Navigating Chaos with Creativity: A Behind-the-Scenes Look at one of Botim’s Latest Growth Marketing Tactics

Introduction

On a bustling Wednesday at SWVL two and a half years ago, a sudden plunge in user sign-ups due to an OTP delivery outage sent shockwaves through our team. Traditional responses would have taken days to mobilize, but Mukund and myself devised a nearly real-time monitoring script in just a few hours. This script didn’t just alert us about SMS failures; it automated user re-engagement through strategic CRM tactics and API integrations, efficiently minimizing the downturn in sign-ups.

This day crystallized my belief in the power of a versatile, agile growth team capable of transcending conventional digital advertising and analytics to deliver sophisticated solutions with unprecedented speed and precision.

Facing the Challenge Head-On

Botim, as a hub for digital transactions facilitating various services such as international & local money transfers, bill payments, cards, recharges & more faced a need for innovation in its routine end-of-month campaigns. Consequently, the Botim Bingo Deals was conceived not just as a promotional effort but as a strategic test of cross-selling and up-selling capabilities without overburdening our product and engineering teams. This campaign was brought to life through modular development and backend automations built end-to-end by the Growth team, without utilizing or consuming any engineering & product bandwidth, which could have otherwise led to disruption in their sprints.

The 1024 Approach

Imagine orchestrating a campaign where each user’s action contributes to a complex mosaic of rewards. To put it in Layman’s terms, the Bingo Deals Campaign required users to complete tasks across eight services, each contributing to a point tally. Points led to rewards, with thresholds at 50 for a 1-month VIP (Botim’s premium subscription for Ad-Free & HD calling capabilities), 70 for eligibility for various prizes, and 75 for a 1-year VIP along with a chance to win an iPhone 15.

Sounds doable & simple? Yes? Well, not really! Imagine building all of this with virtually no involvement from the product/engineering teams, but it was not a no-code deployment either, the entire thing was coded, automated, and deployed by the growth & marketing function without requiring a version release! The key? A riveting blend of HTML5, JavaScript, Python, and SQL

The campaign involved 8 binary service tasks and 2 binary bonus tasks, represented as a series of binary strings (XXXXXXXX_XX). This structure led to precisely 2¹⁰ = 1024 unique combinations of possible user interactions. Rather than manually crafting 1024 distinct graphics — a monumental task — we turned to automation. Abdallah Al Ashqar, our creative maestro, employed Figma coupled with Google Sheets and a few plugins to automate the iterative generation of personalized creatives. This not only saved countless hours but also ensured each interaction was uniquely tailored to the user’s journey.

Script-Driven Creative Generation:

The process involved detailed scripting to interface Google Sheets with Figma’s API dynamically updating visuals based on user interactions & achievements. Each piece was meticulously designed to reflect every individual user’s actions, making every engagement feel personal and recognized.

If you have a knack for details, read further else you can skip to the next section.

Foundation in Figma: We began with crafting a hero creative in Figma, setting the stage for all subsequent personalized visuals. Each element was carefully designed as part of a scalable and robust template to support over 1000 unique creatives.

Layered Design for Personalization: After finalizing the hero creative, we methodically set up individual layers for each component that required subsequent customization. These layers followed a sophisticated naming convention to ensure bulletproof automation & reduce the chances of any kind of failure to virtually 0%.

Google Sheets Scripting: Using Google Sheets, we scripted dynamic data feeds to manage the 20+ layers for each of the 1000+ creatives. This automation allowed for real-time updates and personalization of each design element within Figma.

Data Integration with Figma: A specialized plugin facilitated the seamless flow of data from Google Sheets to Figma, bridging creative design with structured data. This integration enabled the dynamic customization of creatives based on the user’s latest app activity.

Layer Manipulation for Customized Creatives: In Figma, data-controlled layer visibility tailored each creative to reflect a user’s actions. For instance, completing a group call would activate specific layers, visually representing user progress in a fun, novel, and engaging way.

Efficient Creative Selection: A robust naming convention helped our Growth team accurately map the appropriate creative to each user cohort, ensuring a personalized experience that reflected their unique journey and progress in the campaign. Now that the creatives were sorted, we had to figure out how to introduce the campaign to the users so they could understand how it works.

A landing page to introduce & simplify the campaign to the users

A Landing page with gifs and 8 different CTAs sounds like a task that’d require us to wake up our engineers in the middle of the night. However, Ashqar again led this front and designed the Landing Page using HTML, allowing us to easily host it as an H5 miniprogram, which we could easily push to the users without requiring them to update the app. This was then finally made available for our users on the explore page for discovery & deep-linked to all our user communications to drive activation & engagement.

Dynamic Content Optimization: A Game Changer

Our strategy went beyond simple automation; we implemented a sophisticated content optimization algorithm that dynamically adjusted the delivery of the content to go with the customized creative based on real-time user engagement. Utilizing machine learning techniques, this algorithm assessed daily performance data to optimize content distribution, enhancing user interaction and participation in the campaign.

Advanced Scripting and Automation: We integrated a CI/CD pipeline with Apache Airflow for real-time data updates from the H5 Miniprogram, ensuring accurate campaign analytics. Using Python scripts, we dynamically assigned creatives to user cohorts, enhancing content relevance and precision. Campaigns were executed via REST APIs, allowing agile management and personalized user experiences.

Data Science & Analytics: We leveraged clustering algorithms like Affinity Propagation for sophisticated user segmentation and XG-Boost for churn prediction & content delivery optimization.

User Behavior and Engagement

Our content & distribution strategy employed usability heuristics for seamless navigation, aiming to immerse users in a “Flow State” for peak engagement, guided by principles that ensure a fluid user experience across all touchpoints. Flow state is a psychological concept introduced by Mihaly Csikszentmihalyi, which describes a state of optimal experience and deep engagement. The Bonus strategy deployed was primarily devised to artificially invoke the flow state in a user’s mind, especially the ones that had not engaged initially, and after each step they took, they were yet again just a step away from the next milestone.

Every participant received customized content that evolved daily, reflecting their unique journey and motivating further engagement with the diverse services offered by Botim. And this was further achieved by unlocking & optimizing entry points from push notifications to in-app messages depending on the users’ affinity, ensuring effective user funnel management and a cohesive experience.

Results and Impact

The Bingo Deals Campaign led to double and triple-digit growth in service usage and a substantial shift in the baseline activity levels for almost all services two weeks after the campaign had ended.

To name a few:

21 new all-time high records broken across various metrics like new users, users, orders, conversion, and volume across various fintech services

Boosted referrals by 5x without having to tweak the referral award

72% increased users cross-selled & up-selled into at least 1 service M.o.M

The bonus structure drove an increased 10% engagement in services compared to no bonus control group

46% of users exposed to any of the entry points participated in the campaign

As high as a 70% uptick in transacting users depending on the service

Bingo Deals as a campaign not only had an enduring impact on millions of our users but also led to disproportional gains on a lot of our services. Hence, we would love to know if you also ever harnessed the raw power of advanced scripting, data analytics, and machine learning to overcome tough operational hurdles and transform what is essentially a challenge into a triumph?

Join us in this discussion of how business growth can be achieved with intelligent & interesting user engagement-led growth loops.

Special Thanks to Obada Issa & Loay Taha who worked on this tirelessly!

Authors: Sakher Al-Adaileh, Mukund Kabra
Creative Lead: Abdallah Ashqar
Design & Content: Shivangi Premkumar, Simran Karnani, ElFatih Shami, Riza Zaidi, Anthony Romero, Ahmed Mourad
Growth: Obada Issa, Loay Taha, Umar Habib, Arjun Dangi