US outfit Playnomics sounds like the sort of place I'd like to work.

Formally, it describes itself as a "predictive data and personalized marketing platform".

The fun bit, however, is that is comes up with crazy products such as its Churn Predictor, which will warn you when players are about to leave your game.

Even better is its newly announced Acquisition Value Predictor.

I caught up with CEO Chethan Ramachandran to ask some hard questions.

Pocket Gamer: Acquisition Value Predictor (AVP) sounds very scientific. What does it actually mean?

Chethan Ramachandran: The problem we see in the free-to-play industry is that in order to have a sustainable game, developers and publishers have to continuously acquire new users profitably - the future Lifetime Value (LTV) of each user has to be greater than the cost.

Currently, marketers spend money upfront to acquire new users and hope that the user will spend enough money to make that cost back, which can be up to 90 days later.

So the cost side happens upfront, and the LTV remains unknown. In an ideal world, developers and publishers would be able to acquire new users and get an immediate forecast of the lifetime value of that user - not 30-90 days later.

Instead, our AVP forecasts that value immediately (in 1-3 days). The intent is to make user acquisition easier, more predictable, and ultimately more profitable for everyone from independent developers to large companies.

With AVP, instead of waiting up to 90 days to understand the payback period for an acquisition campaign, acquisition managers get a sense in a few days. They can then quickly adjust where they spend money (ad networks, agencies, referral partners, etc) and become really efficient with their spend.

With the acquisition costs being a continuous burden for developers, we think this can really help developers get the maximum bang for their acquisition buck.

You say it can "predict new customer lifetime value with 75% accuracy in just days", but doesn't that assume that the biggest spenders start spend within "just days", or are looking at other metrics too?

Depending on the game, early spend is not necessarily an indicator of becoming the biggest spender. Other indicators can be loyalty levels (regularity of play), intensity of play within each session, or attention levels (length of session time). It depends on the specific game and its specific mechanics.

For each of the hundreds of games integrated with our platform, we capture detailed, in-app user specific events for every unique player (over 300 million player profiles to date). We then apply predictive algorithms that look at the potential for a player to convert to be a spender and the potential volume of purchases over the course of that player's lifetime.

We weight those behaviors based on a scale of importance/priority specific to the game (or genre), which are then applied within the algorithm for each specific game to get a specific predicted LTV for each player.

75% isn't a brilliant level of accuracy. Are there any technical obstacles stopping you getting this to over 85 percent or is it more about the balance between time and accuracy?

Stepping back, the current state of affairs for a developer is having very little to no visibility of the overall lifetime value of a player until 2-3 months have passed. They've spent money upfront hoping to make it back at some point. Given that, we think even just a little insight can be immensely helpful. AVP can give marketers a lot of directional comfort that they're spending money on the right ad networks and channels.

A lot of ad networks have been in the business of unloading sub-par install traffic onto developers, and we want to help the developers get past that. If developers can be more accurate about where they are spending money to acquire users, they then have a better chance to have a profitable game.

In terms of accuracy, in prediction circles a 75% level of accuracy is considered solid - and this is just the beginning of where we're going to take the product. The models will automatically get more and more contextual about specific game types and mechanics and the accuracy will improve as well.

To do this internally, a developer would need a team of dedicated data scientists building models for each channel, every single day. AVP was built to automate this process, and give a marketer the information to make quick and smart business decisions regarding the value of acquisition campaigns and predicted spend - without needing to build out significant internal resources or cost commitment from the developer.

Once a developer has found a "high value player", how do you advise them to encourage them to keep playing?

Every player is different and therefore needs to be treated differently based on their behavior and potential. This is where our core PlayRM product comes into play, which was designed to nurture and retain players.

High value players are a specific group that needs to be monitored closely and catered to with specific content and messaging - new levels, special VIP items, etc. The system can even detect when these players are "at risk" of quitting and send them an automatic message to win them back before it's too late.

To further help developers, we've used the data generated from hundreds of monetization and retention campaigns to design genre specific marketing playbooks that have specific promotion / content / messaging strategies for all key player segments at every step of the lifecycle.

Do you worry this sort of thing could eventually change player behaviour so they start to play like a "high value player" in order to get better treatment?

If players start acting like high value players to receive better marketing promotions and rewards, it's a win for all.

If the concern is players that are trying to 'game' the marketing engine for better offers, we've taken measures within our algorithms and messaging platform to measure the true value a player is worth and reward them accordingly.

Does using AVP metric provide a developer with any useful feedback during the development or soft launch process?

Absolutely. Our AVP tool predicts the lifetime value of newly acquired players, so developers can determine how much money they'll likely earn back from players.

With this information, developers can get early feedback on how well their game is going to monetize without having to wait through the time and expense of a larger scale trial.

By understanding which types of players are likely to have the highest monetization, developers have a powerful tool for game tuning during the soft launch process and a huge advantage for acquisition targeting upon full launch.

Does the AVP metric have any use in terms of helping developers work out if their game is going to monetize well in comparison to other games, or it is merely about identifying the best users?

Our AVP tool is primarily for developers to optimize their acquisition spend for their game(s).

That said, we also publish reports on monetization and engagement metrics across the hundreds of games on our platform, and developers can review this report to see how their game is monetizing compared to the industry at large.

You can find out more about Playnomics' services via its website.