Any business can say it works carefully. A traceable process, an approved work example or a clearly explained analysis shows what that means. Your most valuable content often starts where your team solves questions every day.
1 — Original means contributing something
Google recommends helpful content with original information, research or analysis and demonstrable experience. Our interpretation: publish an answer whose basis you can explain. A third-party statistic in new colours remains a third-party statistic. Your contribution can be a documented application to a specific customer question. [1]
| Material from your work | Possible content | Check before publication |
|---|---|---|
| Recurring customer questions | Decision guide with explained options | Aggregate cases; remove confidential details |
| Acceptance record | Explanation of quality criteria | Permission and scope |
| Your measurement series | Table with methods and limitations | Completeness, units and comparability |
| Practical expert test | Procedure, conditions and observation | Repeatability and alternative explanations |
| Resolved defect | Finding → correction → verification | No unsupported general success claim |
2 — Give every metric a data record
Before designing a number, write down its denominator. What was counted? During which period? Which cases are missing? Who checked the data? Without these answers a chart can be attractive and misleading. A small analysis can still be useful when its limited scope stays visible.
| Field | Example of a useful entry |
|---|---|
| Question | How long until the first human response? |
| Population | All accepted enquiries in a defined period |
| Measurement rule | Business hours from receipt to first human response |
| Exclusions | Report spam, tests and duplicates separately |
| Privacy and permission | Only an approved, non-identifying aggregate |
| Version | Analysis date, responsible person and reason for change |
3 — Worked example: why the mean is not enough
Fictional teaching example, not GlasBox or client data: six enquiries received their first human response after 1, 2, 2, 3, 4 and 12 business hours. Their total is 24 hours. The mean is 24 ÷ 6 = 4 hours; the median lies between the third and fourth values, at 2.5 hours.
| Metric | Result | Interpretation |
|---|---|---|
| Count | 6 | Very small fictional sample |
| Mean | 4 h | The 12 h observation strongly affects the average |
| Median | 2.5 h | Central position; the long wait still matters |
| Range | 1–12 h | Shows how different individual cases are |
None of these metrics establishes improvement: there is no comparable earlier measurement. Even a before/after comparison would need to account for case mix, business hours and measurement rules. Publish a narrowly accurate statement instead of an impressive percentage without support.
4 — Show experience without exposing customers
A work example does not require a customer secret. Explain the starting situation, decision, professional reasoning and acceptance check. Label a recreated example accordingly. Before using real photos, screenshots or records, check permission, personal details and confidential information. Redaction only helps when the remaining details do not allow identification.
A subject expert should review the content and be identifiable as accountable for it. A generated name or invented qualification is not evidence of trustworthiness. AI can help organise and draft. A person remains responsible for the claims, data and approval.
5 — Update when the substance changes
Google describes sitemap lastmod as the date of a significant change; updating a copyright year does not qualify. Visible publication and modification dates should accurately describe the article. Preserve a traceable original publication date and explain material corrections. [2][3]
| Trigger | Action | What stays intact |
|---|---|---|
| Service or pricing logic changed | Update affected claims and examples | Unaffected data |
| New measurement period complete | Add a table with matching methods | Label the earlier version for comparison |
| Source no longer accessible | Verify, replace or narrow the claim | Never invent a replacement source |
| Spelling correction only | Correct the error | Do not present it as a new study |
6 — Your first original evidence-based article
Choose a recurring customer question. Use existing approved material and write a short answer with one concrete example. Add methods, limitations, an accountable person and the next review trigger. Quality lies in a traceable claim, not the largest possible dataset.
Connections within the course
| Part | What you build on here |
|---|---|
| Part 1 · Trust | Your own work makes claims testable. |
| Part 6 · Measurement | Denominators and comparability apply to your own metrics. |
| Part 7 · Content | Evidence improves a specific existing answer. |
| Part 9 · Technical access | The table and methods must be accessible as text. |
Clarify your next step.
SOURCE/01 connects specific findings with evidence and a prioritised task.
Explore SOURCE/01Sources and working method
Checked on 25 October 2026. Response-time numbers and chart are wholly fictional teaching data. They describe neither a business nor a market. Editorial workflows are original recommendations.
No exact document date stated. · Checked: 25 October 2026.
No exact document date stated. · Checked: 25 October 2026.
Document date: 2025-12-10 · Checked: 25 October 2026.
Glossary: terms and connections
| Term / translation | Meaning | Connection |
|---|---|---|
| First-party data / eigene Daten | Information collected through your own work. | Part 6: define measurement rules first. |
| Provenance / Herkunft | Traceable origin of evidence. | Part 5: connect source and claim. |
| Sample / Stichprobe | The cases actually examined. | Part 6: limit the scope of inference. |
| Denominator / Nenner | The base of a proportion. | Part 6: no percentage without a base. |
| Mean / Mittelwert | Sum of values divided by their count. | Here: 24 ÷ 6 = 4 h. |
| Median / Zentralwert | The middle position in ordered values. | Here: (2 + 3) ÷ 2 = 2.5 h. |
| Review trigger / Prüfanlass | An event prompting another review. | Part 7: maintain quality after publication. |
| Lastmod / Änderungsdatum | Date of a significant page change in the sitemap. | Part 9: technical declarations must be accurate. |