Deal Sourcing Intelligence

Intro

A multifamily real estate investment fund wanted to evaluate potential acquisitions more quickly and confidently. We built a custom AI-powered analytics platform that combines the fund’s own historical deal data and criteria with up-to-the-minute market data. The platform automatically scores incoming deals and flags the most promising ones, helping the team screen opportunities 10× faster. It provides interactive what-if analysis and ensures investment memos are backed by data-driven insights, so the fund spends time on the deals most likely to boost returns.

Challenge

The acquisitions team was using generic market research tools that didn’t reflect what a “great deal” meant for their specific strategy. They could see broad market stats, but there was no easy way to overlay their firm’s unique experience and success factors onto new opportunities. When a potential apartment acquisition came in, the team would manually gather data: rent rolls, sales comparables, neighborhood demographics, economic trends, etc., then compare all that to the performance of past deals in their portfolio. This manual crunching could take days for each property and often required multiple team members. In a hot market, that meant they risked missing out on good deals because they couldn’t evaluate them fast enough. There was also inconsistency – the evaluation of deals could vary from analyst to analyst, and insights from past successes weren’t systematically applied to new prospects. The fund needed a way to harness what it had learned over the years about what makes a deal successful (or not), and use that knowledge automatically on new deals as they appear, all while continuously incorporating the latest market data.

How It Works Now

We developed an intelligent deal evaluation platform on AWS that fuses the fund’s internal data with rich external data sources. Using Amazon’s machine learning services, we trained models on the fund’s historical acquisitions and their outcomes.

Essentially, the system learned the patterns of deals that achieved or exceeded the fund’s return targets – for example, it picked up on combinations of factors like cap rates in a certain range, strong local job growth, population migration trends, or sponsor track records that often correlated with winning investments for this firm.

We then set up automated data pipelines that constantly pull in fresh market data: current rental rates, recent sale comps, employment and migration statistics for target markets, and even demographic shifts from sources like the U.S. Census. Whenever a new deal comes across the transom, the platform immediately evaluates it against both the live market data and the fund’s learned “success profile.”

The acquisitions team can log into a dashboard and see an instant analysis of the property. The dashboard shows how the deal’s key metrics stack up against the fund’s benchmarks and past successful deals. It also gives the deal a score (or grade) indicating overall fit. The team can drill down and see specifics – for instance, maybe the deal scores high because it’s in a neighborhood with above-average rent growth and the price is below market comps, but the model might flag a concern that the expense ratio is higher than their typical successful deals.

The platform provides these explanations so the team understands why a deal is recommended or flagged.

Analysts can also interact with the system by adjusting assumptions in real time. If they want to see what happens if they underwrite the deal with a 3% annual rent growth instead of 2%, they just tweak that input and the model instantly updates the projected returns and the deal score.

The system also has a notification feature: it will automatically alert the team when a new listing appears that strongly matches the fund’s criteria (for example, if an off-market deal in one of their target submarkets hits all the right notes that their successful deals have in the past).

Conversely, it can send a caution if a deal they’re considering has one or two red flags that don’t align with their historical success factors.

All of this runs on a secure web application backed by AWS, so it’s fast and can scale as the fund expands to more markets or adjusts its strategy.

Results

The investment fund’s team now screens deals in minutes instead of days. The moment they receive a new offering, the platform provides a data-rich evaluation, complete with a score and rationale, without the team having to do the legwork from scratch. High-potential deals that match the firm’s strategy are immediately brought to the forefront, so the team can focus their time and energy on those winners rather than sifting through every listing. This has significantly improved their pipeline efficiency – they’re no longer spending days on deals that ultimately don’t meet their criteria. The consistency of analysis has also improved: every deal is measured against the same yardstick (the firm’s own success metrics and real-time market data), which means investment committee memos and recommendations are more standard and backed by hard data. The team can easily run scenario analyses to answer “what if” questions, making them better prepared when discussing options with the investment committee or investors. They also have clearer visibility into why a deal is considered risky or attractive, thanks to the model’s explanations for each score, which gives them confidence in saying no to deals that are off-profile. In summary, the fund is moving much faster in evaluating opportunities, they’re concentrating on the deals that have the best shot at high returns, and they have a higher degree of confidence that the deals they do pursue will meet their return targets (since those targets and lessons from past deals are baked into the evaluation process).