Insights / Growth, Search & Measurement · · Updated · 11 min read
Analytics without surveillance: privacy-respecting measurement for a group of companies
A practical guide to limited analytics: separate company reporting, check automatic events and identifiers, and distinguish verified collection from privacy goals.
Analytics setups can grow by accretion: an extra tag, a trial session recorder or a script added by a plugin. A useful review starts by listing what actually collects data, including options enabled in the analytics interface.
Analytics without surveillance is a design goal: measure what a company needs to decide while limiting collection and avoiding unnecessary links between people and brands. This guide proposes practical boundaries; it is not an audit or a claim that every Oryvelon company has implemented them. The group-site section below separates inspected behaviour from recommendations.
For a group of independent companies, separate customer relationships deserve separate measurement decisions. Shared ownership alone is not a reason to combine an individual's activity across unrelated products.
Start from decisions, not from data
The most useful question in analytics design is not "what can we track?" but "what will we decide with it?"
A small company can begin with a short list of recurring decisions:
- Which channels are worth more time or budget?
- Where do people drop out of the path to signing up, buying or booking?
- Do customers come back, renew or buy again?
- Is each product's cost per useful outcome acceptable?
- Did a change we made improve things, make them worse or do nothing?
Counts, rates, trends and cohorts can support these decisions without requiring staff to inspect a named visitor's journey. However, the tools used to calculate those aggregates may still collect identifiers. Review both the report and the data behind it.
Put a short measurement plan in the company's source-of-truth document: the decisions, the metrics that inform them and the events needed to produce those metrics. Before adding an event, identify which decision it serves and which fields it needs.
This has a practical benefit besides privacy. Analytics that nobody uses still costs attention, page weight and review time. A small, deliberate set of events is easier to keep accurate.
One property per company
A separate analytics property for each independent business is one useful reporting boundary. The oryvelon.com property was inspected for this review; the configuration of every company property has not been verified here.
Separate properties alone do not establish that identifiers, advertising integrations, exports or customer records cannot be joined. Check those mechanisms individually before making a public no-tracking claim.
There are three practical reasons to consider separate company reporting.
It reflects distinct business questions. A commerce checkout, an education enquiry and an entertainment feature need different outcome definitions. Combining their conversion rates can conceal those differences.
It supports clear ownership. Document the property owner, access, exports and retention responsibilities for each company. Separation may make a future handover easier, but a property boundary alone does not guarantee a complete or uncontaminated history. See Designing every company so it could stand alone.
It reduces accidental pooling; it does not make tracking impossible. Google's cross-domain measurement documentation describes how identifiers can connect activity across domains. Also inspect user IDs, integrations and exports. The design principle of shared infrastructure, separate data needs technical checks and access rules to hold.
Companies can share a measurement-plan format, event definitions and a UTM convention without pooling person-level records. Document the standards actually adopted; do not infer adoption from a group-level guide.
Aggregate metrics first
The following are hypothetical measurement examples associated with different business models. They are not reported results, verified implementations or confirmation that a named company currently offers every feature listed.
| Company | Aggregate metrics that drive decisions |
|---|---|
| MerchNivo | Trial starts by channel, trial-to-paid rate, weekly active stores, briefings opened |
| KeşifAtlası | Free tests started and completed, paid report conversion, questions routed to people |
| ZodiVela | Readings completed, credit purchases, subscription renewals |
| Noveniq | Sessions by channel, add-to-cart rate, checkout completion, repeat purchase rate |
| Sinem Keser Beauty Academy | Course page views, enquiries, bookings, digital product sales |
| oryvelon.com | Visits by source, outbound clicks to each company, contact enquiries |
These examples describe counts and rates. An aggregate can still reveal information about a person when its group is small or it is combined with other data. Avoid presenting aggregation alone as an anonymity guarantee.
For retention questions, a cohort can show how many accounts from a starting period remain active later. Restrict the output to an appropriate group size and review how membership is calculated: a cohort report may still rely on records linked over time.
For a logged-in product, accepted orders or bookings can provide a reference for business outcomes. Reconcile cancellations, refunds, duplicates and missing records before treating totals as complete. Use verified operational records where appropriate, while limiting access and purpose; see verified data decides.
What the group site measures
A review on September 29, 2026 checked the group site's public analytics script and available event report. These observations apply to the inspected group site, not every company domain.
The public script defines these collection behaviours:
- a GA4 configuration for oryvelon.com and its subdomains; this does not establish that owner or test visits on the live domain are excluded;
- a
company_outboundevent for external HTTP links matching the script's condition, with destination query strings removed and link text included. Its current scope is broader than company links; - a
contact_email_clickevent for mail links, including the linked destination address; - a
generate_leadevent when the thank-you page path loads. This trigger alone does not prove an accepted enquiry, prevent repeat counts or establish lead quality.
The event report also contained scroll, so the previous statement that there was no scroll tracking was incorrect. Google documents an enhanced-measurement scroll event when 90% vertical depth first becomes visible. Interface-controlled collection can extend beyond a site's custom script; inspect both before describing the complete setup.
The selected report did not show company_outbound. A definition in source code is not proof of successful delivery, and an absent report row is not proof of no outbound interest. A recorded click also does not establish that the destination loaded or recorded a session. Use the checklist on outbound clicks, destination sessions and leads to keep those observations separate.
Consent, per company and per audience
Assess collection and consent requirements for each audience, tool and purpose. The following are design and verification recommendations, not a statement that every company's configuration has been audited or meets every applicable requirement.
A review should cover these elements.
Document the scope. Record the audiences, tools, purposes and consent decisions for the particular site. Obtain appropriate advice where needed; copying another company's banner does not establish that its choices fit this setup.
Test the actual collection. Inspect what is sent before a choice, after acceptance and after refusal. Google explains that its default web implementation uses a client ID in a first-party cookie, while deactivating analytics storage through Consent Mode prevents that client ID from being stored. See Google's data-collection documentation. This setting alone is not a complete privacy assessment.
Make the choices clear. Provide understandable options and test refusal as carefully as acceptance. Check that the interface and the behaviour of the tags agree.
Keep purposes separate. Do not treat agreement to product terms as an automatic instruction to subscribe to marketing or enable unrelated analytics. Record each purpose and the applicable decision. See Marketing consent vs product consent.
Check remembered choices. Test returning visits and withdrawal, and avoid repeated prompts whose purpose is to wear down a visitor's refusal.
No cross-brand user tracking
This deserves its own section, because it is the easiest line to cross in a group of companies and the most damaging if crossed.
Combining behaviour across companies may appear useful for marketing, but it changes the measurement boundary. Make that decision explicit and inspect the mechanisms that could connect records.
For a design that avoids cross-brand user tracking, apply and verify these boundaries:
- No shared identifiers. Avoid a common user ID, cookie or device identifier across unrelated company analytics.
- No cross-domain linking between unrelated brands. Check tag and domain settings for unintended session stitching.
- No shared advertising audiences. Do not repurpose one company's customer records for another company's advertising within this design.
- Separate advertising integrations. Audit tag destinations and account links rather than assuming different domains imply different recipients.
- No person-level data joins. Review exports and downstream tools as well as browser tags for cross-company matching.
Apply the same review to shared AI infrastructure: identify what each request contains, where it is logged and who can link records. The architectural discussions in Building an AI gateway and Why we don't merge products into one app are useful starting points, not evidence of a completed audit.
For a portfolio view, ask each company for defined aggregate totals instead of combining individual histories. Record counting units and periods so the totals remain interpretable.
What never goes into analytics
For ordinary GA4 event measurement, follow Google's guidance on avoiding personally identifiable information. Review URLs, titles, event fields and user-entered text; a privacy goal is not proof that those surfaces are clean.
- Personal details in event parameters. Exclude visitor emails, names, phone numbers, addresses, national identifiers and free-text form contents from ordinary analytics events.
- Personal details in URLs. Remove sensitive values from page paths, parameters and titles before measurement. Check password-reset links and pre-filled forms in particular.
- Sensitive categories. Do not attach nationality, eligibility answers or individual outcomes to a person in ordinary analytics events. Review whether even a test-type label could disclose more than the decision needs.
- Children's data. Education products need a specific review of collection, recipients and reporting. Do not describe a student-facing implementation as safe solely because its dashboard displays aggregates.
- Marketplace contact data. Keep contact records within the product's controlled systems, and avoid putting them in analytics events or exposing individual viewing histories.
A periodic check looks for accidental leaks: someone adds a form field and the tag manager captures it, or a URL parameter starts carrying an email. These are much easier to catch in a small, well-documented setup than in a sprawling one.
Retention and access
Analytics data is still data, so it follows the same basics as everything else.
Choose retention for each data type and decision, and verify the actual setting. Aggregated reports, event-level records and exported copies may have different lifetimes; document deletion responsibilities for each rather than treating one tool setting as universal.
Apply least privilege to properties and exports: identify the owner, give access for a defined task, use strong authentication and review continuing need. A quarterly tool review can provide a regular checkpoint. See also Least-privilege access for small teams. These are recommended controls, not verified access claims about every company.
A worked example: measuring a funnel change
Hypothetical example: a guide website tests a clearer explanation of an optional paid report at the end of a free assessment. The new wording explains the report's scope and limitations. The team wants to know whether it improves useful enquiries or purchases; this is not a claim about a launched Oryvelon product or an actual experiment.
A compact evaluation could use aggregate outputs:
- Define the funnel in events.
test_started,test_completed,report_viewed,report_purchased. No event carries the person's nationality, answers or result — only the fact that a step happened and, where useful, which test type it was. - Compare periods or variants. Either run the old and new versions side by side with a random split, or compare a stable period before and after. The split is cleaner; the before-and-after comparison is simpler for a small company.
- Read the rates, not the individuals. Completion-to-purchase rate before and after, with the number of completed tests alongside so nobody over-reads a small sample.
- Check the side effects. Did refund requests or questions routed to people change? Those numbers live in the product and support systems, counted in aggregate.
The reviewer can compare rates without watching a named person's activity. The collection used to produce those rates still needs the boundaries described above. Similar evaluation designs can apply to a trial, checkout or booking flow when that flow exists.
Reading aggregate numbers well
Aggregate data has its own traps, and a privacy-respecting setup does not excuse sloppy interpretation.
Small samples need context. In a hypothetical example, a count rising from 12 to 15 is mathematically 25% higher, but that alone does not demonstrate an improvement caused by a change. Show counts and denominators, comparable periods and relevant uncertainty; there is no universal number of weeks that turns a difference into proof.
Missing data can change the sample. Visitors who are not measured may differ from those who are. Reconcile analytics with operational totals where appropriate, and check those totals for cancellations, duplicates and missing records rather than assuming they are complete.
Small groups can expose people. A segment such as one school or a single course date may become identifying. Set reporting boundaries appropriate to the data and avoid unnecessary breakdowns; a fixed minimum count alone is not an anonymity guarantee.
Compare consistently. Use completed periods of equal length and check changes in collection or traffic mix. A rolling window can help describe a trend, but smoothing does not establish its cause.
Trade-offs to consider
Being honest about analytics means being honest about what this approach costs.
Limited attribution. A restricted measurement design leaves some journeys unobserved. State those limits when interpreting channel reports instead of presenting an exact explanation for every outcome.
Fewer advertising signals. Limiting collection can reduce the information available to advertising tools. Evaluate that trade-off against the actual purpose and applicable choices, without assuming that more collection always improves results.
No person-level cross-selling within this design. Keep campaigns for another company independent of an individual's activity with an unrelated brand; public content can introduce a company without joining customer histories.
More upfront thinking. Deciding what to measure before building takes longer than installing every tag available. It saves time later.
A practical checklist
- Write down the decisions each company makes regularly, then the metrics and events they need. Collect nothing else.
- Create one analytics property per company and per domain; do not create a group-wide roll-up that tracks people.
- Use the same event naming style and UTM convention across companies so aggregates compare cleanly.
- Configure consent per company and audience; make declining as easy as accepting.
- Keep personal data out of event parameters and URLs.
- Set retention to what decisions need, and review access quarterly with two-factor authentication required.
- Report across the group with aggregate totals, never merged user data.
Summary
Useful analytics begins with a decision, a limited collection plan and evidence that the configuration behaves as described. Separate properties and aggregate reports are helpful boundaries, not guarantees of anonymity or an inability to join records. Check automatic events, identifiers, integrations, access and retention. The inspected group-site report included scroll events; a public description must reflect that observation. Keep proposed standards distinct from verified implementation.
Questions and answers
What does analytics without surveillance mean?
It is a design goal: collect only what useful business decisions require and avoid unnecessary person-level tracking. Aggregate reports alone do not establish that the underlying collection is identifier-free.
Do separate analytics properties prevent cross-company tracking?
No. Separate properties provide a reporting boundary, but identifiers, integrations and exports need their own checks. This guide does not establish that every company configuration has been audited.
Can privacy-respecting analytics still support growth decisions?
Yes. Aggregate counts, rates and cohorts can inform channel, funnel and retention decisions. Check collection limits, small samples and outcome definitions before drawing conclusions.