Rapid Active User Growth For An Elite Dating App
The dating space is immensely competitive and has become somewhat of a zero-sum game. To acquire users, dating apps must win them from other dating apps however this is no easy task when starting up, as new users will only invest time in building their profile if there is already a large network of people to engage with. To stand out dating apps must have a unique concept (e.g. Bumble - where women make the first move), and design their features around a specific target audience, e.g. Raya - focusing on celebrities and high-profile individuals.
Victoria’s proposition combines modern dating features with the exclusivity of a private members club, where new users must apply for membership and go through a formal committee review before being accepted.
Victoria needed Growthcurve to find a winning formula for their paid social media marketing, particularly on Facebook and Instagram. We approached the challenge of growing Victoria’s user base on these social media channels from several angles. First we solved challenges Victoria had been facing around tracking and attribution by defining a data dictionary of post install conversion events, guiding Victoria’s mobile app engineering team on implementation - and ensuring our set up would function well across our Mobile Measurement Partner (Adjust) and the ad platforms (Google, Facebook, Snap, TikTok etc) given the limitations presented by iOS-14+ and the introduction of SK Ad Network campaign reporting. We trained ad-network algorithms on a single high-priority post install event (likely to occur within the first 24 hours), which qualified users as ‘likely to be accepted’, based on our analysis of Victoria’s existing 60 day retained paying users. Setting up campaigns in this way meant we were able to feed campaign algorithms signals on what a quality user looks like in the shortest possible time frame.
Simultaneously Growthcurve’s growth engineering team defined, designed and developed a bespoke web-to-app funnel which allowed us to capture an extended set of campaign meta-data against the users signing up from ad campaigns, linking it back to app users once they are accepted in app. With this enhanced level of visibility Growthcurve’s media buying team could confidently invest more in the campaigns, audiences and creative sets generating the most valuable members for Victoria.
Our creative team had the challenge of producing a framework for Victoria’s ads which would communicate the app’s unique dual concept (high-net-worth networking and dating) whilst also speaking to a broad spectrum of business industries for networking and a broad set of tastes for dating. After working with our media buyers to run a series of concept experiments, we discovered (unsurprisingly) that featuring real people in the ads resulted in significantly higher click through rate and conversion to install. However the creative performance was not consistent across different audience groups. With this knowledge and using profile images of real opted-in members we created a suite of personas and began a series of experiments to establish the optimum faces to show to each audience group. This resulted in a significant improvement to both campaign cost per acquisition (CPA) and overall daily volume enabling Victoria to dramatically increase spend levels.
However we were far from done with the creative framework… We reasoned that we could push performance even further by attempting to personalise creative to each viewer, instead of at the audience-group level. You can’t achieve this level of personalisation using standard cold static or video ads on any of the leading ad platform including Facebook so we had to think outside of the box. Where is this type of scaled individualised ad targeting ad creative personalisation commonplace? E-commerce…
‘People as products’ is an awful term, however in Victoria’s case it was one of the breakout creative hacks which our team pioneered to drastically improve campaign performance.
We created an e-commerce catalog of products for each of the real user personas from our previous set of pilot tests, with a number of product variants (instead of a dress, in red, blue or green we used different photos of the same person) and used this to dynamically populate our ads in real-time. The engines that power e-commerce campaigns use keywords from the catalogue feed and a number of other data points about users (which advertisers can’t necessarily see) to surface the image and text combination most likely to get the product sold. In our case as ‘sale’ was simply a qualified new member application.
Once all of the tracking was optimised in the same way we’d optimise e-commerce campaigns - ensuring we fired events back to the ad-networks when each unique item / variant id was viewed, interacted with (considered) and sold - the algorithms began to learn and campaigns could scale cost effectively by optimising ads to individual viewers. Working in this way also ensured we could personalise right the way through the web-to-app funnel we’d developed as the users job function, age range and more was already known by the time they arrived at the funnel landing page, without them having to do a thing.
And of course.. we knew exactly which person to show in landing page pictures to get them to convert…
Dating apps follow marketplace dynamics and so working closely with the founding team we are able to adjust campaign weighting quickly to grow the share of a particular gender within the Victoria ecosystem. This helped keep the optimum balance of male to female to give the best customer experience to Victoria's users. No one want's to wait days for a reply... Given the app's high net worth focus we also worked with the founding team to understand the job profiles of the most valuable users - using our precision targeting and smart funnel to introduce specific numbers of Lawyers, Bankers, and Entrepreneurs in each new cohort.
