How Swvl’s Facebook Ads unlocked a 15740% higher engagement rate, 11745% more shares and 444% more searches on its App!

Swvl
6 min readMay 20, 2021
Swvl Hyper Personalised Facebook Campaign
How Swvl is hyper personalising digital advertising

New user acquisition has always been a priority for Swvl, but our approach had primarily been blanket campaigns; the communication was not tailored to the audiences, the audiences themselves did not have much segmentation, and the channel of choice was by and large limited to Facebook. The result: our campaigns weren’t breaking through the clutter.

Naturally the performance from these campaigns from a brand and business perspective was quite erratic. What’s worse is that the public sentiment on the ads was mostly neutral, occasionally even negative. It was clear now that the old way of doing things wasn’t working. It was time for a revamp.

We decided the way forward was hyper personalised marketing that leveraged data and creativity in an innovative manner. We had to do something that not many brands or businesses have been able to achieve. We needed to disrupt the market.

To make this happen, we knew our advertising needed to achieve three interconnected objectives.

1) Enhance User Sentiment

The creatives and messages needed to delight our users by igniting their curiosity and making them feel special, and acknowledged. The creatives needed to build a relationship between the brand and the user.

2) Improve User Engagement

If we were successful in enhancing user sentiment, we could expect our users to respond to these feelings by engaging with our ads, sharing them with their friends and talking about it on social media. The creatives needed to be something buzz-worthy.

3) Develop Brand Salience

If enough users started talking about the brand, the final goal would be within reach — make swvl the first choice to come to our user’s minds when it comes to intra or inter city commuting. To enable this, the creatives had to be memorable and long lasting.

Our Solution:

Our goal was now to come up with a hyper personalised campaign that achieves all of the above objectives in an efficient and scalable manner. As we brainstormed we realised; what’s more personalised than your name?

We decided we were going to reach out to our existing & inactive user base by tailoring creatives that explicitly called them by their name. We knew doing this with something like email, push, or sms is easy. But pulling this off via Facebook ads, that too at scale, was completely unheard of.

The insight:

Be it brand marketing or performance marketing, the difference between a good campaign & a great campaign is the power & quality of insight. Our insight? Turns out names are not that personal after all.

As we dug through our data, we realised only 120 unique names made up for more than 60% of our total user base. This made our problem tenfold simpler. Instead of hyper personalising the communication for 1000+ names, we only needed to get these 120 right to make the most bang for our buck.

Basic components:

The three basic components in making this campaign work were Audiences, Creatives, and Media Strategy. All we had to do was set up a custom audience for every unique name and ensure that the ad that each of those custom audiences see has their name on it. Then we simply needed to ensure that bigger audiences get bigger budgets automatically. But the final and most critical piece of the puzzle to pull this off at scale was seamless automation.

Using Facebook’s Marketing API & python scripts, we were able to ship 100+ name based audiences directly from our database to Facebook Audience Manager, in a matter of a few minutes. Within a few hours, Facebook was successfully matching most of those users.

Generating creatives dynamically using automated data feeds

With the audiences created on Facebook, we then built a custom feed and connected it to, Smartly.io to dynamically create creatives that matched the audience’s names. This helped save countless hours that would have otherwise been spent in making individual creatives for each name.

Campaign structure to leverage predictive budget allocation and dynamic ads

Now that the audiences and creatives were in place, we designed a simple and scalable campaign structure in order to leverage another powerful feature of Smartly.io called PBA — Predictive Budget Allocation. PBA utilises machine learning to increase and decrease ad set level budgets based on their performance. With this budget optimisation algorithm in place, we didn’t even have to manually optimise the 100+ ad sets. The campaign was now completely automated.

The Result:

The first cohort we reached out to was our inactive and churned base — Users who had used our platform but had not been active in the last thirty days. On the performance front, the campaign clearly indicated three things: the audience was loving it, sharing it forward, and most importantly, coming back to the app to transact. In summary, it was a massive success.

Time to go full throttle:

Once we had the assurance that our users were appreciative of the fact that Swvl had gone out of its way to address them personally, we knew we couldn’t just stop here. It was time to take the personalisation efforts a notch further.

Ramadan was just about to start and we wanted to wish our user base Ramadan Kareem in a way they had never been wished before. This time, instead of using static creatives, we worked with a voice over artist to record a script and 100+ names individually. Using Smartly.io’s creative solutions and data feeds, we were able to stitch the audios together, and generate dynamic video assets to match each name within seconds.

The results of our Ramadan campaign were even better than the previous campaign. Our user base loved it and still couldn’t fathom how Swvl was doing this and at scale.

User Sentiment & Engagement

On the brand metrics front, we were able to unlock: 157 times higher engagement rate, 118 times more share-ability, 28 times more positive comments, and a whopping 60 times increase in positive sentiment, bringing down the CPE (cost per engagement) down by 86%

On the business metrics front, some of the best performing ad-sets had a click through rate of 8.5%, something we hadn’t seen before. On an overall level, the campaign was successful in driving 444% more searches, 66% more revenue, 56% more bookings, 60% higher re-installs at 43% lower costs.

Closing note:

This campaign was definitely one of the most interesting & successful ones we have ever run at Swvl, but with the new iOS 14 privacy changes and Google killing the cookie sometime soon, digital marketing is only going to get trickier.

We welcome these new privacy changes that are kicking in for better & stronger privacy of user data and at Swvl we are in full compliance & support of them. What these changes mean for us is that we will need to get more creative and build even stronger capabilities to ensure our users find our communication meaningful without being intrusive.

Personalisation is an incredible tool, yet so many of us T shaped marketers tend to take it lightly. At Swvl, we strongly believe in being 100% customer obsessed. Personalised messaging, if deployed in a meaningful way, backed with insights, can truly delight your audience and earn you a special spot in their hearts.

#1BillionBookings

Authors: Mukund Kabra & Mehr Javed
Campaign Execution: Mohamed Ayman Fathy, Rajan, Sinan & Reem Ragab
Creatives: Hazem Samra, Ahmed El Raggal, Mostafa & Omar

--

--