Norms databases: how agencies build benchmarks from client studies

What is a market research norms database?#

A market research norms database is a structured store of standardized scores from past studies, such as purchase intent from concept tests, recall and persuasion from ad tests, or brand health measures from trackers. Each score sits beside metadata that says what was tested, where, when and how, so a new result can be read against a relevant benchmark instead of in isolation.

For many agencies the norms database is the most defensible thing they have built. Clients buy the comparison as much as the fieldwork, and a deep, well-kept norm set is hard for a new competitor to copy. That same depth is why the rights question matters: every norm rests on studies that clients paid for.

Norms work at the study or cell level. The database holds the top-box score for a concept in a category and market, not the individual answers behind it. That design choice keeps personal data out of the benchmark and makes the rights review simpler.

What fields does a norms record need?#

A norms record needs enough metadata to make two studies comparable and enough provenance to remove a study later. Agencies that skip provenance fields find out when a client asks to be taken out and nobody can say which cells that client's studies feed.

Which contract clauses let an agency pool client studies?#

The client contract decides whether a study can feed the norms, so the rights review starts with the master services agreement and each statement of work. Read them together, because a statement of work can narrow what the master agreement allows.

Sort contracts into three groups: those that grant aggregated use, those that are silent and those that forbid it. Silent contracts are the hard middle. A cautious agency treats them as ineligible until the client confirms in writing, which costs a few emails and avoids an argument later.

• 1Aggregated-use clause: lets the agency use results in a form that identifies neither the client, its brands nor any respondent.
• 2Ownership clause: separates the client's deliverables from the agency's methods, tools and benchmark data.
• 3Confidentiality clause: check whether it carves out aggregated use or treats every score as client confidential information.
• 4Survival on termination: confirms the aggregated-use right continues after the relationship ends.
• 5Opt-out wording: some clients sign only if they can exclude specific studies or categories.
• 6Respondent notice: confirms the privacy notice and consent text allow research uses beyond the single study.
• 7Sample supplier terms: panel providers may limit how results from their respondents are stored or reused.

How do aggregation thresholds and client opt-outs work?#

Aggregation thresholds are the written rules that stop a norm from revealing any one client's results. A norm cell is published only when it contains a minimum number of studies from a minimum number of distinct clients, and no single client contributes more than a set share of the cell. The agency chooses the numbers; what matters is that they are written down and applied every time.

Thresholds interact with category granularity. A cell for one narrow subcategory in one market may hold studies from only a pair of competing clients, and either could infer the other's scores. When a cell fails the threshold, roll it up into a broader category rather than lowering the bar.

Opt-outs are recorded at the study level. When a client asks to be removed, flag every affected study, recalculate the cells it fed and keep a dated log of the change. A norm that cannot survive the loss of one client's studies was probably too narrow to publish.

How do you keep norms comparable as methods change?#

Norms stay comparable only when the agency versions its questionnaires and tracks method changes as carefully as the scores. A new scale, a new panel source or a shift to mobile-first questionnaires can move scores for reasons that have nothing to do with the ad or concept being tested.

• 1Freeze the wording of each core KPI and give every change a new version label.
• 2Tag every study with its sample source and device mix.
• 3Run parallel studies when a method changes, and record the calibration result.
• 4Publish norms by vintage so clients know which period a benchmark reflects.
• 5Retire studies from the active set by a written recency rule, not case by case.

Illustrative: an ad-testing agency audits its norms#

Illustrative: a fictional creative-testing agency has run ad and concept tests for packaged-goods and restaurant clients for many years. Scores live in a survey platform and a SQL warehouse, and the norms tables are rebuilt on a schedule by an analyst who inherited the process without documentation.

The CEO asks operations to tag every study with a rights status. Most master agreements include an aggregated-use clause, a smaller group are silent, and a handful of enterprise clients prohibit any use outside their own projects. The prohibited studies come out, the silent clients receive a short confirmation request, and a client-share cap is added to the threshold rules.

Several narrow restaurant cells fail the new rules and are rolled up into broader categories. The norms become slightly less granular and far easier to defend when a client's procurement team audits how its data is used.

Can norms be licensed outside the agency?#

Norms can sometimes be licensed outside the agency, but an aggregated-use clause written for client benchmarking does not automatically cover giving tables to an AI developer. Some clients would read that clause narrowly, so third-party licensing needs its own rights check.

Under the SourceX Enterprise Data Value Framework, norms rate well on domain expertise and uniqueness, and their privacy burden is usually low because they contain no respondent records. Recency and reproducibility cut the other way: old vintages are worth less, and benchmarks that anyone could rebuild from public sources add little.

How SourceX approaches a norms database#

SourceX handles a norms database through the SourceX five-step transaction: Supply, Rights, Preparation, Approval and Delivery. The Rights step reads the client clauses study by study, and nothing leaves the agency during the initial assessment, which collects metadata only.

Whatever the agency approves gets a SourceX Evidence Packet: one record of where the norms came from, which licensing rights back them, the permitted use, how privacy was handled and who authorized release. The agency keeps ownership of its norms; any license is limited to the use the agency signs off.