An NGS program does not fail because the sequencer was the wrong choice. It fails because nobody decided, in writing, before the purchase, what volume the program needed to reach, what it would cost per sample at that volume, and what would happen if demand came in lower. The instrument almost always works. The program around it is what falls apart, and it falls apart quietly, over the eighteen months it takes for a grant-funded purchase to become a standing cost nobody planned for.

This section overview covers the five decisions that determine whether a program is defensible: whether NGS belongs in-house at all, how to size demand honestly, what the full cost actually includes, how to structure the business case, and how to measure return once the instrument is running. Platform-level cost, run economics, and the day-to-day operational picture for a program already underway are covered in the main guide to next-generation sequencing in the lab. This piece sits one level up: it is about the decision to build the program, not about running it once built.

Deciding Whether NGS Belongs In-House

Start from the presumption that outsourcing is the default, and make ownership justify itself against that default, rather than the other way around. This runs against instinct in a lab that already has a strong track record with the technique, but the arithmetic is unforgiving at low volume. An independent comparison of the total cost of owning sequencing platforms against outsourcing found that outsourcing was the more cost-effective path for the large majority of labs, and that ownership only wins once volume is sustained over the instrument’s full lifetime, not just in a single busy year.

Make the comparison concrete rather than directional. Using the main hub’s stated cost model, a lab’s fully loaded in-house cost per delivered genome is its variable cost plus its fixed cost divided by volume. If you assume $910.50 of variable cost per delivered genome and $257,000 of annual fixed cost, both taken directly from that model, the in-house cost only falls below an outsourced price once volume clears a specific threshold. Below it, in-house is the more expensive option even though the reagent line looks cheaper in isolation.

Table 1. Break-even sequencing volume against three illustrative outsourced-rate assumptions, using the main hub’s stated $910.50 variable cost and $257,000 fixed cost per year. These are planning assumptions, not sourced market pricing; substitute your own quoted outsourced rate.

Cost is not the only input, and for some programs it is not even the deciding one. Turnaround control, protection of sensitive or irreplaceable samples, method flexibility for non-standard protocols, and the strategic value of holding the capability in-house are all legitimate reasons to own even when the outsourced quote is cheaper. The failure mode is not choosing to own for those reasons. It is presenting a control-and-flexibility decision as if it were a cost-saving one, because that version of the business case collapses the first time someone runs the numbers independently. Say plainly which reason is driving the decision, because the review conversation is completely different depending on which one it is.

Advanced Lab Management Certificate

The Advanced Lab Management certificate is more than training—it’s a professional advantage.

Gain critical skills and IACET-approved CEUs that make a measurable difference.

Sizing Demand Before You Size the Instrument

Demand forecasting for a new NGS program is usually done backward: pick an instrument, then describe a demand level that justifies it. Do it forward instead. Separate demand into three tiers and size the program against the lowest one, not the highest.

• Committed demand: samples with funding already in hand and a named project. This is the only tier a business case should be built on.
• Pipeline demand: grants submitted but not yet awarded, or projects in active planning. Real, but apply a conversion rate rather than counting it at face value; a reasonable starting assumption for an academic core is well under half converting within the funding year.
• Aspirational demand: everyone who says they would use the capability if it existed. Useful for a five-year vision, useless for a purchase justification, because it does not survive contact with an actual invoice.

A worked example: a core serving four research groups has 180 committed samples a year today, two pending grants that would add roughly 220 samples if both are funded, and a further dozen groups who have expressed informal interest representing perhaps 400 samples of aspirational demand. Size the business case on the 180, note the pipeline as upside with its conversion rate stated, and mention the aspirational figure only as context for a possible second phase. A committee that sees 180 stated honestly trusts the rest of the document more than one that opens with 800 and quietly means all three tiers added together.

Seasonality is the other trap. Academic demand in particular arrives in bursts around grant cycles and semester boundaries, and an instrument sized for the annual average will be underused for months and queued for others. Size for the sustained level you can defend, and plan explicitly for how overflow gets handled during peaks, whether that is an outsourcing arrangement, a waiting list with a stated turnaround, or a second instrument once the program has proven the first one out. Deeper detail on batching demand against flow-cell capacity is covered in Running NGS at Scale: Throughput, Scheduling, and Automation, and the readiness question, whether your lab has the foundation to take this on at all before you get to volume, is covered in Is Your Lab Ready for NGS? A Readiness Assessment.

Interested in lab leadership?

Register for a FREE Lab Manager account to subscribe to our Lab Leadership Digest Newsletter.

Subscribe for Free

The Full Cost Stack

Most instrument-purchase business cases price two things: the capital cost and the reagent cost per run. Both are necessary and neither is sufficient. A defensible cost stack has at least six components, and the ones most often left out are exactly the ones that turn a promising first year into a problematic third year.

Storage deserves particular attention because it behaves differently from every other line in this stack. Capital and reagent costs scale with how much sequencing you do. Storage does not shrink even when sequencing volume is flat, because each year of retained data adds to the last rather than replacing it. A program budgeted only for its first year of storage will show a growing, unexplained cost line by year three that has nothing to do with how much new sequencing occurred. Full modeling of that effect is in the main hub, and detailed retention policy guidance sits in Managing NGS Data: Storage, Compute, Retention, and Staffing.

One general caution worth carrying into any cost model: NHGRI’s widely cited cost-per-genome data reflects production cost at large, specialist sequencing centers, not the cost structure of a typical lab building its own capacity. Any headline figure drawn from that dataset, or from vendor marketing built on it, understates what a real program spends, because it excludes exactly the labor, storage, and bioinformatics components that make up the rest of this table.

Building the Business Case

A business case that survives its first budget review has five parts, in this order. Skipping the order is the most common mistake, because it is tempting to lead with the instrument rather than the problem it solves.

• The problem, stated in terms of the science or service it enables. Not "we need a sequencer," but "these four groups cannot currently answer this class of question without shipping samples out and waiting six weeks."
• Demand, at the committed tier, with pipeline and aspirational demand clearly separated. See the framework above. This is the section reviewers scrutinize hardest, and the section most business cases get vaguest about.
• The full cost stack, all six components, against the demand figure just stated. Not against a hoped-for future volume.
• The buy-versus-outsource comparison, with the outsourced rate assumption stated explicitly. If ownership is being chosen for reasons other than cost, say so directly rather than presenting a marginal cost case as decisive.
• A named review date, eighteen to twenty-four months out, with the metrics that review will use stated in advance. This is the single highest-leverage sentence in the entire document, because it converts an open-ended commitment into a bounded one.

That last point is worth dwelling on. A business case with no review date is not more optimistic than one with a date. It is simply unfalsifiable, and unfalsifiable commitments are the ones that get quietly defunded in year three without anyone having agreed on the criteria beforehand. Naming the date and the metrics up front means the year-three conversation is about evidence rather than about impressions. A ready-to-adapt template covering readiness, cost, and the review structure is built out in How to Build a Business Case for an NGS Instrument, and the specific question of whether to run in-house or through a core facility or external provider is developed further in In-House vs. Outsourced Sequencing: Core Facility or Service Provider?.

How Do You Measure Return Once You Are Running?

Report three categories separately rather than collapsing them into a single return figure. Operational return is utilization, turnaround, and repeat rate: the evidence that the program is running well. Scientific or clinical return is publications, grants secured, diagnostic yield, or decisions enabled, and it lags purchase by months to years depending on the setting, so reviewing it too early will make a healthy program look like a failure. Financial return is cost per delivered sample against the outsourced alternative, tracked against the same break-even model used in the original business case rather than against a new set of assumptions invented after the fact.

The instrument that most quickly reveals whether a program is in trouble is utilization, not cost. A program running at low utilization is expensive in a way that no operational tightening can fix, because the cost sits in idle capital rather than in anything a lab manager controls day to day. Watching utilization from month one, rather than waiting for the eighteen-month review to discover it, is the cheapest insurance available against a program quietly failing in a way nobody notices until the numbers are reviewed. Yale’s genome analysis center offers a useful illustration of this discipline in practice: scheduling decisions there are made specifically to protect utilization and turnaround simultaneously, as described in this recent look at workflow trade-offs in sequencing core labs, where instrument time is deliberately held open to absorb urgent cases without abandoning production throughput.

Whichever review structure you choose, use the same metrics named in the original business case. Changing the measure after the fact, even with good intentions, is indistinguishable from moving the goalposts to whoever is reading the review, and it costs the program credibility exactly when it needs it most.

This article was produced under Lab Manager's AI Editorial Guidelines.