How to set up a system that helped us increase net profit margin by 4 times? E-Promo team shares how the implementation of HADI cycles, test schedule and continuous hypotheses generation grows revenue.

Every growth strategy heavily relies on generating and testing various hypotheses: from UX/UI to landing pages to media buying tactics.
We will expand on how A/B testing improves advertising effectiveness.
A/B testing is one component of a comprehensive advertising metrics optimization process. It can be used to collect both quantitative and qualitative metrics. With A/B testing, you can find pain points that you can work on later, and if you don’t do it, you can lose out on potential revenue.

Why do A/B testing?

Every day, companies develop millions of hypotheses to understand how to make their businesses more efficient and their customers happier.

It’s important for startups, small and mid-size businesses to understand how to be painlessly profitable and scale in an environment of uncertainty and competition. Therefore, understanding how to test quickly and well is critical to the development of a marketing strategy. For this purpose a tool like the HADI cycle is used. Many people have heard of it, but as practice shows, they rarely use it.

We apply the HADI cycle throughout the entire process of working with the client.

HADI cycles are relevant to startups and other early-stage businesses to save time and money when you need to launch, grow and scale quickly. However, any existing company can use this tool when launching new products/projects or when various changes are required.

In Internet Marketing, any meaningful metric can be used to help measure performance: CR, CPL, CPO, ROI. Using the HADI cycle, it is important to understand which specific business metrics we are trying to improve by testing.

Multiple hypotheses can be tested simultaneously using a single HADI cycle. However, it is important to ensure that they differ in efficiency.

It will be difficult to judge which idea was correct if you test two hypotheses to improve the same metric in the same HADI cycle.
Based on the business objectives of the customer, we formulate a set of hypotheses, which we may supplement and modify in the process of refutation/confirmation and depending on the results of the previous formulations. In other words, we create the framework of an advertising campaign with a specific set of channels and platforms, and communications appropriate to the achievement of business objectives.
As the work progresses — that is, as hypotheses are tested as data accumulates — the structure may change, sites may be added and removed, communications and USPs may change, and KPIs may be adjusted. Once the final results are in, the cycle can begin again, with new hypotheses for new purposes.

There are two things you need to do before you start working on the test hypothesis:

• Identify the conversion target
• Identify the problem and work out the wording of the problem

Evaluation of the results takes place in several stages:
- After testing the initial hypotheses and accumulating primary data
- In case of changes / adjustments to the strategy (changes in the structure of the advertising campaign), compare before and after
- After one cycle of testing, collect data, analyze, draw conclusions on how effective the built structure was and whether the campaign is ready for scaling and launching the next cycle of testing.
s not confirmed: after the price adjustment, the number of orders decreased and the cost of order acquisition doubled
As you can see from this example, not all hypotheses focus on getting additional advertising dollars to increase the effectiveness of the campaign.
As with everything else, it’s important to prioritize your hypotheses at the start of the testing cycle and to act on the ones that seem most relevant at the time — the ones that can be done most easily and that can yield the greatest return for the least amount of resources, such as increasing the size of the “Buy” button on your site.

More examples

An online store needs to increase two metrics:

• The number of orders added to cart
• Hypothesis: You can do this by making the “Add to Cart” button on the page bigger
• Average order value
• Hypothesis: Adjust the prices on the site and in the advertising campaign

Since the hypotheses test different metrics, it is possible to test them in a single HADI cycle.

Outcome:

• The hypothesis was confirmed: users started to add items to cart 1.5 times more often
• The hypothesis was not confirmed: after the price adjustment, the number of orders decreased and the cost of order acquisition doubled

As you can see from this example, not all hypotheses focus on getting additional advertising dollars to increase the effectiveness of the campaign.

As with everything else, it’s important to prioritize your hypotheses at the start of the testing cycle and to act on the ones that seem most relevant at the time — the ones that can be done most easily and that can yield the greatest return for the least amount of resources, such as increasing the size of the “Buy” button on your site.

HADI cycles are great for testing tactics, audiences, and creatives within each paid promotion channel.

For example, for one of our clients, Europlan, the market leader in car leasing in Eastern Europe, we used HADI cycles to test the following:

• Audiences and creatives
• Hypotheses based on auditory analysis
• New platform products and beta tests
• Approaches and tactics
• Different mixes: tactics+creative

As a result of testing and audience segmentation, the total number of creatives in rotation increased by 45% year over year.

All of this related to one of the key drivers of client development, assets, which included an omnichannel approach, personalization of creatives, adaptation of creatives from online to offline and back again, and UX.

Testing of new tools allowed to increase the number of contracts by 10%. Regular updating of creatives in rotation allowed to maintain a high level of CR with the client.

Read more on how to scale your business with a systematic approach to creative tests in our dedicated case study. In the article we go into more detail on how to find a winning creative that increases the number of users placing orders, which led to x4 increase in net profit margin for one of our clients. Interested in testing out different sales funnels? Find out more tips on sales funnel testing in another article.

Key goals and variables for marketing tests

Marketing testing variables:

• Calls-to-Action (CTAs)
• Offers (such as an ebook, a webinar, a free trial, etc.)
• Headlines
• Pricing scheme
• Targeting
• Video vs Text
• Frequency and timing of publications
• Images
• Landing Page

Examples of testing goals:

• Increase site traffic
• Increase CR
• Decrease Bounce Rate
• Reduce the number of abandoned carts

Recommendations for implementing the tests

• The tests have to go fast, you have to keep track of the test results (whether it’s a creative or a funnel). It will not be difficult to improve and scale the change if it is beneficial.
• It is necessary to plan the tests in such a way that there are no large gaps between them. This would delay the implementation of the marketing strategy.
• Identify the most important business metrics at this stage of the business and generate only those ideas that will have a direct impact on that metric.
• Prioritize hypotheses. The most important and significant hypotheses for the business should be tested first, and the weakest should be discarded to avoid wasting time testing.
• Don’t give up too soon. If the first few hypotheses do not yield results — do not give up. Keep testing.
• Interpret the data correctly and in a timely manner. To see if the implementation of the hypothesis is going in the right direction or not, you need to check the analytics on a daily basis.
• It is not necessary to allocate a large budget for hypothesis testing, especially if the marketing budget is minimal. An accurate hypothesis, even with a minimum budget, will show results.

What else to consider when testing:

• Run only one test at a time (for example, if you run two different tests simultaneously to increase the number of conversions — test ads and landing — it will be impossible to figure out what influenced the result)
• Test only one variable per test (either LP, or CTA, or headers, etc.)
• Conduct even small tests, if they can affect the result (for example, changing the size of the “buy” button on the landing page)
• Test parent elements (for example, create a second version of the landing page and test it against the original)
• Assess changes in performance across the funnel: not just CR landing, but also sales and margins