Hanoi – Mobile apps and games are spreading all over the world, and Vietnam is the No. 1 place in the world when it comes to retaining mobile app users, according to the new Mobile Growth Map from mobile measurement firm Adjust.

Adjust — which focuses on mobile measurement, fraud prevention, and cybersecurity — hopes mobile app companies will use the report to improve retention and identify their highest-value users.

This report draws on data from nearly 3,500 apps released in 2018 and covers growth, retention, and other key metrics across 31 countries and four industry verticals: ecommerce, entertainment, gaming, and utilities.

Fastest-growing app markets by region and industry

The Mobile Growth Map uses the Growth Score, a new metric developed by Adjust to chart the rise of apps in global markets. This metric is calculated by dividing the total app installs per month by the number of monthly active users (MAU) for each vertical and country to reveal the rate of growth from installs relative to the MAU base.

Asia Pacific leads the way with robust growth and is primed to rise. Vietnam, Thailand, and Myanmar are three of the fastest-growing nations. Latin America comes in second place, with Brazil and Colombia rounding out the top five.

Demand for gaming apps and ecommerce apps is strongest in Latin America. In fact, four of the top five fastest-growing countries for gaming apps are located in Latin America. Overall, games dominate the number of installs (33%), time users spend in apps (10%), and the amount of ad spend (74%).

Latin America also dominates the demand for ecommerce apps, with Mexico, Chile, and Colombia enjoying the highest growth in this industry.

Entertainment traction

Entertainment apps are quickly gaining traction. Vietnam, Russia, and Thailand take the top three spots on the Growth Score. This growth is likely fueled by the demand for video streaming services, which are expected to continue gaining steam as industry giants such as Disney jockey for audience eyeballs.

Indonesia is a powerhouse market fueled by the popularity of video apps and streaming services. This dovetails with the findings in the Adjust Global App Trends 2019 report released in May, which names Indonesia the “fastest-growing market.” Along with entertainment and gaming, utilities is a fast-growing vertical in this country. Notably, the performance of utilities is driven by the active use of weather apps.

The “retention factor”

In addition to the Growth Score, Adjust developed its own metric to measure the impact of retention, called the “retention factor.” The retention factor is calculated by dividing organic retention by paid retention, providing readers with the real divide between the two types of users.

With the highest retention of any vertical, gaming averages 34% on the first day an app is downloaded (Day 1), and 15% on Day 7. However, games drop 19% of their total user base between Day 1 and Day 7 — the steepest decline of any vertical.

While this drop appears dramatic, it may also be linked to the impact of hyper-casual games. This high-flying subcategory accounts for a significant share of downloads but has so far failed to drive lasting loyalty among players. Interestingly, North American gamers show the highest Day 1 retention of all countries surveyed.

“Growing your app user base is a critical part of the growth equation, but in a market where most apps are history just 24 hours after the install, marketers need to focus more on engaging and retaining those users,” said Paul H. Müller, chief technology officer at Adjust, in a statement. “To boost engagement and extend the lifespan of the app, marketers must build data-driven capabilities to target users looking to churn and target them at critical points long before retention rates begin their inevitable decline.”

I’m going to moderate a gaming user acquisition session at Adjust’s Mobile Spree event in San Francisco on Thursday.

Machines struggle to make sense of scenes and language without detailed accompanying annotations. Unfortunately, labeling is generally time-consuming and expensive, and even the best labels convey an understanding only of scenes and not of language.

In an attempt to remedy the problem, Microsoft researchers conceived of an AI system that trains on image-text pairs in a fashion mimicking the way humans improve their understanding of the world. They say that their single-model encoder-decoder Vision-Language Pre-training (VLP) model, which can both generate image descriptions and answer natural language questions about scenes, lays the groundwork for future frameworks that could reach human parity.

A model pretrained using three million image-text pairs is available on GitHub in open source.

“Making sense of the world around us is a skill we as human beings begin to learn from an early age … The more we interact with our physical environments … the better we become at understanding and using language to explain the items that exist and the things that are happening in our surroundings,” wrote Microsoft senior researcher Hamid Palangi in a blog post. “For machines, on the other hand, scene understanding and language understanding are quite challenging to hone, especially with only weak supervision, essentially the indirect learning people are able to leverage so well.”

As Palangi and colleagues explain, image captioning and visual question answering quality algorithms usually underperform for three reasons: (1) They can’t leverage context to describe images and perform reasoning about them; (2) they’re not tapping large-scale training data for pre-training; and (3) their architecture isn’t designed to perform well on language, vision alignment, and language generation tasks. The team sought to overcome those with an architecture comprising an encoder (which learns numerical representations of data it’s given) and a decoder (which converts the encoder’s representations into human-interpretable information) pre-trained together and optimized for two kinds of predictions. They say that it created better-aligned encoder and decoder representations in the end, allowing them the use the same model for objectives as different as image captioning and visual question answering.

The researchers evaluated VLP’s ability to caption and reason over images on publicly available benchmarks, including COCO, Flickr30K, and VQA 2.0. They report that it not only outperformed state-of-the-art models on several image captioning and visual question answering metrics, but that it managed to answer questions about images (like those having to do with similarity in clothing design) with which previous models trained only on language struggled.

“With smart model design and smart data selection, we can capitalize on existing publicly available resources to reach even greater heights in language and scene understanding, as evidenced by VLP,” wrote Palangi. “With VLP, we believe we show the potential of unified models to reach the levels of language and scene understanding necessary to successfully complete a variety of distinct downstream tasks — single models that complete multiple tasks efficiently without sacrificing performance. That means more effective and capable vision-language systems without the costs of several separately trained models to achieve the same goals.”

The team leaves to future work strengthening the model’s architecture while adding more data during pretraining. Source: Venturebeat