Benchmarking Questions to Ask
For anyone running a benchmarking study: 20 questions to ask a peer organization, an industry group, or a counterpart on a call so you leave with numbers you can actually compare.
20 questions, each with the reason to ask it · includes a conversation guide
The questions
Open any question to see why it works.
- 1
How do you define the process we are comparing, from first step to last?
Two teams can use the same process name and mean very different work. Fixing the start and end points first is what makes every number after this comparable.
- 2
Which metrics do you actually manage this process by, and how is each one calculated?
The formula matters more than the figure. A cost per order can look wildly better or worse depending on whether returns, overhead, and freight sit in the numerator, so get the arithmetic before the answer.
- 3
What is your current volume, and over what time period?
Scale drives unit economics, so a rate from a team handling 500 cases a month is not a fair target for one handling 50,000. It also tells you whether their averages rest on enough volume to be worth anything.
- 4
Who owns this process, and how many people work on it?
Headcount plus reporting line reveals how much of their performance comes from staffing levels versus method. It also shows you whether the process has a single accountable owner or is split across functions.
- 5
What systems or tools support each step, and how much is manual?
Tooling explains most large gaps in cycle time and error rates. Knowing the manual touchpoints tells you which part of their result you could copy without a platform migration.
- 6
Where do the numbers come from, and how often are they reported?
A metric pulled automatically from a system of record is worth far more than one assembled in a spreadsheet each quarter. Reporting cadence also tells you how quickly they can detect a problem.
- 7
What does the full step-by-step flow look like, including handoffs and approvals?
Cycle time gaps almost always live in the waiting between steps, not the work itself. Walking the flow surfaces approvals you may be able to remove entirely.
- 8
What are your current results on the two or three metrics that matter most?
This is the actual benchmark, and asking it after the definitions means you can interpret it correctly. Getting two or three figures rather than a dozen keeps the conversation focused.
- 9
How have those results moved over the past two or three years?
A trend line tells you whether you are looking at a durable capability or a good quarter. A team improving steadily is a better model to learn from than one that peaked and plateaued.
- 10
What is your rework, error, or exception rate, and what causes most of it?
Speed and cost figures are meaningless without knowing the quality they were bought at. The dominant cause of exceptions often points straight at the upstream fix you need too.
- 11
What is the fully loaded cost of running this, and what is in that number?
Cost comparisons fail on scope, so you need to know whether technology, management time, and facilities are included. This is also where outsourced work often hides.
- 12
What single change produced your biggest improvement, and how long did it take to show up?
This turns a static number into a sequence you can follow. The lag between change and result tells you what to promise your own leadership about timing.
- 13
What did you try that did not work?
Failed experiments are the cheapest thing you can take from a benchmarking call and the part no published report contains. It also signals whether your counterpart is being candid or presenting a case study.
- 14
How much of your performance comes from factors you did not choose?
Geography, product mix, regulation, customer profile, and legacy contracts can account for most of a gap. Naming these keeps you from chasing a target that is structurally unavailable to you.
- 15
Where does demand spike, and how do you handle the peak?
Annual averages hide the periods where a process actually breaks. Their surge approach, whether flexing staff, queueing, or throttling, is often more useful than their steady state numbers.
- 16
What is the binding constraint on this process today?
The bottleneck reveals where their improvement effort is going next and whether they are near the practical limit. If you share the same constraint, you have found the highest value topic in the conversation.
- 17
What target do you hold yourselves to, and where did that target come from?
A target anchored to a competitor, a customer commitment, or a theoretical maximum tells you how ambitious the benchmark really is. Many targets turn out to be last year plus five percent.
- 18
If you were rebuilding this from nothing today, what would you do differently?
This gets past the defense of existing choices and surfaces the design they would actually recommend. For a team starting fresh, their hindsight is worth more than their current metrics.
- 19
Who else does this unusually well, in or outside our industry?
Practitioners know the real leaders, which are often not the companies named in analyst rankings. This is also how you find the out of industry comparison that produces a step change rather than a tweak.
- 20
What would be most useful for us to share back with you?
Benchmarking runs on reciprocity, and offering something specific is what earns you a second conversation and follow up data. It also reframes the exchange as peer to peer rather than an extraction.
Running a Benchmarking Conversation
Practical guidance for the conversation itself.
Do This Before You Ask Anything
Do This Before You Ask Anything
Measure yourself first
Arrive with your own numbers for the same metrics, calculated the same way. Without a baseline you cannot tell whether their answer is good, and you have nothing to trade.
Pick five metrics, not twenty
A one hour call supports roughly five metrics explored properly. Choose the ones tied to a decision you are about to make, and drop anything you would not act on.
Send the metric definitions in advance
Share a one page sheet with your formulas and units before the call so your counterpart can pull comparable figures. This is the difference between real data and rough recollections.
Decide what you can offer
Name in your first email what you will share in return, whether that is your own data, an anonymized summary of the whole study, or an introduction. Vague reciprocity gets declined politely.
Making the Numbers Genuinely Comparable
Making the Numbers Genuinely Comparable
Interrogate the denominator
Per employee, per transaction, per revenue dollar, and per active customer produce wildly different rankings from identical raw data. Always ask what sits below the line.
Check the time window
Ask whether the figure is a trailing twelve month average, a single best month, or a budget target. Best month numbers quoted as norms are the most common distortion in benchmarking.
Separate one-time effects
A migration, an acquisition, a layoff, or a regulatory change can dominate a year of data. Ask what was unusual about the period the number covers.
Normalize for scope, then stop adjusting
Adjust once for clear structural differences such as included steps or outsourced work, and document each adjustment. Endless normalization is how teams talk a real gap out of existence.
Signals the Benchmark Is Not Trustworthy
Signals the Benchmark Is Not Trustworthy
- Round numbers across the board, which usually means estimates rather than measurement.
- No one can explain how the metric is calculated, only what the dashboard displays.
- Best in class results with no explanation of what drove them.
- Refusal to discuss anything that went wrong or any current constraint.
- The figure comes from a marketing deck or an award submission rather than an operating report.
- A vendor supplies both the benchmark and the product that closes the gap.
- The peer set is described only as similar companies with no named criteria for inclusion.
Adapting These for Other Benchmarking Types
Adapting These for Other Benchmarking Types
- Compensation benchmarking: anchor on job scope and level rather than title, and always ask about the mix of base, bonus, and equity plus the effective date of the data.
- Pricing benchmarking: ask for realized price after discounts and rebates, not list price, and confirm what is bundled into the quoted figure.
- Vendor and outsourcing benchmarking: ask which clients resemble you in volume and complexity, and request the range across their book rather than the best case.
- Internal benchmarking across sites or teams: agree on definitions centrally before collecting anything, or the top performer will simply be whoever counts most generously.
- Site visits: ask to watch the process run rather than sit through slides, and interview the people doing the work as well as the manager hosting you.