Marketing automation is taking things that used to be manual and automate them. It’s as simple as that. It’s great and soon most marketing efforts will be automated. It’s however easy to get caught up in nice media agency pitches and promises of a better ROAS.
I believe most people are overlooking an important point.
The case with the display banners
One of the infinitely many use cases of marketing automation is dynamic display banners. You build a banner template consisting of elements that change dynamically as the system learns from the campaign. It’s really just a small website build on a CMS about the size of a banner. The system is then constantly shuffling those elements, building new versions of the banner, usually with the end goal of finding the best performing combination.
Unless you have an advanced and complicated setup that is integrated into your business, then chances are that you look at the campaign as an isolated event and optimize towards getting the most traffic to the landing page. A high clickrate would in this case be a successful campaign.
When you work like this you will most likely also experience a higher landing page bounce rate and in the long run a lower lead quality. This happens because there is a disconnect between the winning versions and the landingpages/products.
I’d argue that until your dynamic advertisement is acting based on real backend sales data, and the landing-/product page automatically adjusts to match the different versions, then you’re better of with a few static versions that are tailormade for your high quality audience.
I believe it is better to become absolutely clear about which audiences to target and then build up the communication from there. Working your way out from an inside point of view. The campaigns might not seem as successful at first, but that just means that you’re avoiding potential bad quality. The real ROAS in the long run, including all factors, is the interesting KPI to be looking at, and this will tell you whether the campaign was successful or not.
It’s not a lost case but be realistic
Running dynamic display banners is the best and fastest multivariate test you can do. If you can manage to learn from them that is. I recommend that you quickly zoom into the 2 or 3 best performing versions based on CTR and optimize your landing pages accordingly.
Your overall conversion rate should then improve.
You don’t need 1.000 different version unless the campaign is based on real backend data and the landing pages are dynamic as well. If you outsource everything to your agency the money you will end up paying for micromanaging the campaign might not be worth the effort, and then you need to consider if the isolated campaign is worth the effort.