Optimizing Marketing Campaigns with Data Analysis

While my colleague Rafael put light on the problems which can come along with having too much data from too many sources in a situation of not so much time but the need for strategic decisions, I will focus on a more tactical topic: Data- and test-based campaign optimization.

Data and testing boost campaign results:

• via data-based idea generation before even briefing the creative department
• via pre-campaign data-based testing
• via In-campaign testing and analysis
• via DCO (Dynamic Content Creation)

Let’s get into it!

1. Data-based ideas generation

We all know that good ideas are the basis of any successful and efficient campaign. So, generating new ideas to test afterwards is always a good idea. Doing this also based on data-based insights has, not surprisingly, proved to be a very good idea.

When is it relevant?

Whenever you have any (non-aggregated) data about your customers, purchase behaviour, audience data etc.

How to do it?

Here we have two examples illustrating how powerful data can be to generate new insights which – provided to your creative department – immediately lead to a firework of ideas:

Example 1:

When we delved into the sales data of a popular coffee brand, we spotted a unique group of high-value customers: “Urban male high-volume black coffee drinkers." Even though the brand usually caters to women who prefer milky coffee flavours, we saw a chance to try something different.

In a surprising move, we created specific campaigns just for these dedicated black coffee enthusiasts. Despite the initial mismatch with the brand's usual audience, it turned out to be a smart strategy. We attracted more of these avid coffee drinkers, introducing them to the brand and turning them into loyal customers.

Example 2:

Working with another client, our goal was to uncover patterns that could guide future campaigns to success. We started with some educated guesses about what might matter, like "product types promoted," "promo mechanics," "segmentation levels," and "product prices." Then, we looked at key success indicators, such as "ROI," "new customer gains," and "re-engagement of inactive customers."

Next, we dove into the data, analysing different combinations within these categories. Our mission was clear: figure out which customer groups respond best to specific types of campaigns. The results not only shaped the upcoming months' campaign strategy, leading to significant improvements in key metrics, but they also sparked creativity. The now simple idea was, to ad specific campaigns with corresponding targeting to acquire more of these male heavy coffee drinkers.

So, example 1 was about the identification of (new) relevant segments in purchase data for addressing this segment with higher relevance. In example 2, we clustered campaigns by different criteria to types and searched for any patterns. Let's break down the process of turning data into effective campaigns:

• Know Your Data: Check what data you have and where it comes from. Sometimes, you might need to gather information from different sources in one place.
• Clean Up Your Data: Make sure your data is accurate. Clean up any messy or confusing bits before diving into analysis.
• Define Your Question: Figure out exactly what you want to know from the data.
• Make Educated Guesses: Come up with some initial ideas about what might influence the results.
• Set Up New Metrics: Depending on your questions, create new metrics (like customer lifetime value or loyalty scores) in your data system.
• Talk to the Experts: Discuss your plan with your data scientists. They might have more ideas or questions to consider.
• Check Your Results: After the analysis, go back to your data science team. Make sure you're interpreting the results correctly and address any deeper questions that come up.
• Brief Your Creative Team: Share detailed insights with your creative team, providing a solid foundation for new campaign ideas. The more they understand, the better their input will be.

Whether the outcome is fresh campaign ideas or strategies to boost success, this data-driven approach sets the stage for the next crucial step: testing!

2. Pre-campaign data-based testing

When is it relevant?

• Big Budgets, Big Wins:  When your campaign has a significant budget. The more you invest, the more you gain by optimizing assets and ideas with pre-campaign testing.
• No Room for In-Campaign Tweaks: When you know in advance that in-campaign testing isn't possible.
• Multichannel Challenges: When you have multiple channels for the campaign, and you can't change assets in all channels while the campaign is running.
• New Ideas on the Table: When you have fresh campaign ideas, perhaps inspired by insights from data analysis.
• End the Endless Discussions: Whenever team discussions about campaign details drag on for more than 30 minutes. Avoid endless debates and opt for testing instead.

How to do it?

• Define Your Goals and KPIs: Clearly define what you want to achieve with your campaign to guide your testing.
• Thoughtful Testing: Decide what you want to test. Whether it's different copy versions, benefits, format types, or overall ideas, plan it out. If you want to derive general learnings, you should first generate general hypotheses you want to test and conceive a test scenario, which will test exactly these hypotheses,
• Test Setup Basics: Ensure your test groups are big enough and free from biases. Set parameters like duration, budget, and target audience.
• Channel Selection: Identify relevant digital media channels and choose one that aligns with your test scenario.
• Tracking Setup: Install tracking capabilities in the selected channel if not provided by default.
• Budget Matters: Allocate a test budget sufficient for significant results, especially for campaigns with small budgets.
• Decision Time: For small budgets, select the test winner and launch. For larger budgets, analyse data collected during the test, refine based on insights, and repeat until desired results are achieved.

As you can imagine: The bigger and the more important the campaign for your business, the more you should invest in pre-campaign testing and analysing the data of the test results.

3. In-campaign testing and analysis

In-campaign testing is not often done properly due to the following challenges:

1. Race Against Time: When a campaign is already live and the daily budget is ticking away, optimization becomes urgent for overall success.
2. Complex Optimization Across Channels: Ideally, optimizations should roll out across all campaign channels, but it gets complicated and costly. What works in one channel may not work in another.
3. Collaboration Hurdles: Close collaboration between the creative and media agency is essential but often challenging due to various reasons:

• Budget Misalignment: Creative agencies get paid for additional work, while media agencies might not, creating misaligned interests, especially in smaller budget campaigns.
• Data Accessibility: Data or insights from data are usually housed with the media agency, creating a barrier for the creative agency.
• Process Bottlenecks: Well-defined and approved processes for fast testing iteration are often lacking, causing delays in decision-making.

When is it Relevant?

Especially for larger and longer campaigns, in-campaign testing is valuable because:

• You get insights from different channels, uncovering significant differences in asset performance.
• Enough data is generated to think of new combinations of campaign variants with specific targeting.
• Over time, a message, or asset might lose its impact, and testing helps identify stronger alternatives.

How to Navigate In-Campaign Testing:

1. Define Your Goals: Clearly outline what you want to achieve with your campaign. This guides metric tracking and data collection.
2. Plan Your Tests: Determine what and when to test in iterations (e.g. weekly) across channels.
3. Establish a Collaborative Process: Set up a streamlined process with the media and creative agencies. Ensure everyone understands their roles, competencies needed, and the importance of campaign success. 

• Data Generation: Define how UTM parameters will be exchanged.
• Data Sharing: Determine the process for sharing data.
• Analysis and Insight Generation: Follow insights generation practices mentioned for pre-campaign testing.
• Decision-Making: Schedule meetings for sharing, interpreting insights, and deciding on the next steps.

Remember, effective meeting routines and alignment are crucial for success in in-campaign testing.

4. Special: DCO

What is DCO?

Dynamic Creative Optimization (DCO) is a digital advertising technique that creates personalized ads for target audiences based on data on these audiences or other relevant data. For instance, a business can use the time, location, and even previous viewing history of a user. Based on this information, ads appear that align with the data, e.g. combining different visuals with different headlines or copy, resulting in a unique advertising experience. DCO is an effective way to personalize your marketing strategy and stand out from competitors.

Here is an example from some of our work: For Original Wagner and Buitoni CH we used Google audience segments and data signals to approach viewers with the utmost customized relatable product moments. With the Wagner Piccolinis DCO campaign, we showcased industry best practices in personalized campaigning – providing exceptionally good results and winning the MAXX award.

Please see here a small insight into the optimization within a DCO campaign for Wagner Piccolinis. This shows only 20 assets from a selection of more than 80 headlines using only one visual.

At DCO, as you see in the example, after the first data analysis and set-up of the campaign, data is used automatically. Based on the data signals of a single customer, the system chooses one of the various options e.g. copy and visuals to create an ad most likely to be relevant for the user. Also, see our case page, Wagner Piccolinis. 

When is it relevant?

DCO is relevant whenever...

1. Your target group is preferring different benefits of your product. E.g. one product feature is making your product perfect for families while another feature is making it perfect for single males, another for single females but also for music lovers etc. Then DCO will combining the different product features e.g. in corresponding headlines with background images addressing the individual showing families, males, females, musicians etc.
2. You only have small numbers of features or claims, but you want to personalise it down to small groups of special interest, e.g. fortnight gamers, fans of a specific pop, rock or metal group, downhill bikers, people with a specific dog breed, etc.
3. You want to show in your campaign your regionality, e.g. by referring to some regional habits, idioms, regional heroes, football clubs etc.

How to do it?

If you are interested in knowing how to successfully launch a DCO campaign with the described challenges above, feel free to contact us anytime!