While artificial intelligence has been used across various industries for years, more recent developments and advancements in AI have propelled its use cases. More companies are looking toward AI tools to help with daily tasks and keeping costs down. This is no different in the gaming industry, with studios using generative AI for creative purposes, marketing and even art.

However, when new technologies emerge, knowing when to climb aboard and how you would even get started can be challenging. In this guest post, LifeStreet’s CEO Levi Matkins shares how generative AI can be used to gain greater success.

Three ways to create value for your business using generative AI

Ad tech providers are constantly competing to deliver faster results, more significant cost savings, and an improved user experience for their clients. One way ad tech players deliver these results is by adopting emerging technologies. But with the constant rollout and hype about emerging technologies - whether it be web3, the metaverse, and now, generative AI - how can a business cut through the noise to determine if, when, and how an emerging product can be adopted to add the greatest value to their business?

At LifeStreet, we’ve started assessing and piloting generative AI across our data, product, and creative teams.

Based on our experience, we’ve come to believe any company that is part of the digital advertising ecosystem can start benefiting from generative AI tools today - especially those in the ad tech space. Given the personalisation potential of mobile ads and the volume of data managed on the day-to-day, ad tech players are particularly well-positioned to benefit from generative AI.

Here are three ways you can identify areas where generative AI can add the most value to your business, including a few examples of how we at LifeStreet have started using this technology to create value for our partners.

Identify time consuming tasks

Consider all the ongoing tasks your company performs and identify the ones that are too time consuming to do at scale. Iterate ways to automate these tasks using generative AI. For example, you could automate certain types of content generation, like creating thousands of customised headlines, summaries, or other types of copy. Scaling these tasks will allow you to streamline time-consuming workflows for less time and money.

Foster experimentation

Encourage every team member to use generative AI and discuss how it can enhance the performance of their tasks. Like many companies, we’ve found generative AI systems like Chat GPT4 very useful for generating new ideas, whether it be icebreaker questions for a team meeting, an engaging summary of an article, or a fresh approach for tackling an existing problem.

We’ve found that using GPT4 as a replacement for Google Search provides easy information without distracting sponsored search results and ads. Likewise, coding assistant tools like GitHub Copilot can materially improve developer productivity. Put generative AI in the hands of your team members and promote knowledge sharing. This will bubble up use cases to the top level of how generative AI can be maximised to scale processes.

Keep tabs on new creative tools

Generating content - such as images, video, and audio - has been one of the most immediate benefits of generative AI. As this technology becomes more powerful, the cost of asset creation will continue to shrink, becoming free or close to free. Very soon, it will take fewer people in an art department of a company to streamline content generation at scale. This, in turn, will power highly personalised products and campaigns in which every component of a customer’s journey is customised to engage them based on their unique experience.

With that said, keep tabs on new generative AI systems that can help your team cut down on costs and streamline content creation. This might be using generative AI to create stunning visuals for your mobile game or natural sounding voice overs for your next promotional video.

How we’re creating value with generative AI at LifeStreet

Our data science, product and creative teams have incorporated generative AI tools into our day-to-day workflows to augment data and optimise mobile advertisers’ growth campaigns. Here’s what we’ve learned along the way.

Data augmentation

With the limitless supply of information available on LLM tools like OpenAI’s GPT, performance marketers can use these tools to compensate for gaps created by data privacy restrictions like Apple’s ATT Framework. Mobile advertisers can radically enrich their data via simple text prompts to build high-performing ad targeting packages and bidding strategies. Because there is no limit to the tags ChatGPT can generate compared to a predefined / manual category list, the possibilities are endless. At LifeStreet, we’ve used OpenAI’s ChatGPT to automatically generate descriptive tags for data augmentation.

These tags are extra inputs that we feed into our models, which lets the models generalise across what would otherwise be sparse signals (i.e. it can learn signals with a given tag convert better or worse for a certain set of impressions in a particular campaign, etc). In this way, we can teach our models that certain users have higher (or lower) install rates for signals with specific tags. This, in turn, fuels better campaign targeting and improves our bidding predictions which helps our models broaden inventory sources beyond the current subset of well performing apps.

We’ve successfully used AI-powered-tagging to customise messaging and target users more effectively. In fact, LLM-powered data augmentation has led to a 9.4% decrease in cost-per-install (CPI) on non-rewarded traffic and a 6.3% lower CPI on rewarded traffic.

Dynamic ads

While AI-powered dynamic text is currently the most common use case for generating ad content, creative models will eventually be able to choose the top-performing AI-generated combination. For any given impression in any given app, models will be able to predict from an unlimited selection of customised ad elements (e.g., format, size, headline, background, CTA position, etc.) which combination is most likely to drive a conversion for that impression.

At LifeStreet, we've observed that when an ad’s text and imagery is customised to reflect elements of the environment in which it is shown, there can be a significant lift in performance. We are at the beginning stages of using generative AI to scale content ideation for personalised ads.

To get started creating dynamic ads using Generative AI, identify which elements you want to customise and use OpenAI’s ChatGPT to generate content ideas for your pre-defined dynamic content areas and input this data into your creative models. Once you have your results, use machine learning to choose the most engaging element or combination of elements to serve a given impression.

Efficient reporting

Generative AI has the capacity to unlock new efficiencies in campaign changelog, performance and spend distribution report summaries. With generative AI, mobile game developers can enhance campaign visibility and access a greater understanding of campaign changes. At LifeStreet, we’re in the preliminary stages of testing the ability to summarise insights with generative AI. This includes weekly campaign changelog summaries that can help us better monitor the impact of optimisations on a weekly basis. We’ve also started testing the ability to generate weekly summaries highlighting notable performance changes or shift in spend distribution.

From Experimentation to Implementation

Generative AI is creating unprecedented efficiencies and optimisation opportunities for all businesses, and especially mobile advertisers. As more companies leverage this technology, generative AI is fast becoming an essential tool in ad tech’s future. To not be left behind, it’s critical that teams identify workflows that can benefit from generative AI and foster experimentation from the bottom up.

Edited by Paige Cook