The use of artificial intelligence in trading has a long and storied history that predates the recent breakthrough of LLMs into mainstream awareness and widespread use. From institutional HFT algorithms and the adoption of trading bots in retail, to early sentiment analysis systems and earnings call analyzers. In 2023, on the eve of the recent AI boom, EY reported that 99% of the CEOs it had surveyed were making or planning significant investments in AI. As far as retail brokerage is concerned, these investments appear to have initially focused primarily on client-facing tools that include market summaries, trade suggestions, portfolio construction and analysis, stock screening, technical insights, and strategy development. This approach has served to aid branding initiatives and to boost conversions at a time when AI was at the forefront of public consciousness. Below we make an argument for why this is changing towards churn prediction and revenue retention, focusing more on harnessing proprietary data that remains underutilized by brokers.Prioritizing competitivenessIn our conversations with clients, we’ve found that as the market becomes saturated and AI fatigue sets in, financial services businesses are increasingly turning their attention to how AI can assist in the retention of revenues and the prevention of client churn. Dormant accounts, high client turnover, and increasing customer acquisition costs repeatedly come up as on-going concerns for brokers. These concerns are being exacerbated by increased competition from a myriad of rival offerings, among them neobanks and neo brokers that are expanding into multiple asset classes, instruments, and regulatory jurisdictions. Online brokers have mastered how to convert leads, but to remain competitive in today’s increasingly crowded market, they’re now called to become experts in revenue retention, and AI can be a valuable ally in this area. Our own approach to the development of AI tools has been two-pronged in that we’ve focused on the development of both front- and back-end technologies that serve the goals of personalization, engagement, churn prediction and revenue retention, each picking up where the other leaves off.Data quality has been an on-going conversation in AI-circles. Our own position has been that brokers have access to high quality client data of their own that is unique to each business and can provide competitive advantages when harnessed correctly. Two types of proprietary brokerage dataOn the one hand, brokers have access to user inputs, which include queries and preferences expressed through searches and engagement touchpoints. On the other hand, they have access to trading behavior, which is available via platform telemetry and account management behavior. The first can be gathered from client-facing interfaces such as AI-assistants and can be used to provide insights that relate to client interests via queries, conversations, and search box interactions. The second is available on the back end by gathering trading data that says a great deal about client trading habits, risk profiles, as well as their money management practices. Each is beneficial to brokers seeking to maximize their revenue-retention initiatives, which include offering a more personalized service, increasing engagement, providing pre-emptive support, and in general having a more holistic view of each client.Client-facing AITo start with, client-facing AI can be used to help users engage with a broker’s offering in a more constructive manner. Recommendations and search bar suggestions allow traders to broaden their horizons while remaining on the platform, helping them to find both what they’re searching for and to be exposed to things they may not yet be aware of. These systems also act as a first line for customer inquiries, answering queries and solving problems quickly and efficiently, lightening the load of human teams and allowing them to tackle high priority issues. Our own version of such a system ships with the DXtrade platform and allows customer support staff to observe these interactions and to seamlessly step in as a second line of support, when required, with in-platform chat, video calls, and screen sharing.It’s important to note that this process generates its own valuable data. Client interactions such as those described above can form the basis for additional client insights, which can be used as the basis for timely alerts, asset suggestions, and other educational content to keep clients engaged and developing as traders. DXtrade’s trading assistant is also integrated with Discord, Telegram, Messenger, and WhatsApp. These integrations allow users to remain “plugged-in” to the trading terminal and to send instructions even when not in direct contact with it. In this way, it borrows from the stickiness of their preferred communication apps, without having to compete with them for attention. Another thing to note is that users benefit from dxFeed market feeds, meaning that the data used to answer market-related queries is of a higher quality.Broker-facing AIOn the back end, effective user profiling can be conducted, and actionable behavioral insights can be distilled at scale. Utilizing machine learning models trained on real client trading data, it's possible for brokers to gain a view of what’s going on with their clients in a way that wasn’t feasible in the past. DXtrade’s user profiling module provides client churn predictions derived from proven engagement metrics, which indicate the users that are statistically most likely to drop off. This information can be directly integrated with company CRMs and is crucial to retention initiatives as it can guide the generation of timely responses that are highly targeted.The system’s extensive segmentation capabilities and its ability to discern between different platform actions that may be of interest to brokerage teams (such as repeated deposit failures, or trading difficulties such as setting stop losses) open up a variety of communication opportunities. This is the key to incorporating systems such as these into the ways different brokerage teams organize their workflows, whether it be via in-platform communications, email, or personalized outreach in advance of accounts falling dormant. Common groundIn the past, overreliance on a constant stream of new conversions and an acceptance of client churn figures as a matter of fact have contributed to a dragnet approach to customer acquisition. This and the reality of fragmented software ecosystems have been obstacles to brokers making the most of the data they generate in an efficient and productive way. As trading evolves from a niche occupation to a lifestyle, brokers are being called to truly know their customers, and to start viewing them as long-term partners. The point here is that churn statistics are not set in stone. Much of the data pertaining to client turnover originate from a time before it was possible to truly provide personalization at scale.Through a combination of learning from what they say and what they’re interested in, as well as what their actions when trading reveal about them, brokers are now able to bridge the engagement gap between acquisition and retention in order to encourage longer, mutually beneficial relationships with their traders. This article was written by FM Contributors at www.financemagnates.com.

Read full article on FINANCEMAGNETS