Whose Data Is the Benchmark?
Two years ago you ran a benchmarking study for a retail bank, Client A. You brought a pre-existing cost-to-income methodology into the engagement; the bank paid for data collection, and the resulting compiled dataset of 40 anonymised peer data points belongs to them under the contract. You are now engaged by a credit union, Client B, for a similar piece of work. Client B's CFO has seen a conference talk you gave and wants the same benchmark, specifically asking for the scores of the twelve named regional banks in it so the board can see where the credit union sits. Separately, a consultant you mentor has asked whether he can use your methodology with his own client, a payments firm. Your contract with Client A says the compiled dataset is their property and that you will not disclose confidential information; it is silent on the methodology, which you were using before the engagement began.
1. How do you answer Client B's request for the twelve named scores?
Read the worked answer
Refuse the named scores, and offer something that solves the CFO's actual problem.
Three separate reasons, and it is worth stating all three because each closes a different door. First, ownership: the compiled dataset belongs to Client A under the contract, so you cannot transfer it. Second, source consent: the twelve banks supplied data on the understanding it would appear in anonymised, aggregated form. Twelve named entities in a regional market is not anonymisation - the participants are identifiable to each other and to anyone who knows the market, and several may be identifiable from combined attributes even without names. Third, reliance: the CFO wants to benchmark against those institutions, which means the output will be repeated in a board pack and probably to regulators, which is disclosure you are not licensed to make.
What you can offer: an aggregated distribution - quartiles, median, spread - with the credit union plotted against it, plus a fresh collection from a comparison set the credit union chooses, run on the same methodology. That gives the board a defensible position statement without exposing anyone.
Say clearly that you will not name the twelve even informally, so the CFO does not spend the meeting trying to find a softer version of the ask.
2. Can you reuse your methodology with the mentoring consultant's client?
Read the worked answer
Probably yes, but the analysis has to be done rather than assumed, and the distinction that matters is between the method and the evidence.
Methodology is the framework, definitions, calculation logic and reporting structure you developed before Client A engaged you and brought with you. Reusing it is generally legitimate, and it is how consultancies build practice. But three checks come before you say yes.
Contract terms. Read the Client A agreement fully, not just the clause you remember. Look for assignment of intellectual property in deliverables, restrictions on serving competitors for a period, and any definition of confidential information wide enough to capture the framework as it was applied. If the agreement assigns everything created during the engagement, your position is weaker than you think.
Contamination. Does the framework, as you would hand it over, embed Client A's data, calibrations, thresholds or observed values? A methodology tuned to one bank's cost base is no longer just a method. Strip it back to generic definitions and rebuild the parameters for the new client.
Competitive sensitivity. If the payments firm competes with Client A, check for a non-compete and disclose the new engagement to Client A if the relationship is material.
Then give the mentee a written licence limited to the generic framework, with no transfer of data and no right to describe it as validated on Client A's results.
3. What should go into your data-collection agreement next time?
Read the worked answer
The agreement you had answered the easy question - who owns the spreadsheet - and left the commercially important ones open. Six terms would fix that.
Purpose limitation and permitted outputs. State what participants are contributing to, in what form results will appear (aggregate, quartiles, individual position only to the contributor), and that named or identifiable disclosure to third parties will not occur. This is what lets you say no to Client B without improvising.
Licence rather than ownership. Take a defined licence to use the contributed data for benchmarking and derived products, rather than a bare assignment of the compiled dataset. Ownership without a licence is the worst combination: you cannot reuse and they cannot either.
Derived data. Agree explicitly who owns benchmarks, indices and models built on top of the contributions, and whether you may reuse the aggregate after a defined period, say 24 months, once it is refreshed.
Pre-existing IP. A schedule listing the methodology, templates and tools you brought in, stated as remaining yours.
Re-identification and small-cell rules. Minimum participant counts before any cut is reported, so no cell traces back to one firm.
Termination and return. What happens to the dataset on exit, and what survives.
Have counsel review the wording; the drafting is standard but the consequences are not.