Why Experimentation is Critical for Successful Marketing
by Matt, on 8 Jun 2021
As a business leader, you are constantly challenged to deliver predictable results – by investors, members of your team, and other stakeholders.
Telling them you plan to “run a few experiments and see what works best” doesn’t often land well.
They expect you to already know what will work. After all, isn’t that why you got the job and earn the big bucks?
I would argue that absolute confidence is a fallacy in any line of work, but there are certainly many roles where implementing proven practices offers a very high degree of predictability.
Not so much when it comes to marketing.
As we will discuss in this blog, the practice of continuous experimentation is proven and does offer the greatest chance of achieving a successful outcome.
What Do We Mean by Experimentation?
The word experiment comes from the Latin experimentum, the action noun from experiri, which combines ex “out of” and peritus “tested”.
So, literally, the action of discovering things by testing.
In our marketing application, experimentation is used exactly as the etymologists would hope – discovering what works through testing.
Why Not Just Implement Proven Marketing Practices?
We can certainly follow proven practices, but the devil lies in the details.
Let’s imagine that you want to run some social media ads.
You know that advertising on LinkedIn has proven to be a successful approach for reaching your B2B target audience – something that’s backed up by numerous recommendations from marketing gurus.
The use of LinkedIn Ads to reach a B2B audience could be called a proven practice, although it certainly isn’t infallible.
However, exactly how to configure those ads is anything but a cut-and-paste exercise.
Each ad combines multiple elements, such as:
- Target audience
- Text that is displayed in the user’s feed
- Additional text that is displayed when the user clicks “…see more”
- Any image(s) that is/are displayed in the user’s feed
- Whether the ad features on-page conversion (e.g. a LinkedIn lead generation form) or links to an off-site landing page
- The daily and weekly budget you set
- Which bidding strategy you ask the LinkedIn algorithm to follow
With a bajillion possible combinations, how you select and combine those elements is a matter of strategic discretion.
There are guidelines and best practices to help with each element, but one crucial factor means there isn’t a one-size-fits-all formula: it depends on the relationship between you, your solution, and your audience.
So What Are We Testing?
How much you know about marketing to your target audience will vary as your business grows.
In the early going, when you are selling to a very small, specific group of early adopters, you might know them quite well.
In many cases, the founders of a new venture have first-hand experience of the challenge their product or service helps to solve, so they are intimately familiar with the needs, wants, behaviors, and beliefs of their target audience.
As the business grows, however, it will want to reach beyond that limited circle of well-understood people to increase its market potential.
This might mean customers in different geographic locations, from different demographic groups, or who are solving the challenge for different psychographic reasons.
What type of marketing will appeal to them?
For example, which of the following images would be most likely to grab the attention of your target customer?
- A cool photo of your product or service in action
- A photo of someone experiencing the problem your product or service addresses
- A chart showing before and after data that demonstrates the value of your product or service
There are many other possibilities, of course, but unless your team has carried out some marketing experiments, your answer is probably “I’m not sure”.
Even if you have a hunch that one might work better than the others, how do you know? How certain are you?
This is where marketing experiments come in.
But wait, aren’t there a LOT of possible combinations to test?
There certainly can be, which is one reason why marketers are always testing!
The key is to test systematically.
Different Types of Marketing Experiment
There are several different types of marketing experiment you can run. Here are some important ones to consider.
Probably the best-known test type, A/B testing compares two versions of a marketing piece – such as an ad, email, or landing page – to see which performs better.
The two versions are delivered randomly to the target audience – usually by running two campaigns in parallel and allowing the delivery channel to decide who sees which version.
It is very important not to change too many elements at once. Keep almost everything the same and just vary one or two elements. Otherwise, it will be impossible to tell which change led to the observed difference in performance.
Examples of elements you might test are:
- Content (headline, above-the-fold text, below-the-fold text, subject line)
- Appearance (image type, image location, page layout, page color, font choice, font size)
- Call to Action (design, location, offer, conversion type)
Certain marketing elements might resonate better with one group of people than another.
As discussed earlier, those groups could be defined in geographic, demographic, or psychographic terms – or some combination of all three.
A common audience test is geo-targeting, which evaluates the effectiveness of different designs across multiple geographic markets.
This can be particularly important when testing visual elements since cultural differences can lead to people having very different reactions to some images and colors.
One of the creepier sides to marketing is how we can observe the detailed behavior of visitors to our digital properties.
Eye tracking is a popular approach that provides insight into how users navigate around a webpage. It produces a heat map showing where they spend the most time and how their cursor tracks between different elements.
This can help you understand whether they find the layout and content easy to understand, and whether it directs them to the places you want them to go.
It might also help you decipher the steps that are involved as they choose to move deeper into your website or to click on a call to action.
Where to Start
With so many possible combinations of content, imagery, and calls to action, the task of figuring out an optimum design can feel overwhelming.
However, not all potential designs are created equal.
Because you know something about your target audience, you should have a working hypothesis (or two) about which marketing designs are likely to work best. Your business strategy, brand positioning, and competitive differentiation may also influence what type of designs are possibilities.
Write those down and start there.
Build a base case design that incorporates what your experience and judgement tell you is most likely to work, then runs tests to confirm your assumptions.
If you’re entering foreign territory – either literally or figuratively – you might need to do some homework.
Study how others are marketing to your new target audience. This might be your competitors or businesses selling different solutions to the same group of people. Are there similarities between their approaches?
You might even ask the customer! We wrote recently about the importance of conducting customer interviews as your business grows, and this is a great example.
Whether you ask a few potential customers what type of ads grab their attention or run a focus group to test the effectiveness of conceptual marketing designs, the feedback you get from real life customers can be extremely helpful.
But, the most effective way to test is to actually collect live data. Try your value propositions, your ad copy, or your design using low cost online advertising to get some indicating data as to what resonates most – if you are able to filter your audience to a relevant group.
Does It Ever End?
In the introduction to this blog, we talked about continuous experimentation. Is that really what every marketing team should be doing? Are there times when it’s okay to stop testing and just be doing?
The short answer is that marketing is never perfect.
Even if you’ve found a combination of elements that measurably outperforms your other designs, it won’t work forever.
Long-running TV ad campaigns give the impression that some companies do find an everlasting formula. But don’t be fooled – even the relentless series featuring the AFLAC duck, GEICO gecko, Lily at AT&T, Flo at Progressive, and Jan the Toyota lady are constantly tested and optimized.
While the ads you see from those brands might seem very similar, they likely aren’t showing the same thing to other audiences. They’ve optimized what works for your market segment and will keep using it until its performance drops.
Reasons to believe continuous marketing experimentation is the right approach include:
- Your company will want to target different market segments – either to expand the current market or enter a different sector or geography.
- Your competition will adjust its marketing strategy to outperform you, so new ideas must be tested for you to stay ahead. If you’re standing still, odds are you’re actually losing ground.
- Your customers’ behavior and tastes will change, influenced by everything they see and hear across all different media channels. If you’re able to spot and test emerging trends, you’ll have an even better chance of outshining other players in the market.
- Just because your preferred approach outperformed everything else you’ve tested, there might still be an even better solution out there (what scientists call a local maximum – being at the peak of a mountain but not necessarily the highest peak in the range).
- Every experiment you run teaches you something new about your target audience, forcing you to question your ideas, beliefs, and best practices.
- Experiments are usually a lot less expensive than running a full campaign, so it’s a smart way to limit the financial exposure of trying something new.
When a marketing leader tells you their strategy involves running experiments to see what works best, give them your full support.
Conversely, if your company is running tried-and-tested campaigns that haven’t been updated in a while, make sure its because ongoing experiments have reconfirmed their ability to outperform newer ideas.
Be sure to set aside part of your overall customer experience budget for marketing experiments – and expect many of them to fail. If you aren’t trying and abandoning ideas, you’re not pushing the envelope far enough.
Nibbling around the edges (testing minor tweaks) or testing low-probability ideas (for example, advertising on channels where your audience spends very little time) are poor uses for your limited testing budget.
Stay focused on experiments that will generate new information about your audience and customers, and that will impact the direction of your marketing strategy.
Finally, be clear about what kind of data you want to collect and how you’re going to use it, and make sure that data is accurately recorded and reported.
A strategic, carefully managed approach to continuous experimentation will allow you to promise predictable results, even if you can’t say in advance precisely what those results will be.