Welcome back to the series of articles by the Space307 Strategy team — Head of Strategy Kirill Shikhanov and Strategy Analyst Artyom Khanganu. In this series, we talk about the Sauron project developed by our team to meet Space307’s competitive intelligence needs. Make sure to check out our first article for insights on the project background, initial development stages and the MVP launch.
This piece will uncover how the Sauron project helps us rank our competitors by various metrics and assess the current competitive landscape.
What is the competitive landscape?
Competitive landscape analysis is the process of examining your market dynamics and assessing your competitors: Market leaders, newcomers and those out of the game. You’ve probably come across infographics rolled out yearly by large agencies, ranking apps with the most downloads.
Our project aims to enhance these infographics so they meet our business needs. We operate in the fintech market, where most products are websites and mobile apps. Knowing the audience and traffic figures of the key players helps us get the measure of the market.
Companies do not typically disclose the exact figures, so we have to resort to approximate values. Let’s dive into how we at Space307 rank our competitors in the mobile app market using the Sauron bot.
Market size: Active users (iOS and Android)
In 2015, App Annie announced their new product — Usage Intelligence, the most extensive database featuring information on how many people use mobile apps and how they do it. That same year, App Annie officially became data.ai. Several companies provide this kind of data, including Sensor Tower, but we prefer data.ai because of its user-friendly interface, API convenience and credibility. Let’s take a look at the Usage Intelligence interface (this is a screenshot taken from the data.ai’s website):
To put together a rating of market actors, we go by the WAU and MAU metrics (number of active users per week and month, respectively). We assess the WAU/MAU of apps available in particular countries, apps issued by the same publisher and a number of specific apps. These metrics show how many users regularly open a particular app.
When gauging weekly and monthly market audiences, we have to work with adjusted figures. When we add up the WAU/MAU numbers of all apps in the sector, we don’t factor in any potential audience overlap (data.ai experts follow the same logic). Calculating the exact number of unique users worldwide is simply not doable. We also add up the iOS and Android user numbers for the same reason.
Market position reporting
We want to know what position we hold in the market and who our biggest competitors are. That’s why we introduced market position reporting, ranking the market players by audience share. For the purposes of the report, we consider each country to be an independent market with unique conditions and maturity levels.
Data.ai provides statistics for each app, even if one player issues several of them. Check out the screenshot below. Binance and Binance.us are presented as two individual players. Our report groups all apps issued by the same publisher and shows them as a single player. We request raw data from data.ai’s API and build a custom ranking.
Members of our team can ask the Sauron bot to generate the market position report to access the ranking. This is what the process looks like:
You can select the required time window in the drop-down menu. The bot then generates the report in the form of infographics:
The infographics show the key players’ figures over the chosen period and how they have changed compared to the previous period.
Look at the screenshot above. The MAU value indicates the market share for November 2023. You can also see how much this number has changed. The line of emojis symbolizes players with more than zero users.
“This sequence doubles the competitor ranking. Players with the largest audience are at the beginning, while brands with fewer users end the line. This is a visually convenient solution as you can get a glimpse into all players and quickly identify the leaders.”
Kirill Shikhanov, Head of Strategy
The report also features market types and concentration values. We divide markets into monopoly, duopoly, oligopoly and competitive markets and assess their borderline states. This classification has certain criteria:
- Monopoly market
The market share of the largest company starts at 50%. The largest player’s share is also five or more times higher than that of the second largest, meaning there are no other major actors.
2. Near-monopoly market
The market share of the largest company starts at 50%. The largest player’s share is also two or more times higher than that of the second biggest, meaning there are no other major actors.
3. Duopoly
The total share of the two largest companies starts at 60%, with each holding a minimum of 25%, meaning both players can be considered market leaders.
4. Near-duopoly market
The total share of the two largest companies starts at 50%, with each holding a minimum of 15%, meaning both players can be considered market leaders.
5. Oligopoly (the market meets one of the three conditions):
- The total share of the three largest companies starts at 50%, with each holding a minimum of 10%.
- The total share of the four largest companies starts at 70%, with each holding a minimum of 10%.
- The total share of the five largest companies starts at 90%, with each holding a minimum of 10%.
5. Competitive market
A competitive market does not fit any of the descriptions above.
We calculate the market concentration using the Herfindahl-Hirschman Index (HHI):
(HHI = Σ((MSn)²)
N equals 5 here as we want to assess the five largest players in the market.
The index value determines the level of concentration:
- The market is highly concentrated if the HHI value equals or exceeds 2,800.
- The market is moderately concentrated if the HHI value equals or exceeds 1,800 but is less than 2,800.
- The market has low concentration if the HHI value is less than 1,800.
The report also provides detailed statistics for top players, including WAU/MAU values, market share and percentage change. If the user wants to see data over a year or a quarter, Sauron adds up the relevant MAU values and calculates the average. The current year or quarter data is compared to the previous periods.
The market position report is the most requested type of report, since it provides a look into the market of a specific region and helps measure the influence of all its players.
Market installs reporting
The second most popular type of report is one that assesses competitors’ user acquisition strategy. The report shows the total number of app installs in one category and installs of apps issued by specific publishers in the same category over the selected period. We assess each player’s share in the installs volume to identify the most active players and market leaders. The information featured in the report helps see the bigger picture. If the number of installs is growing, the market is expanding, and so are the promotion efforts.
The interface and the structure of the market installs report is similar to that of the market position report. However, the market installs report shows the total amount of downloads instead of the number of users:
From a year- or quarter-long perspective, we can compare brands’ budgets or their user acquisition efficiency.
Search interest by Google Trends
Data.ai, SimilarWeb, Sensor Tower and other services use unique algorithms to calculate metrics. Data.ai and Sensor Tower provide more accurate information regarding mobile apps, while SimilarWeb and Semrush are good at assessing web traffic.
Companies regularly launch marketing campaigns or expand into new regions. In such cases, new figures immediately affect user demand and search interest. However, analytical services can’t always register and process the changes right away.
In June 2023, Google accounted for more than 90% of global search traffic, according to SimilarWeb. Google Trends keeps track of all search terms and provides samples of the data acquired. We decided to present Google Trends data as reports to make the Sauron bot more sensitive to our competitors’ moves and provide information on the market changes faster and more accurately.
Scraping data from Google Trends isn’t unachievable. This is where unofficial ready-made libraries come in. The problem, however, is that you can see only five keywords, or data from five brands, within a single query.
Working with the Sauron bot, users can select categories containing more than 50 brands. So, the report needs to feature complex lists with the weighted interest rate. To solve this problem, we developed a custom ETL algorithm in Python using the pytrends library. This algorithm requests weekly statistics for all our keywords, compares them and normalizes the data. The result is the search interest metric, with normalized data falling between 0 and 100. What’s more, the output matches the Google Trends data and score, with 0 for zero interest, 100 for maximum interest and 50 for half the maximum interest level.
“@Akanksha has already described the basics of the algorithm in her article. In our version, we made a few minor tweaks so that the algorithm meets our business objectives. This is an experiment. We are currently assessing its effectiveness.”
Artyom Khanganu, Strategy Analyst
Let’s ask the Sauron bot to send us a weekly report featuring Google Trends data:
The report shows players in the selected category ranked by search interest figures. The number of players in the ranking is defined by the same system that’s used in other reports. You can ask the bot to show any number of players in a category (top 3, top 10, etc.), but the brand’s Google Trends score won’t change.
The search interest report doesn’t show how the numbers have changed compared to the previous period. We want to avoid confusion caused by Google Trends sampling spikes. When fewer users ask the bot to generate this report, the system becomes too sensitive to market changes.
The purpose of the report is to track the dynamics of the top players. We compare search interest metrics with audience and user acquisition data from data.ai. The search interest data and data.ai rankings sometimes match, providing curious insights.
Engineering
To ensure that the bot generates and sends reports correctly and works with multi-user requests, we equipped it with better functionality. The following features now power the competitive landscape reports and the bot operation:
- Navigation
We enhanced the bot with a set of ready-made menus for the chat used by our team to request reports. The set includes dropdown lists, checkboxes, dynamic lists and Slack Block Kit buttons.
2. Computing unit
The system turns user requests into generated SQL to retrieve data from the repository (e.g., Vertica) and use it to form a report.
3. Cash system
We reduced the pressure the app puts on the repository. We also made some space in the system memory for storing generated reports for up to 12 hours.
4. Request control system
- The system tracks random or redundant requests. For example, it can prevent the bot from generating ten identical reports caused by double clicks or other errors.
- This feature also limits parallel computing. Let’s say a user requested a report. If another user later requests the same report, the system will wait until active computations are completed and send identical reports to both users. At the same time, the system can generate reports of different types simultaneously, allowing our team members to assess the current competitive landscape quickly.
5. Logging functions and a simple backend with limited access rights
We can see how the app works and force reset report caches.
This functionality ensures that the bot is efficient and meets our business’s needs for fast and effective competitive analysis.
In the final article of this series, we’ll talk about competitor profiling and share plans for the Sauron project development. Keep an eye out for updates!
