Your Data Can Tell a Story Without Exposing Anyone: A Privacy Protocol for Digital PR
A practical protocol for reviewing permissions, minimizing data, assessing re-identification risks, and documenting approvals before publishing a digital PR campaign based on first-party data.
Why first-party data can add value and create risks
Internal data can help build an original digital PR story: it reflects activity, preferences, or changes observed in a specific context. But having access to data does not mean it can be freely reused. It may include personal information, reveal trade secrets, or be subject to commitments to customers, suppliers, or employees. Even a dataset without names may make it possible to identify someone when combined with public sources.
- Editorial value should come from a finding that can be explained and put into context, not from exposing individual records.
- Authorization to collect or use data for one purpose does not automatically mean it can be published for another.
- Privacy must be assessed across the dataset and its context: what it contains, who might access it, and what additional information is available.
Define the purpose and review permissions before selecting data
Start by writing one sentence that states what question the campaign aims to answer and how the results will be used. Then identify where each data point comes from, who collected it, for what purpose, and what terms of use apply. Review privacy notices, contracts, consents, confidentiality agreements, and internal policies. If the data is personal, the legal basis and compatibility of the new use must be reviewed under applicable regulations; simply having the data in the company's possession is not enough.
- Note the source of each dataset, the date or period it covers, and the person responsible for safeguarding it.
- Check whether external communications were an intended use and whether there are limits, notice obligations, or contractual restrictions.
- If the data belongs to a client or third party, request explicit authorization for the specific use and define which version may be shared.
- If there are questions about personal data, consult the data protection officer or legal team before proceeding.
Apply data minimization: keep only what you need
Once the purpose is clear, limit the analysis to the variables needed to answer the question. Avoid exporting an entire database for convenience if a limited selection or aggregated results will suffice. Also reduce the level of temporal and geographic detail when it adds no value. The less information in circulation, the fewer opportunities for accidental exposure and the easier the process is to review.
- Separate the variables needed for analysis from those that are merely available in the source system.
- Remove direct identifiers and unnecessary fields as early as possible; restrict access to working data.
- Set a retention period and a process for deleting temporary copies when they are no longer needed.
- Do not include test records, internal notes, or free-text fields that could contain unexpected information.
Aggregate, anonymize, and check the risk of re-identification
Aggregation can reduce exposure, but it does not by itself make every dataset anonymous. Removing names or replacing them with codes is usually pseudonymization: the data may still be linked to a person if a key exists or if it is combined with other sources. To assess a publication, consider the attributes shown, the size and uniqueness of groups, geographic or temporal precision, and the information a third party could find.
- Group or generalize variables when detail is unnecessary; for example, use broader time periods or less precise areas.
- Review small cells, rare categories, and unique combinations that could point to a person or company.
- Check whether published results could be cross-referenced with news, public profiles, directories, or other available data.
- Do not adopt a universal group-size threshold: risk depends on the context, sensitivity, and sources available for comparison.
- Consider additional protection techniques with specialists if the dataset is sensitive or publication could have significant consequences.
Separate publishable findings from confidential information
The campaign asset should not be a copy of the working database. Prepare a separate output containing only approved results and the context needed to interpret them. Keep personal data, trade secrets, internal observations, and any detail that could allow records to be reconstructed out of it. Review the text as well as charts, tables, metadata, file names, and document notes: sensitive information can appear in less obvious places than the main body.
- Create a separate publishable version with restricted access permissions.
- Check that tables and charts do not allow hidden values to be inferred through totals, filters, or comparisons between versions.
- Remove comments, change history, hidden sheets, metadata, and source files that should not be shared.
- Agree on which materials journalists, agencies, collaborators, or suppliers may receive, and under what conditions.
Document sources, methodology, limitations, and responsibilities
A data-based story needs an explanation that makes it possible to understand where the results come from and what they mean. Record the sources, period analyzed, inclusion and exclusion criteria, transformations made, and known limitations. Clearly distinguish observation from interpretation: an association does not, by itself, demonstrate causation. The documentation should also state who reviewed the data processing, who approved publication, and which version was authorized.
- Keep a record of data provenance and methodology without duplicating unnecessary personal information.
- Explain gaps, coverage biases, changes in data collection, and factors that prevent the results from being generalized.
- Maintain a record of decisions, reviews, dates, and versions of the publishable asset.
- Make sure claims and visualizations do not suggest greater precision or representativeness than the data supports.
Pre-sharing checklist for the asset
Before sending materials to the media, collaborators, or suppliers, carry out a final review that brings together privacy, quality, and version control. Approval should not be a generic formality: each responsible person needs to know what they are reviewing and which risks they are expected to assess. If the dataset, analysis, or content changes, check again that the approved version is still the one to be distributed.
- Purpose defined and compatible with applicable permissions, notices, and agreements.
- Source, period, and person responsible for each source documented.
- Variables limited to what is necessary; access and copies controlled.
- Direct identifiers, unique combinations, and possible cross-references with public information assessed.
- Charts, tables, metadata, and files reviewed; publishable version kept separate from confidential material.
- Methodology, limitations, and explanatory text checked with the responsible people.
- Approval recorded by privacy or legal counsel where appropriate, as well as by the data and communications teams.
- Recipients, distribution channel, terms of use, and retention plan defined.
Frequently asked questions
Is removing names enough to anonymize data?+
Not necessarily. Other attributes, such as a precise location, a date, or an uncommon combination, may make it possible to identify someone when cross-referenced with available information. Replacing names with codes is usually pseudonymization, not anonymization. The assessment should consider the context and the risk of re-identification.
Can I use data collected for another purpose in a digital PR campaign?+
Do not assume that you can. Review the original purpose, privacy notices, the legal basis where personal data is involved, and applicable contracts or commitments. If the new use is unclear, consult privacy or legal counsel before preparing a publication.
When should I abandon a data-based story?+
When there is insufficient authorization, the risk of identification or harm cannot reasonably be reduced, restrictions prevent its use, or data quality is not good enough to support the conclusions. You can also reframe the question, reduce the level of detail, or work with more aggregated results.
What documentation should I keep?+
Record the data provenance, purpose, permissions reviewed, methodology, limitations, risk-reduction measures, approvals, and the final version shared. Keep only the documentation needed and follow the organization's access and retention policies.
Does this protocol guarantee legal compliance?+
No. It is a working guide for identifying and managing risks, not a legal opinion. Applicable regulations depend on the type of data, the purpose, the people affected, and the context. If personal or sensitive data is being processed, obtain an appropriate review before publication.
Sources and references
- Google Search Essentials — Google Search Central
- Spam policies for Google web search — Google Search Central
- Qualify outbound links — Google Search Central