HADI-cycles for hypothesis testing in marketing

Should your company implement HADI-cycles? Yes, if you want to quickly and effectively test marketing hypotheses. Here are the main advantages:

1. Rapid decision-making. HADI-cycles minimize the time spent on analysis and speed up the implementation of changes.
2. Risk and cost reduction. Companies can avoid major mistakes by identifying and correcting problems at early stages.
3. Evidence-based strategic decisions. This tool ensures higher accuracy and justification for strategic decisions.

It is evident that HADI-cycles are beneficial for companies aiming for flexibility and precision in decision-making. Now, let’s break down their structure to understand how they work in practice. HADI-cycles include four stages: Hypothesis, Action, Data, and Insight.

1. Hypothesis This is an assumption about what change might lead to improved metrics and how this change will impact the selected metrics. It is important to formulate hypotheses that meet specific criteria.

Criteria for a good hypothesis:

• Specificity. The hypothesis should be clear and specific so that it can be easily understood and tested.
• Measurability. This means that the metrics to be used for evaluating the success of the hypothesis need to be determined in advance.
• Testability. The hypothesis should be testable within a reasonable time frame and with available resources.
• Justification. The hypothesis should be well-founded and based on preliminary data analysis, research, or logical assumptions.

2. Action
   Test the hypothesis in practice. It is necessary to develop and implement specific actions to test the hypothesis.
3. Data
   Collect and analyze the data obtained from testing the hypothesis. This stage should be organized in such a way that it is easy and accurate to compare the results before and after the changes. This allows for making informed conclusions about the impact of the hypothesis.

The metrics should be directly related to the hypothesis and reflect the expected changes as a result of the experiment. They can be both quantitative and qualitative. Here are examples of different metrics for two hypotheses:

Hypothesis #1: “Adding personalized greetings in the email subject line will increase email open rates by 15%.”

• Open rate: The percentage of emails opened out of the total delivered.
• Click-through rate: The percentage of clicks on links within the email out of the total delivered emails.
• Unsubscription rate: The number of users who unsubscribed from the mailing list after receiving the email.
• Feedback on email content: User perception of personalized greetings through surveys or feedback forms.
• Perception testing: Interviews with email recipients to understand their reaction to the new subject lines.

Hypothesis #2: “Offering free shipping for orders over $50 will increase the average order value by 15%.”

• Average order value: The average amount spent per order.
• Total number of orders: The number of orders placed by users.
• Percentage of orders with free shipping: The percentage of orders over $50 that received free shipping.
• Sales revenue: The total revenue generated from orders with free shipping.
• Feedback on the free shipping offer: Collecting user opinions on the significance and attractiveness of the offer through surveys.
• Analysis of order reasons: Interviews with customers to understand how the free shipping offer influenced their decision to increase the order amount.

4. Insight
   Interpreting the data and concluding whether the hypothesis was successful. If the hypothesis is confirmed, scale the changes. If not, analyze the reasons and formulate a new hypothesis.

Now that we have detailed the structure of HADI-cycles, let’s move on to specific examples of their application in practice.

Examples of Using HADI-cycles

Case 1: Implementing a chatbot on an IT service provider’s website.

Hypothesis:
Implementing a chatbot on the website will reduce client query response time by 20%.

Action:

• Step 1: Identify typical client queries and frequently asked questions.
• Step 2: Choose a platform for creating and integrating the chatbot.
• Step 3: Develop initial communication scripts.
• Step 4: Write dialogue variants to ensure flexible responses.
• Step 5: Install the chatbot on the website for internal functionality testing.
• Step 6: Launch the chatbot for real client queries.
• Step 7: Identify cases where responses did not meet client expectations and make adjustments.
• Step 8: Relaunch the chatbot and start monitoring and data collection.

More time was required to create and test various communication scenarios to meet client expectations.

Data:
Over a month, the team collected data on response times for client queries via the chatbot and live operators, as well as the number of resolved queries.

Insight:
Data analysis showed that the chatbot reduced response time by 18%, but did not reach the target of 20%. It was also found that clients frequently asked complex questions that the chatbot could not resolve without operator intervention. The company decided to continue training the chatbot to improve its efficiency.

Case 2: Optimization of the Client Case Study Page for a Consulting Firm.

Hypothesis:
Optimizing the client case study page will increase the time spent on the site by 20%.

Action:

• Step 1: Collect data on the current user time spent on the page.
• Step 2: Study user experience to identify key issues and shortcomings.
• Step 3: Design a page layout with interactive elements, videos, and infographics.
• Step 4: Approve the design through internal review.
• Step 5: Develop the new page and upload it to a test server.
• Step 6: Test the page and then transfer it to the main server.

The design approval process between different departments was prolonged. The large volume of multimedia content increased page load times. Content optimization and server performance improvement were necessary.

Data:
For two weeks post-launch, data was collected on the time users spent on the case study page and overall time spent on the site.

Insight:
Time spent on the updated page increased by 12% (falling short of the 20% target). It was found that users had difficulty navigating the new page and could not quickly find the information they were interested in. This prompted the development of a new hypothesis, as the company did not want to increase time spent on the site through navigation issues.

Case 3: Development and Implementation of a Proposal Builder for a Building Materials Supplier.

Hypothesis:
Developing and implementing a proposal builder for the sales department will increase conversion to the first deal by 15%.

Action:

• Step 1: Collect data on products, prices, and sales conditions.
• Step 2: Create a PowerPoint template including all necessary sections and elements.
• Step 3: Discuss the template with the sales department and make revisions.
• Step 4: Develop the commercial proposal template.
• Step 5: Train the sales managers.
• Step 6: Collaborate with each manager to create 20 new commercial proposals.

PowerPoint was used for its accessibility and simplicity. Some employees faced difficulties adapting to the new format as they were accustomed to the old methods.

Data:
For two months, data was collected on the number of commercial proposals created and the number of deals closed.

Insight:
Data analysis showed that conversion to the first deal increased by 26%. The proposal builder was decided to be retained and further refined.

Case 4: Advertising to Clients Already in the Negotiation Stage.

Hypothesis:
Advertising to clients who have met with a manager but have not yet signed a contract will increase the conversion to payment by 10%.

Action:

• Step 1: Conduct a photoshoot of the team.
• Step 2: Develop banners with photos of the managers.
• Step 3: Export the database for advertising from the CRM system, segmented by sales funnel stages and the managers the clients met with.
• Step 4: Launch contextual and targeted advertising.

At some sales stages, there were not enough clients to launch the ads, requiring combining different stages.

Data:
Over a month, data was collected on the number of clients who moved to the next stage of the sales funnel.

Insight:
Conversion to payment increased by 6%. While the hypothesis was not fully confirmed, it still significantly impacted the company’s revenue, as it operates in the B2B sector. Some clients called their managers to confirm if they had just seen them on the internet.

In practice, HADI-cycles can come with various challenges and issues. To avoid common mistakes and make the most effective use of this tool, it is crucial to understand the typical errors that may arise and know how to prevent them.

Top 7 Mistakes When Using HADI-Cycles

1. Unclear Hypothesis Formulation Mistake: Hypotheses are formulated vaguely or too broadly, making them difficult to test and interpret. How to avoid: The hypothesis should be measurable and have a clear success criterion.
2. Choosing the Wrong Metrics Mistake: Using metrics that do not reflect the real changes caused by the hypothesis being tested. How to avoid: Carefully select metrics that are directly related to the hypothesis and allow for objective evaluation of its impact.
3. Ignoring Seasonality and External Factors Mistake: Overlooking seasonal fluctuations and external factors that may affect experiment results. How to avoid: Conduct repeated tests at different times to check the stability of results.
4. Disregarding Qualitative Data Mistake: Focusing solely on quantitative data and ignoring qualitative aspects. How to avoid: Include both quantitative and qualitative data in the analysis.
5. Simultaneously Testing Multiple Hypotheses Mistake: Testing multiple hypotheses at the same time on the same sample, complicating result interpretation. How to avoid: Test hypotheses sequentially or on different samples to avoid cross-influences.
6. Neglecting Negative Results Mistake: Ignoring or underestimating negative results, which can lead to repeated mistakes. How to avoid: Analyze negative results as thoroughly as positive ones, as understanding the reasons for failures is crucial for improving future hypotheses.
7. Overcomplicating Hypotheses Mistake: Formulating overly complex hypotheses that are difficult to test and interpret. How to avoid: Break down complex hypotheses into several simpler ones for step-by-step testing.

Understanding these mistakes and working to prevent them will help you build more accurate and well-founded marketing strategies.

Effective marketing is data-driven and flexible marketing. HADI-cycles give companies the opportunity to experiment and learn from their mistakes quickly and at a reasonable cost. I sincerely wish you success in applying HADI-cycles in marketing!