DECISIONS, NOT DASHBOARDS
Understand and unlock the value of your users.
You pay a lot to acquire every user. Ilara makes every dollar of it count, in minutes, per player.
250,000+ players scored in live games — 5,800 more every day.
- 01 Rescue paying players going cold whyspend −60% / 7d $18.4K
- 02 Offer a loot box after a match win why154,018 paths/wk $12.9K
- 03 Show the starter pack to likely payers whytop 1% of spend $9.7K
Real Data. Real Impact.
Measured on live games.
| 9× | improvement on IAP conversion, per offer shown |
|---|---|
| 95% | fewer popups shown. Same volume of sales |
| 38% | fewer players stuck in FTUE |
| 98% | accuracy in predicting 30 day player retention |
| <30 min | to identify player spending behaviour |
The problem with LiveOps now
You pay to acquire every user. You learn what they’re worth weeks later.
Your dashboard tells you what happened after the event. Ilara tells you what is going to happen before it does, so you can do something about it.Your dashboard tells you what happened. Ilara tells you what is about to, while you can still act.
We tell you within a single session.
Ilara Features
Every player evaluated.
Every action measured.
We score every player in their first session, predict who is likely to pay, and attach the action most likely to convert them.Every player scored in their first session, with the action most likely to convert them attached.
-
Operate
A live view of your players. What you are doing and what to do next, player by player.
-
Predict
Churn, payer likelihood, stuck and fraud. Scored before the first session ends.
-
Personalise
Recommendations are tailored per player, not per cohort, so the right offer reaches the right player at the right time.
-
Protect
Machine learning identifies cheating and gives recommendations to reduce in-game fraud.
Your AI co‑pilot for game operations.
Ask in plain English.
It answers with the next move.
No dashboard to decode, no waiting on the data team. You ask, and the answer comes back with a suggestion attached. It answers at your level, whether you are the CEO or the analyst.It answers at your level, whether you are the CEO or the analyst.
- Show me players at high churn risk who spent over $50 last month.
- Working
- search_segments build_offer
- Send the 312 whose sessions halved a return reward tonight, before their next login window closes. They are the only part of the 847 still opening the game, and they are worth about $142 each. Want to see the reward first?
- Spotted While I was in there I found 1,100 more with the same drop who have never paid. The reward would be wasted on them, but a free-currency nudge costs almost nothing and might catch a few.
- Who is most likely to make their first purchase this week?
- Working
- predict_payers rank_offers
- Put the starter pack in front of 2,140 players at their next session start. It does about three times better with this group than anything else in the store. I would keep it away from everyone else so it stays worth showing.
- Spotted You did not ask, but 380 of those 2,140 have already bounced off the pack’s price twice. They want a smaller first rung, not the pack again.
- Where are new players getting stuck?
- Working
- analyze_funnel set_flag
- Offer a skip on tutorial step 7 after the third failed try. About one in five stalls on that tap and quits, and the skip brought roughly six in ten of them back when we tested it. It is a flag, so nothing needs rebuilding.
- Spotted Worth knowing before you ship it: the players who clear step 7 unaided go on to convert at twice the rate of the ones who skip. Keep it to the third failure rather than offering it up front.
- Is any of last week’s revenue fraudulent?
- Working
- score_transactions hold_payouts
- Hold 1,304 payments across 217 accounts before Friday, and let the rest through. That is about 8% of last week’s receipts, and they match refund abuse closely enough that paying them out costs more than reviewing them. I can queue them now.
- Spotted One flag on that: 40 of the 217 accounts have spent over $500 legitimately before this month. I would put those in front of a person rather than let the hold catch them.
- 30+ Tools
- 16 Domains
- 1 Conversation
See it on your own game’s data first. No risk.
No sales pitch. Send us your game’s data and we can show you what we can do. No strings attached.No sales pitch. Send us your data, see what we can do.
No SDK. No new tracking. Nothing to rebuild.
- 01 3 in 4 players never return sooffer a return reward on day 1
- 02 Top 1% bring half the money soshow the starter pack to them
- 03 1 in 5 stall in the tutorial sogive them a skip button
- 04 8% of sales look like fraud soflag those accounts for review
1
easy-to-digest report
The Ilara Difference
What nobody else publishes
-
Others report cohorts on day fourteen.
We score every player before their first session ends.
-
Others hand you another dashboard.
We run your data and show you the money you are already losing.
-
Others rebuild the models for every game.
Our features and labels carry across titles, so a new one never starts from zero.
Ilara Insights
Our blog and case studies.
-
Article
We showed the offer to one player in twenty and sold the same number of packs
A scoring model picked who saw the starter pack. A randomised control measured what that was worth.
-
Article
One in five new players was stuck in the tutorial, and worth exactly nothing
Of the 420 players trapped there, zero ever paid. A skip button rescued one install in thirteen.
-
Article
797 accounts failed the studio’s own fraud rules. One carried a flag in-game.
The rules were already written, and they fire. Nothing downstream acted on them.
-
Case study
Scoring who sees the offer
The full write-up: what the model was shown, how the control was held back, and how the impressions were counted.
-
Case study
The players who never reach your game
The full write-up: the 420 who never paid, what the skip changed, and the figures we left out for being underpowered.
-
Case study
Fraud is already inside your revenue numbers
The full write-up: the patterns found in exports the studio already had, and the two limits on what it proves.
How the first run works
In three simple steps.
No SDK. No new tracking. Nothing to rebuild.
-
1
0
lines of SDK work
Send us your data
Whatever your analytics tool already writes. No SDK, no new tracking, no engineering ticket.
-
2
23
models run against your own players
We analyse it
The full scoring stack, run on your players. We publish what each model gets right, misses included.
-
3
9×
conversion per offer shown, against a control
We send back the insights
What to do next for each player, ranked by what it's worth, in plain English. Plus a call to walk you through it.
Get in touch
One live title and a week of events is enough. Every play comes back priced, and measured against a control.One live title and a week of events is enough.
FAQ
Live mobile titles with real daily users and a live economy - and publishers with catalogues, where one relationship covers many titles.
Today, yes, deliberately. Any app that buys users, converts some and loses the rest daily has this problem. Games is the hardest version, so we’re solving it first.
Analytics tells you what happened to a cohort. Ilara predicts what each user will do next, picks the move, and measures it against a control. Dashboards report. We decide.
Those give operators faster hands to run rules they author. We predict which move to make and publish whether it worked. You can run both. Ours picks the target, theirs sends the message.
No. Scoring runs without one, and every model comes with plain-English reasons attached.
None for the first run. We read the export your analytics tool already writes and run the models on your own players. Going live in-session is a small integration after that, once you have seen what the models found.
Some are. We publish accuracy per model, we ran twelve against a month they’d never seen and four drifted, and every live play carries a holdout so you can see the lift rather than take our word for it. Ask for the test behind any number on this site.