an overview of sales, sales compared to goals, sales vs marketing costs, etc.
Criticisms of this setup
CRM/CDP
And you're like "wait a minute! You haven't talked about Zapier and Mailchimp up there on your graph".
Yes, because it needs improvement.
As you've noticed, we use Mailchimp as a CRM to collect customer information and send email campaigns later.
We could do better but this has not been the priority of the management. And to be fair, to deploy a proper CRM (or a CDP) is quite expensive and probably not worth it for this business at this stage.
So in the meantime we connect Mailchimp with Woocommerce via Zapier.
It is therefore more difficult for us to have a user-centric strategy with 1 platform that connects all the dots on a user-level.
BigQuery really needed?
Another concern I have is: does such a business really need a data warehouse in BigQuery?
And there are many tools out there that connect all sources and even create dashboards directly in their software (Funnel.io or Klar for example).
But you know what? All these tools are actually more expensive than the current setup. So yes, BigQuery needs maintenance. And yes, we should be careful not to exceed the free plan for Fivetran or BigQuery.
But this setup for DesignYourBike is cheap. Here's the overview
Cost
- Google Analytics 4: free
- Looker Studio: free
- Fivetran: free
- Google Big Query: free
- Google Tag Manager server-side (with Stape): 20€ monthly
- Mailchimp: 60€ monthly
- Zapier: 30€ monthly
Be careful, as I explained above, some tools here are free regardless of your usage, some of them have a free tier and then start charging. Always check their conditions.
Results
Ok that's all beautiful.
But does it bring anything?
Well, I wrote a lot already so I leave you with numbers of 2023 compared to 2022 and a quote from one of DesignYourBike's founders:
Matthieu took over our Online Marketing, built up data collection, reports, dashboards, helped us to better understand our KPIs, reduce unnecessary expenses and drive several very successful projects. Great guy, great job.