What questions should I ask a spatial biology vendor?

Ask which sample types are validated rather than merely compatible, the documented failure rate on tissue like yours, data volume and export formats at your configuration, software licensing terms and what happens if a subscription lapses, site requirements, service response times in your region, training depth, and whether a platform generation change is expected. Require written answers.

How do I evaluate a spatial platform demo?

Insist it runs on your own tissue, including your most difficult or oldest sample type rather than your best block. Observe the full hands-on workflow, not just the result. Require the raw output files rather than a curated figure, and have whoever will analyze the data open them before you decide. Record real timings from sections in to data out.

What hidden costs come with spatial platforms?

The commonly missed ones are data storage across your full retention period, compute and analysis software licensing, analyst time, facility modification before installation, service contracts starting after warranty, additional training when staff leaves, and sequencing where the workflow requires it. Consumables at realistic volume, including repeat runs, are usually larger than expected.

Are vendor performance claims verifiable?

Increasingly, yes, at least for spatial transcriptomics. Independent groups have published peer-reviewed benchmarks running competing platforms on matched tissue sections, so claims about sensitivity, resolution, and background can be checked against data the vendor did not produce. Equivalent head-to-head evidence for spatial proteomics platforms does not yet exist.

Evaluating Spatial Biology Vendors: Key Questions

Useful spatial biology vendor evaluation is not about collecting answers. Any competent sales team will answer everything you ask, plausibly and in good faith. The value is in asking questions where a vague answer is itself informative, and in knowing how to verify the rest independently. This is organized around that rather than around a checklist.


Key Takeaways Require written answers to a written question set sent identically to every shortlisted vendor. Verbal answers are not comparable or quotable later. Ask for documented failure rates on tissue like yours. Reluctance to give a number is the answer. Peer-reviewed benchmarks now exist for transcriptomics platforms, so performance claims can be checked against independent data. Reference calls only work if you choose the reference. Ask the vendor’s customer who else you should speak to. The answers a vendor cannot give in writing are the ones that will cost you money later.

Key Takeaways

Require written answers to a written question set sent identically to every shortlisted vendor. Verbal answers are not comparable or quotable later.
Ask for documented failure rates on tissue like yours. Reluctance to give a number is the answer.
Peer-reviewed benchmarks now exist for transcriptomics platforms, so performance claims can be checked against independent data.
Reference calls only work if you choose the reference. Ask the vendor’s customer who else you should speak to.
The answers a vendor cannot give in writing are the ones that will cost you money later.

Questions About Your Samples

Sample compatibility is the fastest eliminator in this category, and the area where a vendor can be technically accurate while leaving you with a wrong impression. Watch for the word compatible, which is weaker than validated and much weaker than routinely run.

Ask This	Weak Answer to Listen For	How to Verify
Which fixation and preservation types are validated, as opposed to compatible?	"It works with FFPE." No distinction drawn between validated and possible	Ask for the validation data or the application note, by document number
What is the documented failure rate on tissue of our type and age?	"Failures are rare." Any answer without a number	Ask a reference customer running your tissue type for their own repeat rate
What sample quality thresholds apply, and how should we measure them?	"Send us anything and we will take a look"	Ask which metric and what threshold. Published protocols specify these
Will you run the demonstration on our tissue, prepared by our staff?	Redirection to their reference tissue or a curated dataset	The answer itself is the verification. Reluctance is informative
What happens to a run that fails partway through?	Vague goodwill assurances with nothing in the contract	Get the reagent replacement and rerun policy in the quote, in writing

Table 1. Sample questions, with the answers that should prompt a follow-up and the independent check for each.

The demonstration request in the fourth row does most of the work. A run on your own difficult material, ideally your oldest sample type rather than your best block, tells you more than any specification.

Compendium

Rear view close-up of a female biologist pipetting cells in a laboratory

How to Optimize Culture Conditions for Scalable, Reproducible Stem Cell Research

Learn key strategies to enhance stem cell and iPSC culture across expansion, scale-up, and downstream applications

What Do Real Throughput and Failure Rates Look Like?

Nameplate and realized throughput differ substantially. Ask about the gap directly rather than about the specification.

What is the realistic time from sample received to data delivered, including any step performed off site?
How long does an instrument run take at the panel size or capture area we intend to use, not at the smallest configuration?
What proportion of runs need repeating, and what are the most common causes?
What is the realistic elapsed time from purchase order to first usable data?

The second question matters more than it looks. Run time often scales with panel size or imaged area, so a figure quoted for a small configuration can badly understate occupancy for the one you will buy. Published comparisons of both Spatial Transcriptomics Platforms Compared: Visium, Xenium, CosMx, MERSCOPE, and GeoMx and Spatial Proteomics Platforms Compared: PhenoCycler, Imaging Mass Cytometry, and MIBI include observed run times from independent studies, which sanity-checks whatever you are told.

Questions About Data and Analysis Support

The most under-asked area, and the most expensive to get wrong, because format and licensing decisions bind you for the life of the instrument.

Ask This	Weak Answer to Listen For	How to Verify
How much data does one run generate, at our configuration?	An answer in gigabytes when the honest answer is terabytes	Ask for a real dataset from a comparable run and check the size yourself
Is raw data accessible without your proprietary software?	"Our software handles everything you need"	Request a sample export and open it in an independent tool
What open standards does export conform to?	No named standard	Ask specifically. Named formats can be tested; “standard formats” cannot
Is analysis software perpetual or subscription, and how many seats?	Bundled pricing with no breakdown	Ask for the licensing terms as a separate line in the quote
What happens to our access and our existing data if we stop paying?	Reassurance without contractual language	Have the answer written into the agreement, not the email thread
Who supports analysis, and is it included or billable?	"Our team is always available to help"	Ask for the support scope in writing and the hourly rate beyond it

Table 2. Data and software questions. The last two rows are the ones most likely to produce an unwelcome surprise in year three.

Bring your prospective analyst to this conversation. Only the person who will work with the output can tell you whether it is usable, and an instrument producing data nobody in the building can process has not passed evaluation.

What Will This Actually Cost Over Five Years?

Purchase price is the smallest number in the decision, and the one vendors lead with. Ask for the rest explicitly, because none of it is volunteered.

What is the per-sample or per-section consumable cost at our stated annual volume?
What is included in warranty, what does the service contract cost afterward, and can it be negotiated into the capital purchase now?
What site work is our responsibility, and what are the power, environmental, and vibration requirements in writing?
How many staff does training cover, for how long, and what does additional training cost when someone leaves?
Is a chemistry or platform generation change expected, and will current consumables remain supported?

Ask that last one plainly rather than guessing. Vendors often say more than you expect under a confidentiality agreement, and buying weeks before a generational change is avoidable. To model all of the above against published reference rates, use the estimator in How Much Does Spatial Biology Cost? Budgeting for Instruments, Consumables, and Data.

Reference Checks

A vendor-selected reference will be positive, which is limited rather than useless. Two techniques improve the yield: ask that reference who else you should speak to, which often produces a name the vendor would not have offered, and ask questions that invite candour rather than endorsement.

Ask the Reference Customer	Why This Question Works
What would you do differently if you were buying again?	Invites reflection rather than a verdict, and almost always surfaces something concrete
What surprised you after installation?	Targets exactly the costs and constraints that were not disclosed up front
What proportion of your runs need repeating, and why?	Gives you a real failure rate from someone with no incentive to minimize it
How long did it take to get from installation to reliable production?	Reveals the method development period that quotes never include
How responsive has service been, and where is your engineer based?	Regional support quality varies far more than vendors acknowledge
Who else should I talk to about this platform?	Produces a reference the vendor did not curate

Table 3. Reference check questions designed to produce candour rather than endorsement.

Reach at least one customer running your tissue type at roughly your scale. A glowing reference from a lab running fresh-frozen mouse tissue tells you little if your cohort is decade-old human FFPE.

Which Answers Should Worry You?

Some responses are worth treating as findings rather than as gaps to be filled in later.


Signals Worth Slowing Down For Reluctance to run the demonstration on your tissue, or to hand over raw output files from it. No documented failure rate for tissue like yours, or a failure rate offered only verbally. Data volume quoted in units that seem too small, or no answer at your configuration. No named export standard, and no willingness to provide a sample file to test. Software licensing bundled with no line-item breakdown, or no contractual answer on lapsed access. Service response times described regionally but not contractually, or an engineer based implausibly far away. Only vendor-selected references offered, and no route to an independent user.

Signals Worth Slowing Down For

Reluctance to run the demonstration on your tissue, or to hand over raw output files from it.
No documented failure rate for tissue like yours, or a failure rate offered only verbally.
Data volume quoted in units that seem too small, or no answer at your configuration.
No named export standard, and no willingness to provide a sample file to test.
Software licensing bundled with no line-item breakdown, or no contractual answer on lapsed access.
Service response times described regionally but not contractually, or an engineer based implausibly far away.
Only vendor-selected references offered, and no route to an independent user.

None of these means a platform is wrong for you, and none is evidence of bad faith. Gaps often reflect what a vendor has not measured rather than what it is concealing. But an unmeasured failure rate becomes your repeat rate, and an unwritten support commitment becomes your downtime. Treat each gap as a costed risk in the business case, not an item to resolve after purchase.

Buyers do have one new advantage. Independent peer-reviewed benchmarks now exist for several spatial transcriptomics platforms, including a technical comparison across six cancer types that ran competing systems on matched tissue sections, so performance claims can be checked against data the vendor did not generate. Putting a published figure in front of a sales team and asking them to reconcile it is among the more productive conversations available here. Fit this into the wider process in Choosing Between Spatial Biology Platforms: A Lab Manager's Buyer's Guide, and see the operational picture across the whole workflow in Spatial Biology in the Lab: A Manager's Guide to Evaluating, Implementing, and Scaling Spatial Technologies.

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