Reduced Lead Cost in Meta Ads by 3.8x, Testing 420 Hypotheses in Ad Combinations Over 3 Months

Key Takeaways from the Case

The project is a startup, so demand and interest in the product were not yet established. In the early stages, we had to quickly test various combinations in our ads and landing pages.

Our goal was to make the advertising channel profitable and find an effective strategy for generating leads from Meta Ads for B2C clients in the UAE.

Key stages that helped us achieve results:

  • Continuous growth hypothesis testing for 3 months: we tested 420 ad combinations across 64 campaigns with 4 different landing pages.
  • Development and launch of a landing page in just 2 days.
  • Setting up automated lead qualification based on additional questions in the application form.
  • Establishing a traffic source identification system for leads from WhatsApp (transmitting UTM tags and ClientID into the CRM).
  • Creating Meta events for optimizing advertising campaigns.

Results of our efforts:

As a result of the hypothesis testing and the operation of a single campaign, we acquired 61 leads at a cost of 27 AED each. Based on this advertising campaign and forecasted ad spend metrics, we concluded that the B2C segment for the project can be scaled up.

How the Project Changed Users’ Lives

Thanks to the advertising, users became aware of the furniture rental service in Dubai. Meanwhile, the business established a functioning advertising channel that can be scaled.

Business Challenge and Its Solution

We worked under tight deadlines. Within 3 months, we needed to establish effective advertising as a lead source for further scaling.

The entire process can be divided into 3 major phases of testing different landing pages.

  • Phase 1: Marquiz Test

Test period: from June 17, 2024, to July 20, 2024.

Results: We acquired 67 leads at a cost of 66 AED each. The outcome showed that quizzes as a landing page did not perform well; the applications received were low quality but within the KPIs. As a result, we redirected traffic to the main website and kept the quiz in a pop-up on the site.

  • Phase 2: Website Test

Test period: from July 20, 2024, to July 31, 2024.

Results: We garnered 20 leads at a cost of 145 AED each. The findings indicated that traffic to the site could be scaled; however, while lead quality improved, we needed to work on reducing lead costs. We decided to create a dedicated landing page.

  • Phase 3: Landing Page Test

Test period: from August 1, 2024, to August 18, 2024.

Results: We obtained 61 leads at a cost of 27 AED each. We discovered that Facebook ads could generate high-quality leads for 27 AED, which was an acceptable cost for us. Most importantly, this combination showed the potential for scaling up.

Craftsmanship, Implementation, and Technical Details

The success was influenced by a systematic approach to internet marketing. We implemented continuous growth hypothesis testing through weekly sprints.

Here’s how our hypothesis testing is structured, illustrated by an example of a successful hypothesis with the launch of a landing page (LP) completed in two days.

  • [Hypothesis]

The hypothesis was: “If we drive traffic to the LP, we will receive 32 applications (with a CPL of 100 AED).”

  • [Metrics for Tracking]

All metrics for tracking were entered into a shared table, allowing us to create a “Plan-Fact” comparison for the values.

  • [Statistical Significance]

For each hypothesis, we establish a metric for statistical significance. When enough data is collected, we can assess whether our hypothesis is successful or not. We use calculators like Mindbox for this purpose.

In our case, the metric was clicks. With 570 clicks on the ad, we would gather enough data to determine whether the hypothesis was successful or not.

  • [Mockup]

Before launch, we create a mockup to visually present the hypothesis. In this example, we developed an LP as a new landing page.

We also document the creatives tested in each hypothesis, categorizing them as effective or ineffective after the test concludes. Effective creatives are scaled, while ineffective ones are disabled.

  • [Audience]

We specify the target audience for each hypothesis:

  • Broad, ages 22–55: UAE, all languages;
  • Interests, ages 22–55: rental + Dubai;
  • Interests, ages 22–55: children and parenting;
  • Broad, ages 22–55: singles.
  • [Execution Time]

We aim for each new hypothesis to take no more than 6 hours to launch. In this case, team preparation took around 16 hours.

  • [Actions]

We outline actions step-by-step, using the SMART approach.

  • [Budget]

We document the budget for each hypothesis. In our example, it was set at 3200 AED.

  • [Result]

Once statistical significance is reached, we draw conclusions and set the next hypothesis.

Result of the LP Testing Hypothesis

The hypothesis was successful. We achieved a cost per lead (CPL) of 39 AED, which was 2.56 times lower than the projected success.

Conclusion for Future Searches

As a result, we identified a successful combination in Meta ads that is well-suited for scaling lead generation and sales.

Through subsequent tests, we were able to expand upon this result, optimizing ad campaigns to obtain 61 leads at a cost of 27 AED each.

Insights, Hypotheses, Creation Process, and Client Collaboration

We launched the landing page (LP) in just 2 hours, a breakthrough that allowed us to quickly identify a successful advertising combination and scale the results.

We integrated artificial intelligence capabilities into the landing page creation process.

Steps Taken

  1. Developed the Structure and Mockup: Created the landing page structure and designed a mockup in Figma.
  2. Used a Plugin for Fast Transfer: Utilized a plugin to quickly transfer the mockup while retaining all layers and styles from the client’s website.
  3. Built the Landing Page: Developed the landing page in accordance with the design of the main site, making sure it was mobile-responsive.
  4. Edited a Video for the First Screen: Created a vertical video for the landing page’s hero section.
  5. Assembled Required Blocks: Compiled the necessary blocks and transferred them to the Tilda platform.
  6. Adapted for Various Screen Resolutions: Ensured all blocks were optimized for different screen sizes.
  7. Published the Landing Page: Launched the landing page on a subdomain.

After the website was created, we proceeded with the advertising setup.

Advertising Setup

  1. Established Tracking Systems: Set up tracking systems for traffic sources for applications coming through WhatsApp messages (transferring UTM tags and ClientID to the CRM).
  2. Configured Target Events for Meta Ads: Adjusted target events for Meta ads.
  3. Automated Incoming Lead Qualification: Developed an automatic system for qualifying incoming leads.

As a result, advertising campaigns were optimized based on CRM data concerning qualified leads and successful deals.

Outcome

We developed an effective algorithm for rapid setup that enables us to acquire leads at half the cost compared to other landing pages, achieving a share of qualified leads exceeding 70%.

  • Meta has been designated as an extremist organization in Russia.