Appearance
25.1 — What This Industry Actually Is
Hold one tablet in your hand. Before it reached you, somewhere between eight and fifteen years passed, and somewhere around fifty separate organisations touched it.
A university lab found the biological reason the disease happens. A biotech company turned that reason into a chemical that interferes with it. A contract laboratory poisoned rats with it on purpose to find the dose that harms. A contract research organisation ran it through three rounds of human trials across two hundred hospitals in nineteen countries. A regulator read forty thousand pages about it. A factory made it under rules written after people died from a factory that did not follow them. A wholesaler shipped it. A pharmacy dispensed it. An insurance company paid for most of it and argued about the rest. And a software system you have never heard of recorded every one of those steps, because if a step is not recorded, in this industry it did not happen.
That last sentence is the one that pays your salary.
The industry in one picture
"Life sciences" is a business term, not a scientific one. It means every organisation whose product is knowledge about living things turned into something you can buy: a medicine, a vaccine, a diagnostic test, an implant, a scanner, a surgical robot.
"Healthcare" is the delivery end: hospitals, clinics, doctors, insurers, pharmacies — the machinery that gets those products into an actual human being and settles the bill.
They are two halves of one chain, and almost every job in the chain belongs to one of seven groups.
The discoverers. Pharmaceutical companies, biotechnology companies, university labs, and the medical device companies that design the physical objects. They carry the risk of finding something that works.
The testers. Contract research organisations, called CROs, run clinical trials for the discoverers. A CRO does not own the drug. It owns the people, the process and the systems that get a trial done correctly in twenty countries at once. Central laboratories, imaging labs and specialist safety groups sit alongside them.
The makers. Contract development and manufacturing organisations, called CDMOs, make the physical product for companies who do not own factories. Plus the packaging suppliers, the sterile-container makers, the cold-chain shippers.
The regulators. The FDA in the United States, the EMA in Europe, CDSCO in India, PMDA in Japan, and about a hundred more. They do not make anything. They decide what is allowed to exist, and their rules shape every system you will ever build here.
The providers. Hospitals, clinics, surgical centres, pharmacies, laboratories, the doctors and nurses inside them. They are where the product meets a patient.
The payers. Insurance companies, employers, and government programmes — in the United States, mainly Medicare and Medicaid. They decide what gets paid for and at what price, which is a second approval process running in parallel with the regulator's.
And the service layer — consultancies, information-technology companies, data companies, staffing companies. This is where you sit. The service layer exists because every group above has more work than it has people, and because most of that work is now software.
Follow the money, because everything else follows it
In 2024 the United States spent $5.3 trillion on health care. That is 18.0 percent of the entire national economy, and about $15,474 for every person in the country. Medicare accounted for roughly 21 percent of it and Medicaid about 18 percent, which means the government is by far the largest customer in the system.
Now notice what that money is not. It is mostly not spent on medicines. Hospital care and physician services take the largest shares. Prescription drugs are around a tenth of the total. A great deal of the argument in this industry — over prices, over coverage, over who is to blame — comes from people comparing different slices of that number without saying which slice they mean.
The path a single dollar takes is worth memorising, because it explains behaviour that otherwise looks irrational.
You pay a premium to an insurer every month, or your employer pays most of it for you, or a government programme covers you. You get sick. A hospital treats you and sends a claim to the insurer. The insurer pays a negotiated amount, not the amount printed on the bill, and passes some cost back to you as a copay or deductible. For medicines, a middle organisation called a pharmacy benefit manager, or PBM, negotiates with the drug manufacturer for a rebate and decides which drugs are on the covered list. The manufacturer's public price and the money it actually keeps are two very different numbers.
Every one of those handoffs is a transaction in somebody's software. That is why healthcare information technology is enormous and why claims data is the most valuable data set in the country.
Why it takes so long and costs so much
These are the numbers that explain the culture of the whole industry.
Out of every hundred drugs that begin the first human trial, roughly seven eventually get approved. Recent industry analysis puts the overall success rate from Phase I to approval at around 6.7 percent, down from about 10 percent a decade earlier. The worst step is Phase II, where only about 28 percent of programmes survive — that is the phase where you first find out whether the drug actually treats the disease in patients rather than merely doing something interesting in the body.
The cost of getting one drug to market, counting all the failures that had to be paid for along the way, is estimated by Deloitte's annual analysis of large pharmaceutical companies at about $2.23 billion per asset in 2024. Other methods that count only the direct research spend on the successful drug give figures closer to a few hundred million. Both are honest; they measure different things. The larger number includes the ninety-three failures out of a hundred, and the cost of money tied up for a decade.
And the output is small. In 2025 the FDA's drug centre approved 46 novel drugs — 34 new chemical molecules and 12 biological products. Forty-six new medicines for the whole world in a year, from an industry spending hundreds of billions.
Hold those three numbers together and the industry's personality makes sense. When 93 percent of attempts fail, and one success has to pay for all of them, and a mistake can kill people and end the company, you get an industry that is slow, cautious, obsessed with documentation, and willing to pay well for anyone who reduces its risk. It is not bureaucratic by accident. It is bureaucratic on purpose, and Chapter 25.2 shows you the bodies that each rule was written over.
Where a services company actually earns its money
A pharmaceutical company's own staff are expensive and are meant to be doing the science. Almost everything that is not the science gets handed to somebody else. The work that flows out is remarkably consistent, and it is worth knowing the list by name, because these are the projects that land on your desk.
| Work handed out | What the client actually needs |
|---|---|
| Clinical data management | Trial data captured, cleaned, locked |
| Safety case processing | Every side-effect report handled on the clock |
| Regulatory publishing | Submission documents assembled to spec |
| Computer system validation | Proof that software does what it claims |
| Quality system support | Deviations, CAPAs and audits kept moving |
| Commercial and patient data | Analytics, reporting, patient program systems |
| Revenue cycle and claims | Bills coded, submitted, denials chased |
| Interoperability work | Hospital systems made to talk to each other |
Read that table as a menu of problems rather than a list of departments. Each row exists because the client has a volume of work that grows faster than their headcount and a regulator who will punish them if the work is done badly. Those two facts together are the entire business case for your company, and they are also where every innovation opportunity comes from: anything that keeps the regulator satisfied with fewer human hours is worth real money to a client.
One warning that will save you an embarrassing meeting. Efficiency alone does not sell here. A proposal that makes a process faster but weakens the evidence trail will be rejected by a quality director without discussion, no matter how good the demonstration was. The winning proposal is always "same or better evidence, less human effort". Chapters 25.20 and 25.21 explain what "evidence" means in this industry precisely enough that you can make that argument yourself.
The words you will hear in week one
Three of them cause more confusion than the rest combined, so they are worth settling now.
GxP is a family name, not a single rule. The x stands for whichever practice applies: GLP is Good Laboratory Practice, which governs the safety studies done before humans; GCP is Good Clinical Practice, which governs trials in humans; GMP is Good Manufacturing Practice, which governs the factory; GDP here means Good Distribution Practice, which governs storage and shipping. "Is this system GxP?" is the first question anyone will ask about a piece of software you build, and it means: if this system is wrong, can a patient be harmed or can a regulatory decision be wrong? If yes, the system carries a whole extra world of rules.
Sponsor means the organisation that owns a clinical trial and is legally answerable for it. Usually the pharmaceutical company. Not the hospital, not the CRO, even when the CRO does all the work.
Indication means the specific disease a medicine is approved to treat. The same molecule with two indications is, in regulatory terms, two different products with two different labels. When a client says "we are filing a new indication", they mean a new approval for a drug that already exists.
What this Part will do for you
By the end of it you will be able to sit in a room with a regulatory director, a quality lead, a clinical operations manager and a hospital's chief information officer, follow the whole conversation, and know which of them can stop your project and why.
The order is deliberate. History first, because the rules only make sense once you know what went wrong. Then the making of a medicine from the first hint of a cause to the factory floor. Then the regulators. Then the quality and data-integrity machinery, which is the part of this industry that touches software most. Then everything that happens after a product is approved — safety monitoring, medical affairs, patient programs. Then the hospital and the American payment system, because that is where a large share of service work actually is. Then health data itself, and finally the engineer's own chapter, the roles directory and the acronym dictionary.
Next: Chapter 25.2, the history — including the two disasters that between them created the entire modern approval system, and the reason a drug label exists at all.