Correlating Facebook Ads Analytics with Business Metrics
How to build Facebook advertising analytics

Alex Bouchard
/
/
Correlating Facebook Ads Analytics with Business Metrics

When a business grows primarily by Facebook ads it’s important to understand how much you can spend on advertising to keep your ROI from dropping. For that, you need to track the correlation between Facebook ads analytics and business metrics from different data sources. In this post, Alex Bouchard from From Rachel will tell you how to put the numbers in place and run data-driven experiments.

***

Our business – selling tights & nylons through subscription and an online shop – has mostly been built around Facebook, meaning we’re getting about 60%-70% of traffic from Facebook ads directly. Until recently, we have had a hard time evaluating our performance since all our data points were split between Facebook and twenty different services.

Now we have a growth dashboard and can compare our KPIs relative to the amount spent on Facebook ads within a given timeframe.

facebook ad analytics

Powered by Statsbot

To get Facebook ads analytics, we’re relying on Segment to pipe all the data into a BigQuery database and Statsbot to get visualizations and ad-hoc analytics. To get our sales, website performance, and the product metrics, we connected Google Analytics, Mixpanel, and two other databases to Statsbot. This way we get values from it all over the place.

What is important to track

Before Statsbot, we had this impression that however much money we put into Facebook ads it would always bring more profit. “Oh! Infinite money! – we rejoiced. We can just put as much money as we want into Facebook and it’s all going to come back.” Obviously, it doesn’t work that way.

On our marketing and growth dashboard, money spent on Facebook ads is a key metric we use to compare all our other KPIs to. It’s important to consider the variation rate in ads spent relative to our other metrics. Imagine that our revenue from the Facebook channel is usually more than Facebook costs by 15%. If Facebook spends went up by 30%, how does a 30% increase in the revenue look? As expected, right?

If you were to look at it disregarding the context, this increase would seem like a stunning success! However, in our case it’s actually poor performance. The variation of a metric week over week is useless without weighting in the variation in ads spent for that same period.

To get a better ROI, we look at which articles specifically are getting the most traffic and are generating the most engagement, and then we correlate that by where we put the money on Facebook. These are metrics we keep on the marketing dashboard as well.

Additionally, we have some metrics that are negatively correlated with the Facebook ad spends, for instance, a conversion rate from blog to the product. A large proportion of customers first come to the website through content articles, which have a tendency to bring more people and are less expensive for us, but they have way lower conversion. Later, they’ll get retargeted with an ad that’s very specific for some product, which they can add to cart right away. But the tendency is:

The more people we bring on the site through content pieces, the lower the conversion rate is overall.

Note: Purchases are probably going to go up, but conversion rate does go down.

When you have the numbers in place, it’s time for the magic experiments

Experiment #1: Google vs. Facebook

We tried to reduce the amount we spend on Facebook ads by about 30%. As a result, we got a bigger loss than if only 70% came from Facebook. Then, on the contrary, we put in way more money one week to see what kind of impact it had. We could see the correlation between growth metrics and ad costs pretty clearly.

It gave us the idea that about 10-15% of our Google traffic actually is from Facebook leads, because there are people who see us on Facebook and then just google the company.

Experiment #2: desktop vs. mobile

Our conversion rate on mobile was really poor, and we realized why. Over time we developed a habit of seeing desktop users converting better, so most of our conversion-related posts were targeted to desktop users. And very recently we started experimenting more with targeting conversion posts on mobile and – ta-dah! – we actually realized that mobile ads convert better in terms of ROI.

We ended up putting a lot of money into mobile conversion posts specifically. We did very similar ads for mobile and desktop and ran them at the same time to see if it was a good assumption or if we just got temporarily lucky with mobile.

We went from practically no conversion ads on mobile to making an ROI of 475% on them. We are still running desktop conversion ads but with a lower budget to maintain an interesting ROI.

Experiment #3: scaling content marketing

It’s one thing to scale the advertisement budget, but it’s another to scale the content to advertise with. It’s an ongoing effort to test how much content we can produce while maintaining quality and marginal returns.

By looking at the traffic per article and their respective budget, we have been able to balance our production and it helps us enormously by being the biggest driver of growth. In fact, that’s how we managed to maintain around 300% growth year over year.

It’s an ongoing effort to benchmark the daily article output to their performance and playing with those numbers to see when interest drops off.

facebook ad analytics

Powered by Statsbot

Experiment #4: using videos

One of the things we focus on constantly is engagement, such as likes, comments, reposts, and mentions. If you compare with the industry average right now, we’re doing way better in terms of engagement. That makes it easier to retarget those people that engage with us with more product-focused ads.

Recently, we started making and promoting videos to see how those are performing.  Considering the cost of the videos sits at around 800$ right now, we have actually been disappointed by the performance, especially engagement. Additionally, the ROI has been negative so far. We are now trying to reduce the cost and make videos that are more tightly related to our products rather than being more lifestyle/fashion advice types of videos.

Conclusion

Now we’re pretty sure we’re getting better at how we understand Facebook ads analytics. We’re more optimized all across the board and we’ve been able to fulfill demand better on products and content. Collecting data from different places gives you control of your experiments on the way to getting a better ROI.

You may also like: