Recently I’ve been running a variety of social ads to promote the Festival of Marketing, but due to a glitch in our purchase process, measuring the exact return has been more difficult than usual.
Luckily I’m able to see the silver lining in any cloud, so I thought I’d use this opportunity to talk about how data correlation can help uncover hidden social ROI.
We got a problem:
First of all let’s look at my problem. As this was the first year that we’ve run the Festival of Marketing, the entire thing was a learning process, and in order to actually hit some very tight deadlines we needed to streamline the site development process.
As we already have a fairly well-optimised checkout installed on Econsultancy.com, so it made sense to use this when visitors to Festivalofmarketing.com proceeded to purchase.
From a UX perspective this is a minor niggle, it could be improved but it’s not the end of the world. When it comes to measuring social revenue though, it’s a complete nightmare.
I’ve written before about the way that social data is often sucked into the ‘Direct’ funnel, meaning it generally delivers better returns than you are able to measure. This is a big issue with organic social, but it also occurs with PPC, and we've got to justify those budgets after all.
In this case, the user is actually redirected to an entirely different site to complete purchase, so even more tracking code than usual is lost along the way.
Just to clarify the user journey in this case, here's a poor-quality diagram:
So how can I work out how effective my social ads are?
Correlation, my dear Watson...
First of all, we need to consider which data sets we do have access to.
Firstly, we can check our social posts and ads for CTR information. Here’s a screenshot showing add response from LinkedIn:
Note those peaks on the 12th, 17th, 23rd etc, and also that almost vertical skyrocket at the end (more on that later). In the case of PPC it’s all laid out neatly for you. If you’re measuring organic stuff you’ll probably have to plot this yourself.
Clicks will never match visits in analytics, but we can at least correlate this graph with traffic data from Festivalofmarketing.com:
Here we can clearly see corresponding (If slightly shallower) peaks and troughs in the data.
Finally, let’s take a look at Econsultancy.com’s analytics.
This shows converting traffic where FestivalofMarketing.com was the referring source:
Again we’ve got some corresponding peaks, so it’s highly probable that our social PPC is influencing these sales peaks. But can we work out exactly how much cash social made us?
Yes... and no. The only real way to check this is to get our historical data hat on. Traditionally events are more likely to lead to social last-click transactions than our other products (which is why all this matters to me, I don’t want my social traffic and revenue falling year on year)
Looking back over the year, social was responsible for about 18.9% of on-site conversions.
Not all of our conversions have a monetary goal though (and incidentally, why not help yourself to a free Facebook trends report), so conversion value is a little lower, around 15%.
As I mentioned, events typically have a higher social last-click value for Econsultancy, but given the amount of conjecture I’m going to go with the lower average figures, so we can reasonably assume that social was a factor in 15% of our online ticket sales.
Last-click purchase is always significantly lower than attribution for social, and in our case it tends to hover around 8%.
I’m not going to say how much we actually made in sales because the stock market would crash, fire and brimstone would fall from the skies, and I’d probably be cleaning out my desk, but I can say that I spent about £2,500 on social ads, so from these figures I can see that we’re making a significant positive return from a mix of organic and paid social.
It’s important to underline the fact that social doesn’t work alone.
The Festival was promoted across all of our and our partners' channels, so social traffic would also be driven from a variety of sources on Twitter, Facebook etc, and backed up by editorial content, email, telesales, direct mail and more, but as you can see here, our total site revenue shows a similar correlation.
Total site revenue over period:
Incidentally, that peak at the end of the month is down to human nature. As much nail biting as it creates, there will always be a last minute rush for remaining event tickets just before deadline, because people are lazy.
Here social actually gets a boost because people see a tweet saying ‘Last chance!’ and it nudges them into action at the last minute.
Social makes money
Once again, social isn’t about to give up an exact figure here, but by correlating data from a variety of sources we can get around these hurdles and show that it actively creates revenue.
This is a rather unique example, but similar data sets should allow you to identify the revenue created by events held on third party sites, and help to isolate social traffic driven by partners and other sources.