Tag: Documentation
All blog posts with this tag.

- Nov 16, 2026
Marketing site, blog, docs and SaaS app: what to measure separately
A B2B SaaS business operates a marketing website, a blog, a documentation hub and an authenticated product. It asks for "one dashboard for everything." The request sounds sensible but bundles questions with different answers. Pricing pages help prospects evaluate an offer. Documentation helps existing users solve problems. The application is where customers do their work. Blog articles may create awareness without producing immediate sales requests. Adding all pageviews together is rarely enough to guide those decisions. The goal should be consistent team reporting, adapted metrics for each surface and no unnecessary individual tracking across properties. A fictional portfolio makes the problem concrete Consider "North Workshop", an illustrative B2B SaaS company serving European SMBs. It runs four properties:a commercial site with product, pricing and demo-request pages; a blog publishing guides and industry analysis; a documentation hub explaining setup and product use; an authenticated application where customers carry out tasks.Marketing, engineering and support share reporting responsibilities. There is no dedicated business-intelligence department. The system therefore needs to answer ordinary questions without becoming a data infrastructure project. This company is fictional. The following model is design guidance, not a claim about any current vendor's capabilities. Marketing: understand acquisition and accepted demand The primary question is whether relevant visitors reach relevant pages and can complete intended actions. Useful starting indicators include entry pages, detectable sources, important page consultations and requests actually accepted by the application. A pricing page with many views is not automatically effective if visitors misunderstand the offer. A low-volume page may yield a few well-qualified accounts. Metrics must connect to questions, not exist just because they are easy to capture. State the denominator of each conversion rate. Accepted inquiries divided by visits is different from accepted inquiries divided by pricing-page views. Our acquisition page guide provides a practical framework. Lead qualification belongs primarily in the business process. Raw form contents need not be copied into audience analytics to count completed requests. Blog: measure discovery and understanding A B2B blog should not be judged only by immediate conversions. An educational article can help a technical decision-maker long before a commercial conversation. Review entry pages, content trends, detectable discovery sources and carefully selected actions such as a resource download. Google Search Console provides a separate perspective on Google Search impressions, clicks and queries under its own definitions. Search impressions are not website visits. Do not merge the two into one unlabeled traffic count. A useful editorial decision might be updating a high-value article, adding a missing answer raised by support, or improving links between related guides. Documentation: support problem-solving Documentation serves a different purpose. Teams want to know which articles are used, which explanations need improvement and whether readers can find useful instructions. A troubleshooting page may receive more traffic because it ranks well or because a product defect has spread. The direction of change is not intrinsically good or bad. Interpret it with support tickets and release information. Documentation also attracts crawlers and AI agents. Automated requests should not automatically be treated as human reading sessions. Avoid inventing a "resolution rate" from pageviews alone. Reliable resolution measurement requires a meaningful signal, such as explicit feedback or a carefully defined service outcome. SaaS application: adoption, retention and operations Inside an authenticated product, the questions change. Do customer organisations activate important features? Does useful work recur? Which operations fail? These often require product analytics, operational diagnostics and business events beyond ordinary public-website pageviews. PostHog, for example, documents a web analytics view centred on traffic and a product analytics view focused on events, adoption, funnels and retention. That distinction does not prove that two separate software tools are always required. It shows that the questions and data models are different. Teams should explicitly separate service operations, product usage and public audience measurement. A single indistinguishable event stream can easily become more invasive than necessary. A four-surface measurement mapSurface Main decision Starting metrics Avoid in audience reportingMarketing Improve acquisition and inquiries Entry pages, sources, accepted requests Free-form contact fieldsBlog Prioritise content and revisions Article entrances, trends, sources Individual reading profilesDocumentation Improve help resources Frequently used pages, justified outcomes User secrets and personal queriesProduct Understand activation and usage Defined business events, adoption, incidents Blanket copies of application payloadsThe map is a contract between owners, not a vendor-feature checklist. Centralised dashboards are not cross-site identity tracking Multi-site teams often need property-level comparisons, common date ranges and shared reporting. Those goals can be met with aggregate metrics without identifying the same person across every property. Linking individual activity across sites raises separate questions about purpose, lawful basis and transparency. The CNIL's narrow conditions for audience-measurement exemptions should not be extended by assumption to cross-site identity linking. Keep reporting centralisation distinct from personal behaviour linkage. The former can strengthen governance without requiring the latter. A shared metrics dictionary Harmonise concepts that can genuinely be compared: timezone, reporting period, entry page, observable source and selected event definitions. Conversions are often surface-specific. A documentation download, a marketing demo request and first successful product use cannot responsibly become an interchangeable "conversion" metric without further context. Document for each measure its source, definition, unit, calculation, filters, owner and limitations. Our multi-site dashboard guide describes how to organise portfolio and property views. Assign collection and access owners Who may add a script to documentation? Who approves marketing conversion events? Who can export product data? These are governance decisions as much as reporting questions. Assign an owner to each surface and a shared owner for collection principles. Review access when staff or contractors change. The CNIL recommends least-privilege permissions and regular reviews. A lightweight process can work: write the purpose, choose necessary fields, test with synthetic input, update the dictionary and review periodically. One tool or several? One product may fit when permission models, reporting definitions and required analytical depth are compatible. Several products may make sense when purposes, sensitivity or event types differ materially. Avoid multiplying tools for fashion. Also avoid forcing everything into one platform just because a single dashboard looks convenient. Compare total administration, maintenance, data quality, export and training costs. The measurement map should precede vendor selection. Where Pomelo fits Pomelo focuses on privacy-first audience measurement intended to remain readable for teams managing several properties. That makes its position relevant to public-facing sites and shared web reporting. Web reporting does not necessarily replace product analytics, business intelligence or application diagnostics. Authenticated product usage may require complementary instrumentation, depending on verified needs and available functionality. A practical ownership ritual for four surfaces A short quarterly meeting can keep the measurement map useful. Ask each surface owner to bring one decision made from reporting, one metric whose definition caused confusion, and one collection operation that may no longer be necessary. The marketing owner might show that a pricing-page redesign changed the number of accepted inquiries. The editorial owner might identify a guide that needs updating, supported by Search Console and observed entrances. Support might explain why documentation traffic increased after a release. The product team might report that a feature's activation event was redefined, making historical comparisons less straightforward. Do not force those stories into a single performance score. Record the distinct decisions and the evidence limits, then identify any shared reporting changes that would genuinely benefit the whole organisation. Governance also requires someone to challenge a proposed new event. Before it is added, the requesting team should explain the decision it will support, the minimum fields needed and what happens if the event is never collected. This simple exercise can prevent an expanding set of unused identifiers and arbitrary dimensions. The result is a portfolio report that respects differences among surfaces. A shared language becomes possible without demanding that each website or application adopt identical instrumentation. Conclusion: map decisions before dashboards The best system is not the one that puts everything on the same chart. It is the one that assigns every surface a useful decision, defensible metrics and accountable owners. Start with a four-surface map. Harmonise what deserves comparison, separate different purposes and resist equating centralised reporting with cross-site personal tracking. FAQ Can all four surfaces use one analytics product? Yes, where definitions, permissions and analytical depth fit. Reporting goals still need to differ by surface. Must users be identified across blog and product? Not for most traffic and content decisions. Any individual linking requires a distinct justification. Does Search Console replace blog analytics? No. It reports Google Search visibility and interactions, not all website use. How can documentation success be measured? Start with content demand and support questions. Add a credible resolution signal only if it can be defined and tested. Pageviews alone do not prove resolution. SourcesPostHog, web analytics dashboard PostHog, product analytics PostHog, retention analytics Google Search Central, getting started with Search Console CNIL, audience-measurement tool conditions CNIL, managing permissions
Just the right measure, for all your websites.
Pomelo is opening gradually. Join the first teams measuring right and help shape what comes next.
