How to Attract More Customers with Digital Marketing Optimization
Raketa is the first Ukrainian delivery service in the Foodtech 2.0 format. In March 2018, the service was launched before Glovo, Uber Eats and Bolt Food arrived in Ukraine. However, for a long time, the service was available In the city of Dnepr in Ukraine. The company faced strong competition when launching throughout Ukraine. But due to the lockdown, the demand for delivery increased, which helped Raketa attract new customers and become the market leader. We have prepared a case study based on Raketa's experience, why the right strategies and an optimized advertising campaign bring more results than a huge budget.
Working with Raketa, newage. created and implemented digital strategies for display campaigns to increase:
- relevant audience;
- KPIs of the Raketa mobile application;
- number of active users of the application/visitors to the Raketa website.
- optimization and efficiency of tools to better understand the interests and scenarios of user behavior;
Also the newage. team prevented cost overruns to cover non-targeted Internet audiences and improved personalized marketing. Throughout the campaign, we tracked changes in the engagement of existing and new users.
Our team completed these tasks in three flights in May, June and September-October. We have achieved maximum coverage in the largest cities of the Ukraine — Kiev, Dnepr, Lvov, Kharkov, Vinnitsa and Odessa. To implement the strategy, we used the well-proven See-Think-Do (Care) model, where the level of the conversion funnel depends on our assessment of the user’s purchase intent.
Using this model, we:
- covered the hottest demand;
- warmed up segments with lower purchase intent;
- created a strong brand image in the minds of users;
- analyzed the display media campaigns of competitors.
We used YouTube ads in TrueView, Bumper and Discovery formats to introduce the user to the brand and increase their awareness. We also used banners to increase reach and frequency of exposure per user. We divided the audience into hot, warm and cold. For each of them we applied a specific method, tools and KPI. As we moved along the funnel, our task changed from conversion KPIs "here and now" to media indicators that a person saw our ad, became interested and downloaded the application, or visited the website.
Table showing the actual strategy:
Cold - Google ads / Direct - Special interests of a wide audience - Trueview / Bumper / Instream - 3/m, - CPC/CPV/CPM/ vCPM
Neutral - Google ads / Direct - Special interests of the audience who have previously been in contact with similar service - Trueview / Bumper / Banner - 3/m / 5/d - CPC/CPV/CPM/ vCPM
Warm - Google ads / Direct - Special interested audiences of buyers - Trueview/Bumper / Banner / Discovery - 3/m / 5/d - CPC/CPV/CPM/ vCPM
Hot - Google ads - Remarketing and Discovery - Bumper / Discovery - 3/m / 5/d - CPC/CPV/CPM
In this strategy, we divide our target audience into these segments:
- users with an interest in ecommerce;
- users with an interest in food and restaurants;
- taxi app users;
- look-a-like audience of Raketa users;
- people working on freelance/remote work.
We worked with a user acquisition strategy in which click-through data cannot be applied because a visitor without a Raketa app on the phone is first redirected to the AppStore or Google Play, where the visitor becomes a user. Only after installing the application, the user interacts with ads. The reason we tracked post-view conversions is that the user doesn't immediately order food after installing the app.
We have identified a 7-day window as the optimal timeframe for tracking delayed conversions. Post-view and cross-device data make up the majority of tracked conversions and have been used to optimize ad placements, creatives, and audiences.
Conversions distribution across platforms (session start):
The graphs showed that post-view and cross-device accounted for over 98% of all user conversion data from Raketa app ads. We recommend tracking these metrics because any other data will give a distorted picture of the situation.
Frequencies
The analysis of additional data (post-view + cross-device) made it possible to find the most favorable frequencies for advertising exposure and find out what the cost of delayed conversions is. For example, for YouTube placements in Trueview format, it was important to track the frequency for both individual formats and the entire campaign:
* CPA is a coefficient.
When analyzing the data, we clearly see a sequence of frequencies that bring maximum value to action at optimal cost. This analysis can optimize frequency for each ad format, audience, and device.
Audience
Post-view and cross-device data identified the most effective audience segments, based not only on the number of generated clicks, but also, the actions generated by users directly in the application. So for iOS, the most effective segment turned out to be “users with an interest in ecommerce.” Among IOS, it was the only one that provided useful traffic.
Analysis of the activity and conversion rate among the audience determined the segment of users who are interested in topics related to horeca, among them IOS users have more orders and more often installed the application than Android users.
The segment of people who prefer ordering a taxi to public transport equally divided among themselves the share of clicks and total conversions. It was also found that only IOS users with an interest in ecommerce gave high conversion rates.
We realized that the desktop always performs worse than mobile advertising. Mobile ads have always shown the best results. It sounds obvious, but we often check the obvious because basic things are often more complicated than they seem.
On mobile, we deliberately raised the bids for the test. Even if this made the impression more expensive, it made it more valuable. And that's okay — optimization doesn't mean choosing only cheap solutions. The task of optimization is to give the best result and save only on what doesn't impair efficiency.
Results
Holistic analysis gave the following results:
-
campaigns were optimized 75 times by the number of installs among users who were exposed to ads (from the first week to the last), this required 3 flights, more than 15 stages of optimization and analysis of results (more often than once a week)
- Reduced the cost of the attracted user by 15 times.
The effectiveness of our methods exceeded the client's expectations. At some point, the client paused advertising to improve logistics for the increased number of orders. The customer redirected the rest of the marketing budget to attract couriers. So, as a result of our advertising campaign, not only the number of active clients and users of the application has grown, but also the number of couriers :)