Why BD&L Teams Lose Deals Before the Negotiation Begins

The deals that sting most are the ones where everything looked right. The team was experienced, the asset fit the portfolio, the analysis was thorough. And yet, by the time the work was finished, someone else had already signed the term sheet.

This is not an uncommon story. And it is rarely a talent problem.

The deal was lost upstream, not at the table

Business development teams are paid for their judgment. But judgment cannot begin until the picture is built, and today that takes too long.

The information a deal turns on is public. Trials, filings, deal comparables, the China pipeline. All of it is available. But it is scattered across sources that were never built to connect: trials in one subscription, deal comparables in another, the patent estate in a third, the early China pipeline mostly out of reach. No single source is complete, and the sources frequently contradict each other.

So before any analysis begins, someone spends a day just reconciling them. A single competitive landscape can run for weeks of senior time. And the moment it is done, it starts going stale.

The hours meant for judgment go to gathering instead, not by choice, but because the gathering has to happen first.

That delay loses deals quietly. By the time the analysis is finished, someone else has already signed the term sheet. The deal was never really lost at the negotiating table. It was lost upstream, in the time it took to turn public information into a view worth defending.

Three forces that made slowness fatal

It was not always this costly to be slow. For most of the field's history, information was scarce, and the team that gathered the most knew the most. A landscape that took weeks was acceptable when the landscape barely moved. Three forces broke that arrangement.

First, information stopped being scarce. PubMed now holds more than 40 million citations and adds roughly 1.6 million a year. ClinicalTrials.gov has passed 530,000 registered studies. Once the data is available to everyone, holding it is no longer an advantage. The edge moves downstream, to what a team does with the information everyone already shares.

Second, the pace stopped being linear, and AI is the accelerant. A study of more than 15 million PubMed abstracts found that at least one in seven written in 2024 showed signs of being produced with a large language model, a fingerprint that barely registered before late 2022. The pile to sort through grows heavier every quarter, not lighter.

Third, and most important, machines crossed a threshold. Agentic systems can now follow a defined methodology, resolve entities across sources, produce cited output, and keep it current. The weeks of gathering went from an unavoidable constraint to a choice.

In a market this fast, you don't get outsmarted. You get outrun.

The stakes are not abstract

More than $300 billion in annual revenue loses patent exclusivity between 2025 and 2030. Since 2018, more than 70 percent of new-drug revenue has come from assets sourced outside the company that sells them. The external assets a BD&L team fails to reach in time are the ones most likely to replace what is going off-patent.

The pace is already visible at the top of the market. The 2025 contest for an obesity biotech escalated from roughly $5 billion to about $10 billion in two weeks. When a deal can double in a fortnight, the team that reaches conviction last is the one negotiating from behind, if it is still in the room at all.

What legacy tools fail to fix

It is tempting to read this as a people problem, as if BD&L teams just need to be faster or more disciplined. They don't. The teams are good and the effort is real. The burden is structural rather than personal.

Legacy data tools were built to return data, not to build conviction. They fail BD&L on several fronts at once. The team is buried in more inbound than it can cover. Nothing is resolved to the asset, so the full picture gets stitched together by hand across four or five systems that were never built to talk to each other. The China pipeline, which now represents 28 percent of the innovative drugs large pharma in-licensed in 2024, sits behind a firewall and in native-language filings most tools cannot read.

And because nothing refreshes on its own, a landscape is stale the week after it is built. Every valuation is rebuilt from memory under time pressure, making it hard to defend the moment the committee challenges it.

What changes when assembly is no longer the bottleneck

A system built specifically for BD&L conviction starts from the other end. It carries the data a deal turns on, resolved to the asset and kept current, so the work the old model leaves to the analyst is already done.

Ferma's own testing compressed a drug-asset competitive landscape from about two and a half analyst-days to roughly three hours, in line with independent academic findings on agentic diligence. The same workflow that took a team from 33,575 drugs to five acquirable CDH17 ADC targets in a single run now runs as the opening move, not the bottleneck.

The gains compound in three ways: the same team evaluates far more inbound, every deliverable arrives with its evidence attached so valuations hold up under committee scrutiny, and institutional memory no longer leaves when an analyst does.

Frequently Asked Questions