Funnel Analysis Case Studies: Training Facebook’s machine learning, adding WhatsApp CTA, reducing privacy anxiety (examples in 3 industries)

This article was published in the MarketingSherpa email newsletter.

“OK, I complied ads and hypotheses together in a Sheet. Your feedback on all of it would be appreciated.” This comment is from the MECLABS Super Funnel Research Cohort group on LinkedIn, where marketers are working together to increase conversion on their digital marketing funnels.

To spur your own best thinking for improving your funnel, in this article we bring you case studies from your peers at three different funnel stages.

First, an e-learning provider that added a WhatsApp CTA to increase conversion. Then, a sock brand that tested ad spend strategy to make the best use of Facebook’s machine learning algorithm after Apple’s iOS14 privacy update. And speaking of privacy, we end with a case study from a real estate firm that improved privacy communication on a form and increased conversion.

To be clear, the team at MarketingSherpa and MECLABS did not perform any of the work in the below case studies. All credit goes to the brilliant marketers quoted below, and the teams they work with. Just like with the funnel group on LinkedIn, our goal is to bring marketers together, and provide evidence and training to spark new ideas in you as you address your greatest challenges and seize your biggest opportunities.

Quick Case Study #1: How e-learning provider is getting a 405% higher conversion rate on WhatsApp

“We learned the importance of good calls to action (CTA) the hard way,” said Iqbal Ahmad, Founder & CEO, Britannia School of Academics.

When the team started marketing its accredited qualifications online two years ago, they used to have more than 2,000 learners visit their websites every week, most of them through paid clicks. And yet, only a handful would convert. Slowly and gradually, they started making improvements to the landing page experience that included more effective CTAs, such as live chat and the ability to apply online.

“Through Mouseflow, we noted that a lot of learners would start the online application form but close the window looking at the size of the form,” Ahmad said. The team added an additional CTA – “Quick Enquiry Form,” whereby the learner could submit a query by filling a few details (please note: Britannia is based on the United Kingdom, and thus, used the British spelling of “inquiry.”)

This change brought a much better interaction rate, and it allowed the team to start a chain of communication with the learners – an opening they were not getting through the detailed application form.

While this improved results, they continued to work to optimize the CTA and discovered a strategy that changed the way they used to think about a CTA. “While browsing through our CRM, we saw an option to integrate WhatsApp contact feature on our website,” Ahmad said. The team thought…why not give it a try? After all, it was free of monetary cost, so they felt they had nothing to lose.

So they added a WhatsApp CTA to the website.

Creative Sample #1: “WhatsApp Us” CTA on education provider’s mobile website

When visitors clicked on the CTA, they were given the opportunity to continue to WhatsApp on their phone, or download WhatsApp if it wasn't already installed.

Creative Sample #2: After clicking on WhatsApp CTA on website

Once the chat starts, the team can then exchange messages back and forth, until the learner agrees to pay the fee and start the qualification. "We use Bitrix24 to integrate all chats, calls, and emails into one lead management system. Our IT guy helps us with that, but I think each CRM is different," Ahmad said.

People who engaged on WhatsApp had a very high conversion rate – 91% (31 enrolled out of 34) signed up for the courses by making full payment. This is in sharp contrast with an 18% to 40% conversion rate when the learner initiated contact through other modes of communication. For example, live chat has an 18% conversion rate (134 enrolled out of 732).

The team carried out an in-depth analysis of such a substantial difference in the rate of paid conversions and made the following discovery.

When a learner fills out an inquiry form and the team responds by email, it often lands in their junk/spam folder or is simply ignored if they are busy at work or with family at the time that they receive the response. Within a few hours, the email gets buried under a pile of other marketing- or product-related emails that the learner gets every day. Or the learner may simply lose interest because of the delay in replies back and forth.

Live chat communication has its cons, too. The team loses touch with the learner as soon as they close the website, app, or simply put down the phone.

The situation is very different when the potential learner initiates contact through WhatsApp. First, the team has the contact number of the person and their willingness to have a relatively informal chat to discuss needs and any concerns. Second, there is no risk of messages going to junk, being blocked, or being buried under other marketing messages. Even if they get busy, the team resumes the chat as soon as the learner opens WhatsApp again. Sooner or later, the learner tends to convert since often the team has the right qualification training for them, which is why the potential customer reached out in the first place.

“We know that if a learner starts a WhatsApp chat, all bets are on us converting the lead,” Ahmad said.

Now, the main issue the team wants to address is that most people prefer live chat over WhatsApp chat.

The team is planning to start a pilot on one of their websites where they will further highlight the WhatsApp chat option and show the live chat feature as a less-highlighted option. “There is a high probability that the proportion of WhatsApp inquiries in relation to other modes will increase, and our overall conversion rate will increase accordingly. Even if it does not, the fact remains that we are converting many more leads – thanks to WhatsApp chat feature,” Ahmad concluded.

Quick Case Study #2: Sock brand gives Facebook machine learning algorithm “room to run” by testing cost caps, increases gross income to $195,000 in September

"Over the last few years, our Facebook advertising results have either been slowing down or barely holding steady. It seemed the opportunity for growth in that channel was diminishing for a variety of reasons,” said Nate Banks, VP, Crazy Compression.

One of the reasons may be Apple’s iOS 14 update. “For [the] local, high-grossing sock producer and manufacturer, we found after the iOS changes that the results had dipped to a point of breakeven,” said Chris Anderson, Owner + Digital Marketer, Social Ktchn (Crazy Compression’s social media advertising agency).

While the margins for this brand didn't need to be exorbitant, they did need to be better than what they were getting.

Creative Sample #3: Facebook ad for sock brand

“We decided that Facebook's ‘Lowest Cost’ optimization was not cutting it, and tested Cost Caps. These are notoriously difficult to 'tame,’ but we needed something that would produce both higher efficiency (return on ad spend) as well as a solution that would allow us to scale without losing ROAS,” Anderson said.

Training the machine learning algorithm

In order to test, the team needed to choose a cost per result that wouldn't be “too greedy.” They initially went with $32.15, which is essentially a breakeven number. They also chose a daily budget of $5,000. This would allow the machine learning algorithm room to run.

After two to three days, they were not seeing much spend at all, and the campaign was still in the learning phase. The next few days were higher spending, without great results. At around day four to five, it hit for the team. They had a spend of $300 that day with a ROAS of more than 3x. Then, they worried they had gotten in a bit of trouble – the spending increased to $2,000 in one day.

The good news: the ROAS stayed at 1.5x, which was a huge profit. Again, for this particular business, the manufacturer also being the seller makes for great margins and not needing 4-10x to give great returns. On top of that, the more they sold, the higher the margins. So, volume really was the focus, as long as ROAS was above 1.5x.

After some adjustments and additions, they employed several cost cap campaigns, using lower CPAs (cost per acquisition/sale) and even higher daily budgets, across multiple interests and lookalike ad sets. This approach brought them from $75,000 spent and $90,000 made in August, to $107,000 spent and $195,000 made in September, which broke the record for most gross income in a month. The sock brand had to hire six more employees just to keep up with production.

Overall, they were able to control the CPA in a way that allowed them to scale with great profits, compared to the “lowest cost” setting they originally were running for the client.

“When the cost cap was brought to my attention, it seemed like a new and innovative way for Facebook to deliver our ads to the right customers. The results have been incredible and now the bulk of our investment is in Facebook Ad/Cost Cap," Banks said.

Quick Case Study #3: Real estate firm increases conversion 14% by adding privacy message

Throughout the Purple Egg Real Estate website, there is a form that people can fill out to get an offer on their home. Having customers fill out the form is the main goal of the website.

Creative Sample #4: Form on real estate website

It’s a simple form with only three fields, and it’s the same form that came with the templated website with only a small modification.

“As any marketer, I am always looking for ways I can increase the number of leads,” said David Hampshere, Founder, Purple Egg Real Estate.

One day Hampshere was playing with the form, which has a second step with a lot more questions.  On this second step at the very bottom, he saw an image about privacy.

Creative Sample #5: Privacy message

“I’m kind of a privacy freak, so such statements resonate with me. However, I thought it was pretty strange to inform the customer that we care about their privacy after they filled out both forms and are ready to hit the submit button for a second time,” Hampshere said.

The website already had a privacy policy link in the footer, but he decided to see what would happen if he copied this privacy message and made it the first thing to show up in the first step of the form.

Creative Sample #6: New form with privacy message on real estate website

Hampshere ran the new form with the privacy message for two months and compared it to the previous form that did not have the message. The form with the privacy message generated a 14% higher conversion rate.

Creative Sample #7: Result of privacy message implementation

	Impressions	Leads	Conversion
No Privacy Policy	856	52	6.07%
With Privacy Policy	913	63	6.90%

“As a side note, I regularly check my rankings in Google results. I was quite surprised to see this statement about privacy popping up in the results!” Hampshere said.

Creative Sample #8: Privacy message on search engines results page

“I did a Google search on why Google was doing this instead of using my meta description (I verified the page had one). While there was no definitive answer, one suggestion was that Google may elect to choose unique words on the page and ignore the meta tag.  It’s also interesting that the other verbiage is a rotating testimonial, so it looks like Google ignored my meta tag completely,” he said.

Related Resources

Marketing Funnel: 3 quick case studies to help you increase conversion

Messaging – Case studies and research from MECLABS Institute and MarketingSherpa

Marketer Vs Machine: We need to train the marketer to train the machine