Frequently Asked Questions
1 - Why is askblu.ai focused on player retention?
The AERM principle in mobile games means: Acquisition, Engagement, Retention and Monetization.
It works in this order: you need a good acquisition to have a good engagement (First Time User Experience) ratio, you need a good engagement (FTUE) to have a good retention, and a good retention will lead to better monetization (via In App Purchases, via Ads , for Premium games, and for Subscription models).
Improving FTUE in a game (quality of the on-boarding process in the first session) involves many factors: the quality of the first session, the look of the game, the simplicity, the UX, the art style….
=> Right now askblu.ai is not focused on improving FTUE.
Improving retention is key to increasing the Daily Active Users in a game, and thus the monetization.
This is where askblu.ai is focusing, targeting the “core” retention mechanism: the FLOW (keep the player in the flow).
Remember: it is usually 5X to 25X more expensive to acquire new players than to retain existing ones.
2 - What are askblu.ai main features?
For each game you can activate:
- the Data-Driven Difficulty Analysis feature
- the Real-Time Player Personalization feature
- both features (Data-Driven Difficulty Analysis + Real-Time Player Personalization)
Date-Driven Difficulty Analysis (DDDA)
askblu.ai will ANALYZE the GLOBAL DIFFICULTY TUNING of the Stage (your DEFAULT value) to check if it is optimum for your overall audience of players (on the TOP of the Gaussian view of you players, from “cool” to “competitive”).
The Difficulty Dashboard page in your GAME MENU in the Web portal will give you this information for each stage, so if your default value is not good for a specific stage, you can change the “default tuning” of the stage in the game accordingly (make it “easier” or “harder”) then resubmit your game.
If you use the 2 features, the Player Personalization will work only on stages with a Difficulty Analysis feedback. Thus your game can have some stages ready for Player Personalization, and some other stages still in Difficulty Analysis (waiting for more players).
If there is a large diversity of possible actions in a stage, the DDDA might not be very efficient, so you can turn on only the Player Personalization.
Real-Time Player Personalization (RTPP)
With Player Personalization, before a Blu_Player starts a stage, askblu.ai tells your game the best difficulty to select (Easier, Default, Harder) for this specific player.
(read FAQ §7 below to learn more about Blu_Players)..
askblu.ai RTPP will answer:
- “default” is the player seems to be in his/her “flow zone”
- “easier or harder” is the player seems to be at risk of churn because of furstration / boredom.
Blu/Ref full metrics
In the web portal, you get all the metrics (retention rates, sessions funnel, stages funnel…), all displaying data for BLU players (with personalization) and for REF players (without personalization). Check the DEMO GAME in your account to see all these metrics. Use the tooltips to display details on the reports.
3 - What should I code in the game when I get the easier/default/harder answer (from askblu.ai) at the start of a stage?
Before starting a stage, askblu.ai will return one of these 3 answers for Blu-Players:
- if you already have a specific parameter for the difficulty level of the stage, keep it for “default” value, and 2 more values: one a little harder and one a little easier (you don’t want to make big changes: it is “easier”, not “easy”, and “harder”, not “hard”)
- if you have several parameters for the difficulty level of the stage, you can compute a new parameters based on them, and set this new parameter
IMPORTANT: for one stage, the difficulty behind “default, easier, harder” should be the same for a specific game version and all players, so askblu.ai can learn, check the global tuning, and then personalize.
=> if you were doing some “local personalization” (like “if player did this then set difficulty easer, if player did that the set difficulty harder”), then REPLACE IT with askblu.ai answers ( if “askblu says easier” then set difficulty easier, if “askblu says harder” that set difficulty harder) for the Blu_Players (but you can keep if for the Ref_Players, when GetBluDiff() returns REF).
If you and some testers want to FEEL your gameplay with all stages in EASIER or HARDER, just go to YOUR GAME > SDK Integration > SDK Debug Mode:
With this feature you can “force” all request to answer one value, while the SDK is in DEBUG MODE..
Do not forget to DISABLE this mode before submitting your game.
4 - What should I do if the first stages cannot be lost by the players?
It is quite common in casual & hyper-casual games to have the first stages that cannot be lost (they are used as tutorial for the game mechanic for instance).
In this case:
- put the events stageStarted(StageTag) / StageEnded (WIN), so these stages will appear in the Metrics & Funnels
- do not request askBluDiff() / GetBlueDiff(), because there is no difficulty to tune.
5 - What should I do if I want a specific stage to be difficult (pain point)?
The idea here is setting up the “default” value of a specific stage, called “pain point”, quite “difficult” so people will have to buy (a bonus for instance) to win the stage and continue.
You should keep this pain point, but set an “easier” value that has a real effect.
This is what happens without players personalisation:
- a player “in the flow / well engaged” will not be frustrated by this difficulty, and might try hard or buy a bonus to win the stage and continue BUT
- a player “not in the flow / badly engaged” will be frustrated by this difficulty, and will prefer to CHURN if he is blocked,,, but he won’t buy!
With askblu.ai RTPP, we are telling the game if the player is at risk of churn because of boredom or frustration, so:
- a player “in the flow / well engaged” will not be frustrated by this difficulty, askblu.ai will not answer “ easier” but “default”, the difficulty will not change, so the player might try hard or buy a bonus to win the stage and continue BUT
- a player “not in the flow / badly engaged” and frustrated will receive an “easier” answer, the difficulty will be lower, which can prevent the player for CHURNING if he is blocked. (And will never buy to bypass the stage and continue)
6 - What should I do when I update the game?
There are 2 ways to change the difficulty tuning of the stages in your game:
- an App Store Update (you modify your game, then submit as an update)
- an Online Update (you modify some difficulty tuning data on a server, those data are downloaded by the game)
In both cases, if you use the DDDA, you need to indicate to askblu.ai the stages that have a difficulty level changed (for easier/default/harder) so that the learning will restart only for those stages:
- go to the AI DASHBOARD page of the game,
- for each stage that has been changed, click the corresponding box in the last column (you can tick all).
- for an App Store update, it’s automatic: askblu.ai detects the new update (new version number) with at least 200 players (so the update is live, no internal tests)
- for an Online Update, you just hit the “ONLINE UPDATE” button on the AI DASHBOARD page of the game.
The learning will restart for all the changed stages (daily process).
7 - Can I use askblu.ai for Endless or Idle types of game?
Endless and Idle types of games are important game categories.
askBlu.ai will be compatible with endless games (players never win) and Idle games (players never fail).
Do not hesitate to contact us to speed up things !-)
8 - Will askblu.ai slow my game?
Not at all. The events you send are asynchronous.
The request “askBluDIFF” is also asynchronous for the game: you send the request as soon as possible, then you prepare the stage graphics and data, and you fetch the answer with “getBluAnswer” just when you need it.
If the answer is not ready (no enough time to get the answer, bad connection….), the getBluAnswer method will return your default value, thus there is no waiting time and no risk for your game tuning.
9 - Are all players in a game going through askblu.ai?
For each game, you can setup 2 main parameters for our AI engine (MyGame > game Setup > features):
- Exploration Speed (for “Difficulty Analysis”): “slow” generates less perturbations but takes longer to analyse, “fast” generates more perturbation. Will be set with askblu.ai team depending of your DAU.
- Percentage of daily new players becoming Blu_Players (for “Player Personalization”)
Blu_Players are impacted by askblu.ai: will be set with askblu.ai team (then directly available to you), can be low (20 or 30%) when you want to see the impact of askblu.ai on the metrics and funnels, then you can go up to 100% to get the full potential of askblu.ai for all your player (there is always a little “ref/passive” population for Funnels and Metrics).
The other players (not impacted by askblu.ai) are called the Ref_Players.
10 - Why do we display the Stages Funnel and the Sessions Funnel?
In a game, it’s important to check which percentage of player finish your different stages. You can find “stage drops”: stages where you lose too many players….
With a regular analytics tool, you need to create events, and setup yourself such funnels.
With askblu.ai, you automatically have the funnel of your stages, for both Blu_Players and Ref_players, so you can check the efficiency of askblu.ai.
But having more players in later stages does not mean that players play more, especially if the stages are very easy for the players.
That’s why we display the Sessions funnel of a game: percentage of players reaching session X. That’s the real retention metric.
11 - What player data is askblu.ai using ? Is it RGPD compliant ? Is the SDK using the Apple IDFA?
askblu.ai is only using data about the sessions (start/stop) and the stages (start/stop-> win, lose…).
All this data is fully anonymous: currently with use the device identifiers to keep track of the players in the platform: we have no information about his/her name, age, or any other data coming from over apps of from the OS. Thus askblu.ai is fully RGPD compliant.
We do not use the Vendors IDs or Advertiser IDs of the device, we produce our own Unique User ID.
Here is an article on our BLOG explaining why we do not need to use the Apple IDFA.
12 - What technologies does askblu.ai use?
askblu.ai uses 2 main advanced AI technologies:
- Machine learning: clustering and neural networks algorithms are fed with features (variables based on the events coming from the players). Each game has its own model, updated every day.
- Predictive analysis: askblu.ai learns, by reinforcement, how to predict and correct the churn risk due to the player’s frustration/boredom.
13 - What happens is a player is not connected?
askblu.ai work in real-time, receiving events, and answering the askBluDiff request, so it needs connected players.
If a player loses the connection for small periods of time during a session, the events are stores on the mobile, and sent to askblu.ai as soon as the connection comes back.
-> requests are not answered, but no events are lost, which is good the the Learning process.
If the SDK detects that a player is way too often non connected, this player will be flagged “locked” and will become unknown for askblu.ai.
14 - My game has an iOS version and an Android version, should I create 1 or 2 games in askblu.ai?
Because players, acquisition channels, usage… are not the same on iOS devices and Android devices, you have to create TWO games in askblu.ai, one for iOS and one for Android.
Not only you will get the metrics and funnels for each platform, but most important askblu.ai creates a SPECIFIC Machine Learning Model for each game!
15 - As a publisher, can I manage games from different studios?
In the OPTIONS MENU (top/right), as the OWNER of the account, you get the MANAGE USERS option.
This allows you to invite other people on your account (Managers): a producer to check the metrics, a game designer to check the AI Dashboard, a developper to integrate the SDK….).
As a publisher in charge of optimizing the monetization of published games, you can setup these games in the askblu.ai portal, and then you can invite some peoples from the studio. You can restrict each Manager to only 1 or several games, so someone from one external studio will only see the games from his/her studio.
16 - What do I do if I soft-launch with an A/B test?
If you are doing an A/B test in your game, here is the best solution:
- create 2 games in askblu.ai, “My Game A” and “My Game B”, thus you have 2 keys (key_a and key-b)
- in your game, at launch, if Cohort A then init askblu.ai with key_a, if cohort B then init askblu.ai with key_b
That’s it, then you can check all the metrics and funnels and the Difficulty analysis for each version of the game.
When you chose one version and launch you game, just keep the “live” game and we’ll delete the other one in askblu.ai.
17 - What is the askblu.ai Pricing?
askblu is currently in PAID BETA stage, open to publishers and studios willing to test with 1 or 2 games, and give us feedbacks.
For more information the pricing, visit the dedicated page on the web site.
You can apply for Usage Credits to test askblu.ai at no costs!
In the ACCOUNT menu (top right of the web portal), you have access to:
- “Usage & Costs”: a detailed presentation of your monthly costs, per game.
- “Billing & Credits”: a view of the “Usage Credits” you got, and the list of monthly billings and payments.