Diagnostics as a Service or your own lab, and what running blood testing really costs

Wellness brands that end up buying Diagnostics as a Service have usually tried building it themselves first. Earlier this year I spoke with the head of product at one of them, a company that had spent nine months bringing blood testing in-house. When I asked him how it was going, he told me that his best engineer had spent the whole previous quarter writing courier tracking code.

He was not complaining about the money, because the budget had been signed off and it was holding. What bothered him was that nobody had ever decided this, since there was never a meeting where the team agreed to become a logistics operation. The work reached them one sprint at a time, and by the time he noticed what had happened the roadmap he had shown his board was two quarters behind, and the reason for that was a van.

I have heard versions of that conversation more than once, and the shape of it is always the same, because the money is planned for properly while the effect of the work on the people is not. None of that effect looks like a line in a budget until it is already happening.

So this is the other half of the business case, which is the four things you take on after the budget has been signed off, roughly in the order that you meet them.

First, your engineers stop building your product

You hired a product team to build the thing your customers actually buy, and a blood testing operation gives that team a second job.

Somebody has to own each of the following, but not one of them is your product:

• Turning a customer order into a laboratory requisition, in the exact format that laboratory wants it.
• Getting the right tube to the right person before the appointment takes place.
• Booking a courier, and knowing quickly when that courier did not appear.
• Reading results back in and mapping every marker to a reference interval you trust.
• Deciding what happens when a value comes back outside that interval.

Each of those jobs sounds small on its own, but together they add up to a team, and nobody sets out to build that team deliberately. You take the people you already have, you give them one more responsibility each, and the roadmap you promised slips by a quarter and then by another one. Eighteen months later you have a working courier tracker and a product that looks much as it did before.

Your engineers also discover that laboratories do not share a single way of working, because each one wants its data in its own shape and most systems were built for the volume a clinic sends rather than a consumer brand. Several steps that look automatic also still need a person to type something in. That work does not go away as you grow, since it grows along with you. Our developer documentation shows what the same job looks like when somebody else has already done it, and our article for product teams goes through the integration in more detail.

Second, a year passes before anybody pays you

While that team is forming inside your company, you have nothing at all to sell.

Quality systems and laboratory accreditation both need months of documented work and an external assessment before a single customer can buy anything, so you are hiring, writing procedures and going through audits while your revenue line stays flat. The difficulty is not really the spend, because the harder part is defending that spend every quarter with nothing finished to show for it.

That timeline becomes considerably longer if you decide to sell your own home test kit, which is the decision that people underestimate most often. Under the in vitro diagnostics regulation, a test that a customer performs on themselves is class C by default.¹ That classification brings a notified body assessment, a study proving that untrained people can use your kit correctly, and a requirement that the kit warns a user when a result is not valid.²

There is also a queue in front of you, because only nineteen bodies across Europe are designated to carry out in vitro diagnostics assessments, those nineteen are based in twelve countries, and nine member states have none at all.³ How long you wait for one of them is not something you control.

Sending a sample from a professional blood draw to an accredited laboratory does not put you in that position at all, and the difference is the one between manufacturing a medical device and buying a service. Most companies learn that difference after the packaging has already been printed.

There is one date worth checking against your own plan. The class C legacy transition runs until 31 December 2028, but the extension only applies to devices that met three earlier deadlines. Those deadlines were a compliant quality system by 26 May 2025, a notified body application by 26 May 2026 and a signed written agreement by 26 September 2026.⁴ The application window has therefore already closed, and missing it forfeits the extension entirely.

Third, you have to hire and buy in a field you do not know

At some point in that first year you need a quality manager and somebody who understands regulatory affairs, but if you have never worked in diagnostics you will not be able to tell a strong candidate from a plausible one. The cost of getting that hire wrong reaches you eighteen months later, in an audit.

The same gap appears with your suppliers, because couriers, kit manufacturers and laboratories will not give you a serious quote until you look like a serious account, and you do not look like one yet. So you negotiate with no volume behind you, no history and nothing to compare the offer against, which is why most companies accept the first workable price and find out later what they agreed to. If you want that decision written from the buyer's side, our procurement guide covers what to ask for.

Fourth, some samples have to reach a laboratory within hours

By this point you have a team and a supplier, and then the samples start moving, which means your working day is set by biology rather than by your own plans.

A large blood count with a differential has to reach a laboratory within about eight hours, and some coagulation samples give you closer to four. Uncentrifuged serum must not go into a fridge at all, because below four degrees the cells begin leaking potassium and the result stops being true.⁵

An eight-hour limit rules out the post entirely, and what it requires instead is a courier collecting from a place you control and driving to a laboratory close enough to receive the sample in time. That is a network, and a network is a daily operation, which means somebody in your company is watching a van and telephoning a customer whenever that van is late.

There is an awkward consequence for your product as well, because the markers customers most want to see are often the ones with the shortest time limit. The broad panel you want to sell and the convenient home kit you want to ship do not combine as neatly as your plan assumed, and our article on a first launch works through that choice in detail.

Then somebody has to tell a customer to give blood again

Samples fail for ordinary physical reasons, because they clot on the way, they arrive warm, the tube was underfilled or the label came off in transit.

None of that means something has gone wrong with your operation, since it is the normal condition of the work. A meta-analysis of 26 studies covering more than 16 million sample requests found a pooled rejection rate of about two in every hundred, with clotting, breakdown of the sample, insufficient volume and labelling errors accounting for most of it.⁶ A 2025 study of roughly 11 million specimens at a single laboratory found that 98.4 percent of all recorded errors happened before the sample was analysed.⁷

So your support team ends up writing the message. They have to tell a customer who has already paid, already fasted and already given up a morning that none of it counted and the whole thing has to happen again.

That costs you more than the sample does, because you lose the tube and the shipping, you lose the money you spent acquiring that customer, and more often than the spreadsheet admits you lose the customer as well. It becomes a routine part of your week rather than an exception to it.

You should expect to hear about it publicly too, because customers who feel let down by a health test tend to say so where your next prospect can read it.

Eventually a result comes back that nobody can answer

The last of these is the one people have thought about least, and it is also the one with real consequences for a person.

Sooner or later a value comes back that is genuinely worrying, and it reaches you on a Friday evening when there is no clinician employed anywhere in your company.

At that point you need answers to questions you have probably never written down:

• Who contacts the customer, and how quickly that has to happen.
• What that person is allowed to say, and what they must not say.
• Where you send somebody who needs real medical advice the same evening.
• Who is responsible if the message is wrong or reaches the customer too late.

Every company that offers testing has to deal with this situation eventually. The ones that handle it well had decided the process in advance, while the ones that handle it badly improvised in the evening with a frightened person waiting on the other end. The responsibility belongs to whoever sold the test, so your brand is on the result and the consequence is yours.

What you cannot find out before you commit

Having read all of that, you might reasonably want to put numbers to it, and some of those numbers cannot be had in advance.

Courier and cold chain costs are quoted per customer, so you cannot price them until you are already negotiating, and sampling device suppliers work in exactly the same way. Nobody will tell you what a lay-user usability study costs until they have seen your kit.

Your own failure rate is also unknowable before you start, because the published rejection rates above are a guide rather than your number. You will not know how many of your samples come back usable, or how many customers complete a test after buying one, until real people have gone through your real process. Those are the figures your business case depends on, and they are exactly the ones you cannot have on the day you have to decide.

When building it yourself beats Diagnostics as a Service

Sometimes building it yourself is the right decision, and pretending otherwise would be dishonest.

• Build it if diagnostics is your product rather than a feature of something else, because then the quality system is what your customers are paying for.
• Build it if your volume is large enough that a partner's per-test price across a year costs you more than running the operation yourself.
• Build it if you intend to sell testing capacity to other companies, because at that point you are properly in this industry.

If none of those three describes your company, then the real choice is between renting the capability and rebuilding it, and rebuilding it takes a year of your team's attention.

How Aniva delivers Diagnostics as a Service instead

Aniva does the work described above so that your own team does not have to.

• One contract covers the laboratory, the kits and the couriers, together with the ordering software, the results and all of the compliance paperwork.
• Your customers see your brand on the dashboard and on the report, and on your own sub-domain if you want one.
• Aniva owns its blood draw network across Germany instead of subcontracting it, which is what makes an eight-hour sample window workable.
• Diagnostics as a Service, or DaaS, is available to partners across Europe, and what changes at each border is covered in our article on European coverage.
• Four weeks from a signed contract to your first blood draw.

Bring your expected volume and the panel you want to sell, and the first call will give you a real number for your model. Book a 30-minute demo, or send your engineers to the API documentation first.

A summary of the main points

• The budget is the part you will plan properly, while the staffing is the part that surprises you.
• Your product team takes on ordering, kit logistics, courier tracking and abnormal results, and your roadmap is the first thing to slow down.
• A quality system and accreditation take months before anybody can buy anything, and selling your own kit adds a notified body and a queue.
• An eight-hour limit on a large blood count decides whether you build a collection network or rent one.
• Sample failures are normal, and the conversation asking a customer to give blood again becomes part of your week.
• A worrying result will come back on a Friday evening, and the responsibility belongs to whoever sold the test.

Questions we are asked most often

What is the hardest part of running blood testing in-house?

Most companies expect the cost and are surprised by the staffing, because ordering, kit logistics, courier tracking, result parsing, reference intervals and abnormal result handling all need an owner. In practice the product team you hired for something else takes them on, so your roadmap slows down first and the reason for it becomes clear much later.

How long does it take to set up your own blood testing operation?

Plan in months rather than in weeks, because a quality system and laboratory accreditation both need documented work and an external assessment before you can sell anything. If you intend to place your own home test kit on the market you also join a queue, since only nineteen bodies across Europe are designated for in vitro diagnostics assessments.³ Working with an established provider takes a few weeks instead.

Does selling a home test kit make me a medical device manufacturer?

Placing your own self-test kit on the EU market brings you under the in vitro diagnostics regulation, where a self-test is class C by default.¹ That means a notified body, a study showing that untrained users can operate the kit, and a requirement to alert the user to an invalid result.² Sending a professional blood draw to an accredited laboratory avoids that route, though you should confirm your own position with a regulatory advisor.

Why can a blood sample not simply be posted?

Some markers degrade within hours of the sample being taken, so a large blood count with a differential needs to reach a laboratory within roughly eight hours and certain coagulation samples give you about four.⁵ That rules out ordinary post for those panels, and what it requires instead is a courier running from a collection point to a nearby laboratory.

How often do blood samples fail?

A meta-analysis of 26 studies covering more than 16 million requests found a pooled rejection rate of about two in every hundred, with regional subgroups ranging from 0.55 percent to 2.82 percent.⁶ Separately, a 2025 study of roughly 11 million specimens found that 98.4 percent of recorded errors happened before analysis.⁷ Your own rate depends on your collection method and cannot be known before you run real samples.

Who is responsible when a result is worrying?

The responsibility belongs to whoever sold the test, which is why the escalation path has to be decided before you launch. Write down who contacts the customer, how quickly they do it, what they are allowed to say, and where the customer goes for medical advice. Interpretation of an individual result belongs with a treating clinician.

Notes and sources

1. Regulation (EU) 2017/746 on in vitro diagnostic medical devices, Annex VIII, Rule 4(a): devices intended for self-testing are classified as class C, except for pregnancy, fertility and cholesterol testing and certain urine tests, which are class B. legislation.gov.uk
2. Regulation (EU) 2017/746, Annex I, point 19, on devices intended for self-testing, including point 19.3 on procedures allowing a user to verify performance and be warned of an invalid result. legislation.gov.uk
3. European Commission list of designated notified bodies, checked 21 August 2026: 19 designated under Regulation (EU) 2017/746, based in 12 countries, with 9 member states having none. Single Market Compliance Space
4. Regulation (EU) 2024/1860, amending the in vitro diagnostics transition dates. Class C legacy devices run to 31 December 2028, conditional on a compliant quality system by 26 May 2025, a notified body application by 26 May 2026 and a signed written agreement by 26 September 2026. EUR-Lex
5. LADR preanalytics guidance on whole blood, EDTA plasma and serum stability. ladr.de
6. Getawa S, Aynalem M, Melku M, Adane T. Blood specimen rejection rate in clinical laboratory: a systematic review and meta-analysis. Practical Laboratory Medicine 2022;33:e00303. Pooled rejection rate 1.99 percent across 26 studies and 16,118,499 requests, with subgroups from 0.55 percent in the Americas to 2.82 percent in Asia. doi:10.1016/j.plabm.2022.e00303
7. Lin Y, Spies NC, Zohner K, McCoy D, Zaydman MA, Farnsworth CW. Pre-analytical phase errors constitute the vast majority of errors in clinical laboratory testing. Clinical Chemistry and Laboratory Medicine 2025;63(9):1709 to 1715. Of 87,317 recorded errors across roughly 11 million specimens, 98.4 percent were pre-analytical. Single institution, and the figure is dominated by haemolysis. doi:10.1515/cclm-2025-0190

This article is general information for companies deciding whether to run blood testing themselves. It is not legal, regulatory or medical advice. Laboratory accreditation and reference intervals belong to the analysing laboratory, and interpretation of any individual result remains with a treating clinician.