Interview: Leveraging Data in Ad Design
Back in July, I was part of a panel with Shanee Ben-Zur, CMO at Crunchbase, at Superside’s Ad Designpalooza. As part of our panel discussion hosted by Superside’s VP Marketing, Amrita Mathur, we were asked a few questions centred around data and advertising design, which I’ve answered below.
If you are keen to understand when and when not to test your creatives, how to prove incremental value, secure buy-in from leadership, and what the future of advertising holds, then this article is for you.
Everyone knows they should be testing their ad creative, but is there a time when you shouldn’t?
In my opinion, it depends on several things, including the digital marketing maturity of your organization, how large the sample size you want to use is, whether or not you have enough budget and stakeholder alignment, and if you have good measurement in place.
Depends On Your Digital Marketing Maturity
Companies who are nascent in their digital marketing maturity are most likely using manual processes that take time. With limited resources and very manual process in place, it’s likely that the simplest solution that delivers the highest impact with the least cost to time and budget will suffice. Sometimes this can be good old managerial intuition, and other times, this might materialize as running tests with much more limited scope such as A/B testing vs multivariate.
Depends On Your Sample Size and Quality
Regardless of how far along your organization may be in your digital marketing journey, if you don’t select the right sample or a large enough sample size to test with, then the conclusions you draw from them will not be generalizable to the larger population. For example, imagine trying to select a restaurant for ten people but you only consider the opinions of two? Even if the two people you’ve considered like the same restaurant, it’s highly unlikely that this would remain the same if you asked six. Furthermore, you have to ensure that the people you ask even like restaurants to begin with or you’d introduce a lot of bias into your testing. Similarly, with ad-design, you wouldn’t want to waste your time testing if you can’t drive enough traffic to the creatives you’re testing or if they have no relevance to the audience to whom you are showing them to.
Not Enough Budget Or Resources
This one’s easy. If you don’t have enough budget, you run the risk of your tests never reaching a point where significant conclusions can be drawn i.e. statistical significance. This could mean many things in the advertising world but a good example would be having your ads go offline due to insufficient budget. In scenarios where there are not enough resources (e.g. people, time) you may need to simply rely on your intuition to set an initial hypothesis and then use that as baseline benchmark to measure future performance against.
When Stakeholders Are Not Aligned
Different stakeholders measuring success by different standards and if you don’t have that alignment prior, one stakeholder’s success could potentially be another’s failure. It is always a good idea to gain buy-in from key stakeholders about whether a test is necessary and if the results of the tests would be put into action as opposed to simply languishing in ego-stroking win-books.
When You Can’t Measure Or Track Your Tests
If you are still dealing with deduplication issues, pixel tags not firing, attribution woes, or don’t have any measurement in place, then I would avoid testing until you are confident in your ability to generate fairly clean data consistently.
What are your thoughts around data squashing creativity? Has marketing become a numbers game only?
I believe that data and creativity need to be applied simultaneously in today’s digital advertising world. While it can be tempting to lean into one or the other, a lot of the hypotheses that inform testing come from qualitative observations, while quantitative results generate the insights for subsequent tests. Creativity is a precursor to innovation in my opinion.
Your insights are only as good as the data that powers them, so relying too much on metrics rather than using a balanced approach between creativity and data may lead you to draw false conclusions if your underlying data is corrupt (remember the restaurant example from above).
Is there such a thing as too much data? Tips for focusing on the key data points rather than trying to combine it all?
Yes. If you are trying to look at every single detail of an image, you will miss the whole picture. Whenever you are testing your ad design, there needs to be some kind of prioritization or weighting that allows you focus on key things that drive impact. It may be great to test a blue versus red button, but is that as valuable as testing messaging that may appeal to more profitable verticals?
Simply put, you’ll have to make trade-offs due to the age old problem of scarce resources. In this case, while there is a ton of data out there, you don’t have the time to analyze every datapoint, so you’ll have to be agile and prioritize according to what your targets are, what your managers care about, and most importantly what your target audience values.
In my experience managing many different campaigns, I find that most managers care about the return on ad spend or profitability, while most consumers care about relevance of the ads.
Advertising is expensive, and as such, people will always want to know how the budget is being spent, and the performance of it. Do you have any advice on communicating your ad (or other marketing) performance upwards?
Focus on the key metrics that both your internal stakeholders and customers care about. Most managers care bout return on ad spend (ROAS), cost per acquisition (CPA), or lifetime value (LTV). They also care about marginal cost (i.e. how much it will cost them to acquire one new customer) and marginal return. Managers also want to know about the opportunity that exists if there was more budget available.
The best way to surface these results is in a dashboard in my opinion. That way, it empowers your key stakeholders to assess performance how they see fit, but also reduces the time impact to you.
If you are communicating performance to marketing stakeholders however, there may be more interest in the details (e.g. test performance, the optimizations that led to results, etc). Regardless of who you are communicating to, it’s important to discuss and secure buy-in for next steps.
What are your thoughts on copy-based AI especially for small teams?
I love it. You’re probably already using copy-based AI if you’re using Google Ads. Google Ads has a feature called Dynamic Search Ads and Responsive Search Ads, which essentially use machine learning to surface headlines, descriptions, and keywords in ad-copy in the combinations that are most likely to lead to a conversion.
Automation saves time and I’m a huge advocate of leveraging as much automation as possible, more so if you are a small start up without enough resources.
Are there specific data points we should look for when it comes to top-of-funnel marketing assuming if you’re a young company?
There’s a good chance that if you are a young company, you are still trying to build awareness and capture intent. It wouldn’t be strange to focus on traffic to site but unless you can convert that traffic efficiently, you will end up burning through your budget. It’s important to consider some kind of cost per acquisition or return on ad spend.
However, the answer to this question also depends on what type of company you are and what your marketing objective is. For example, if you are a blog and only care about whether people read articles, you may consider new users to site, time spent on page, and metrics that are more relevant to engagement. However, if you are an e-commerce site, then no matter the stage of your digital marketing journey you should be concerned with ROAS and sales!
The important thing here is to start small and continuously optimize until you hit the sweet spot between growth and cost!
What data analytics tools would you recommend?
There are a plethora of tools out there but some of my favourites are:
What do you predict for the future of ad data? Are we going to be able to see everything, or will there always be some ambiguity?
We are already seeing shifts in the privacy landscape which will make it harder for marketers to target. However, I think the future of ad data lies in streaming video, affiliate marketing, blockchain advertising that distributes revenues fairly, and hyper-personalization.
If you’d like to understand how your business can grow regardless of external conditions, feel free to get started with us at Stallion Marketing.