Evidence blocks are compact pieces of evidence that link a statement with context, evidence, source, date, limitations, and a responsible person. These evidence blocks make content transparent and help you develop citable facts from performance data, case studies, key figures, and local facts. The crucial factor is not the number of sources, but their unambiguous attribution.
In my work with owner-managed SMEs, I frequently encounter two problems: Some service pages contain sweeping claims without evidence. Other posts link to multiple sources but fail to explain which source supports which claim. An evidence block connects a specific statement with the information necessary for a professional review.
An evidence block consists of seven core elements: statement, context, evidence, source, date, boundary, and responsible person.
A source citation alone does not make a statement reliable. The context shows what the citation refers to. The measurement method explains how a result was obtained. The limitations prevent a single case from being presented as a universally valid finding.
Evidence Blocks: Blocks for verifiable content
An evidence block is neither a quote box nor an isolated statistics tile. A bibliography at the end of an article also serves a different purpose. The evidence block is positioned as close as possible to the relevant statement and answers seven questions:
- Expression: What exactly are you claiming?
- Context: For which business, time period, market or use case does this statement apply?
- Document: What data, observations, or documents support this statement?
- Source: Where does the information come from?
- Date: When was the data collected or published?
- Border: What cannot be explicitly deduced from the evidence?
- Responsible person: Who reviewed the content, calculations, and approval?
For ongoing maintenance, I'm adding three operational fields: status, review date, and next review . This makes it clear whether a document is up-to-date, needs to be reviewed again, or needs to be archived.
The difference to structured data is important: An evidence block visibly presents the subject-matter context within the content. Structured data additionally marks selected information in a standardized, machine-readable format. Both approaches can complement each other, but they are not interchangeable.
Which types of sources you should distinguish
A typical SME website uses four types of evidence. Each type of source requires appropriate wording and a clearly defined boundary.
External sources
External sources include, for example, government information, standards, scientific publications, or official documentation. First, check if a primary source is available. An industry article summarizing a study is weaker as evidence than the published study itself.
Google's guidelines for helpful and trustworthy content include clear source citations, evidence of expertise, and background information about the author or the publishing website. This is a recommendation regarding the trustworthiness of content, but not a guarantee of a specific ranking. The complete documentation is linked in Source 1.
Own measurement data
Your own measurement data can be valuable because it reflects concrete processes and results of your company. However, your own key performance indicators (KPIs) must be published along with the time period, data source, and measurement method.
"Our new website generates more inquiries" is not a reliable statement. "Comparing the twelve weeks before and after the relaunch, the number of qualified form inquiries increased from X to Y" is verifiable, provided that tracking, selection rules, and potential side effects are documented.
experience
Experience is not a measurement. Therefore, formulate it openly: "In our project practice, we frequently observe..." is more technically precise than "Companies always need...". Experience can provide guidance, but it should not be presented as representative statistics.
Estimates and assumptions
Estimates are useful when a decision needs to be made before complete data is available. Specify the assumption, the basis of the calculation, and a realistic range. A statement like "This saves 17,4 percent working time" would be falsely precise if it is based on only three informal observations.
Five evidence block patterns for different types of evidence
The following sample blocks show the structure with fictitious demonstration data . The values are not published project results from Berger+Team. Replace the sample values with verified data from your company before publishing.
1. Sample performance data
- Expression: The average processing time for recurring quote requests decreased from 46 to 29 minutes.
- Context: The study examined 38 comparable inquiries from a small B2B service company before and after the introduction of a standardized inquiry template.
- Document: Time logs from two eight-week comparison periods.
- Source: The company's own project time tracking system.
- Date: Fictitious comparison period from January to April 2025.
- Measurement Method: The active processing time was recorded from the opening of the request until the offer was ready for dispatch. Waiting times were excluded.
- Border: The result applies to this process and cannot be transferred to other companies without verification.
- Responsible person: Sales management; technical approval by the management.
- Status and inspection date: Example status "approved"; next review after six months.
2. Sample case study
- Expression: After a service page was revised, the proportion of technically relevant inquiries increased during the period under review.
- Context: Fictional, anonymized case study of a South Tyrolean specialist company with a German- and Italian-speaking target group.
- Document: 41 requests before and 47 requests after the revision were evaluated using the same internal criteria.
- Source: Example inquiry logs and CRM categories of the company.
- Date: Twelve weeks before and after the publication of the new page.
- Measurement Method: Only requests for defined services, within the served area and with a realistic project timeframe were considered technically appropriate.
- Border: Seasonality, ongoing recommendations, and other marketing measures could not be completely isolated. The case study therefore shows a temporal correlation, but not a clear causal relationship.
- Responsible person: Project management and responsible person in the company's sales department.
- Status and inspection date: Example status "approved for anonymized publication"; annual review is planned.
3. Sample Website Metrics
- Expression: 72 percent of the test subjects found the performance information they were looking for without assistance.
- Context: A clickable website prototype was tested with 25 people from the defined target group.
- Document: 18 out of 25 people solved the task within 90 seconds.
- Source: Fictional moderated user test.
- Date: Exemplary test procedure on March 14th and 15th, 2025.
- Measurement Method: All participants received the same task and the same starting point. Success and time taken were recorded.
- Border: The small sample reveals specific barriers to use, but is not a representative statement about the entire market.
- Responsible person: Person responsible for user guidance and project management.
- Status and inspection date: Example status "completed"; retest after significant changes to navigation or page structure.
If you want to interpret website metrics, the most important website KPIs for business owners can also help you . Not every number is automatically relevant for decision-making.
4. Example of a documented method
- Expression: The positioning was developed in three steps: inventory, decision workshop and validation.
- Context: This method is used in owner-managed companies with a small decision-making team.
- Document: Workshop minutes, documented decision criteria and approved positioning statement.
- Source: Own project documentation.
- Date: Version date of the respective released method documentation.
- Measurement Method: Not applicable, because the Evidence Block describes a documented process and not a statistical result.
- Border: Documented implementation proves the application of the method, but not automatically its economic success.
- Responsible person: Strategic project management and owner of the company.
- Status and inspection date: Example status “Method active”; check after every third project or in the event of a significant process change.
5. Patterns for local facts
- Expression: Berger+Team is a freelancer collective based at Reichrieglerweg 17 in 39100 Bolzano, South Tyrol.
- Context: This information describes the company's location. Berger+Team supports SMEs in South Tyrol, Italy, and the DACH region (Germany, Austria, Switzerland).
- Document: Current company master data of Berger+Team, owned by Florian Berger.
- Source: Internally released company data.
- Date: Status of this article's publication.
- Test method: Comparison of address, service area and available languages with centrally maintained company data.
- Border: The company's location says nothing about whether a service is provided exclusively on-site or also digitally.
- Responsible person: Florian Berger.
- Status and inspection date: Active; semi-annual review together with contact and performance data.
Berger+Team supports projects in German, Italian, English, French, Spanish, and Dutch. The specific language requirements are determined for each project.
Inconsistencies can quickly arise with local information: The contact page lists a new phone number, while an older post still shows the previous one. Or a service has been added but is missing from one language version. A well-maintained brand fact sheet, serving as a central source of information, reduces such discrepancies.
Before and after: a claim becomes evidence
The following example is a didactic model using fictitious numbers. The initial situation corresponds to typical situations I encounter in my work with SMEs.
"With the new website, we are receiving significantly better inquiries."
The statement leaves essential questions unanswered: What does "better" mean? What time period was compared? How many inquiries were there? Was advertising running concurrently? Without this information, the claim cannot be professionally assessed.
A more reliable version reads:
"In the twelve weeks following the relaunch, 31 out of 47 website inquiries met our predefined criteria for a suitable project request. In the twelve weeks prior, this figure was 19 out of 41 inquiries. The evaluation criteria included service type, project region, timeframe, and basic feasibility. The data originates from our CRM system. Seasonal influences and concurrent referrals could not be completely ruled out."
The second version includes the time period, sample, measurement method, source, and limitations. It does not claim a definitive cause that cannot be substantiated by the available data.
The operational process for your SME
A documentation process doesn't require a large editorial team. For small businesses, I recommend an eight-step process:
- 1. Select statement: Start with statements that are relevant to a decision. These include results, performance limits, procedures, regional availability, and specific expertise.
- 2. Clarify the statement: Replace words like "fast", "leading", "better" or "successful" with verifiable characteristics.
- 3. Check the primary source: First, find the original document, the raw data, or the responsible public authority.
- 4. Document the context: Record the time period, target group, sample, region and initial situation.
- 5. Add measurement method: Explain what was measured, counted, or evaluated, and what data was excluded.
- 6. Name the border: Explicitly state in which cases the result does not apply or which influences remain unclear.
- 7. Technical approval: Designate a responsible person who can verify the content of the statement and evidence.
- 8. Publish and maintain: Enter the inspection date, status, and next inspection date.
Don't distribute responsibility across the board to "the team." Technical specifications are reviewed by the responsible specialist, sales figures by the sales team, and key company data by management. At Berger+Team, clients speak directly with the people working on the project. The freelance collective operates without unnecessary intermediaries.
Where evidence blocks should be placed on the website
In guidebooks
Place the citation block directly after a key factual claim or at the end of the section where you explain the statement. A general list of sources is also helpful, but it does not replace the need for clear attribution.
In FAQ sections
In an FAQ, the direct answer should come first. Then add a brief explanation if the answer depends on figures, legal requirements, studies, or your own measurements. For simple definitions, a clear technical explanation is often sufficient.
On performance pages
Performance pages primarily require evidence blocks detailing the approach taken, scope of services, project experience, and measurable results. Use a suitable evidence block for each key statement, rather than repeatedly rephrasing the same claim. A geo-audit for SMEs can systematically review existing statements, sources, and attributions.
In case studies
A case study should distinguish between the initial situation, the intervention, the outcome, and the limitations. The difference between correlation and causation is particularly important: a result following a change does not automatically prove that this change alone caused the result.
Evaluate source quality
The most popular website is not automatically the best source. Check the source quality using verifiable criteria:
- Proximity to the original information: Does the statement come from a primary source or from a summary?
- Responsibility: Is the publishing organization responsible for the topic from a technical or legal perspective?
- Topicality: Is there a release date, version date, or review date?
- Transparency of methods: Are the data basis, sample and measurement method comprehensible?
- Conflicts of interest: Does the source have a financial interest in a particular outcome?
- Verifiability: Can another person access the source or control the internal data?
- Fit: Does the source actually support your statement or only a similar topic?
An official source may be outdated. A current source may be methodologically weak. An internal analysis can be relevant if its data basis and limitations are transparently documented. Therefore, source quality must be assessed individually for each statement.
Evidence blocks and machine-readable origin
The traceable origin of information is also the subject of technical standards. The W3C PROV data model describes provenance using entities, activities, and agents. Relationships such as wasAttributedTo and wasAssociatedWith can express attributions to responsible actors.
For a typical SME, a full technical implementation of PROV is usually not necessary. However, the model shows how origin, processing, and allocation can be described in a structured way. The specification is linked in Source 2.
In the Schema.org vocabulary for CreativeWork, `citation` denotes a reference to another work, ` dateModified` the date of the last modification, and `author` a person or organization listed as the creator. These properties support the machine-readable identification of reference, recency, and authorship. They do not prove that a statement is technically correct. The definitions are linked in Source 3.
The technical labeling should match the visible, verified content. Further priorities can be found in the article about structured data on SME websites.
What Evidence Blocks cannot do
Documentation blocks increase verifiability and facilitate the attribution of a statement. However, they are not a seal of approval and do not automatically guarantee trustworthiness.
- Evidence Blocks do not guarantee rankings. Google evaluates content and websites based on various factors.
- Evidence Blocks does not guarantee AI citations. Whether a system accepts a source depends, among other things, on the request, the system, and the available data.
- Evidence Blocks do not replace a primary source. The block summarizes and categorizes; the original source remains authoritative.
- Evidence blocks do not correct bad data. A precisely formatted estimate remains an estimate.
- Evidence Blocks do not replace professional or legal review. This applies particularly to health, finance, data protection and legal statements.
- Evidence blocks must not violate confidentiality. Internal performance data and data from clients require documented approval, sufficient aggregation, and, if necessary, anonymization.
If a number is not reliable, a reasoned range is better than an exact number without a sound data basis.
Checklist before publication
Check content and document
- Is the statement formulated concretely and without exaggerated generalization?
- Is it clear for which context and time period the statement applies?
- Is there verifiable evidence?
- Was the primary source used where possible?
- Are the date, sample size, and measurement method documented?
- Is the limit of the statement clearly stated?
- Are empirical values, measured values, and estimates clearly separated?
Check release and maintenance
- Has a responsible professional approved the Evidence Block?
- Is confidential or personal data protected?
- Is there an inspection date and a date for the next inspection?
- Is the supporting document directly linked to the corresponding statement?
- Do the visible content and structured data match?
Questions and answers about Evidence Blocks
How often should I update an evidence block?
Review dynamic performance data and key performance indicators at least every six months or after a significant process change. Stable business metrics can be reviewed annually, while legal, technical, or time-critical information requires more frequent review.
Who should be responsible for a receipt pad?
The person responsible should be the one who can assess the data and the professional context. The editorial team can formulate the text, but the final approval should rest with the responsible expert or management.
Can I use anonymized data from clients?
Sufficiently anonymized and approved data can be used if no conclusions can be drawn about individuals or confidential business information. Document the approval process, aggregation, and limits. If there is any uncertainty, a legal or data protection review is required.
What do I do if no primary source is available?
Use the highest quality secondary source available and indicate that the original data could not be verified. For business-critical claims, it is often better to omit the claim altogether or to phrase it more precisely.
Do I need an evidence block for every statement?
No. Evidence primarily consists of statements that influence decisions, contain figures, make a comparison, or promise exceptional performance. Clearly identified opinions and general transitions do not automatically require a full block of evidence.
How do I handle multilingual evidence blocks?
Transfer text, meaning, unit, date, and limitations correctly into each language version. The source can remain the same, while terms and contextual information must be adapted to the specific language.
Are structured data necessary for Evidence Blocks?
No. An evidence block initially functions as visible, understandable content. Structured data can additionally identify the author, modification date, and references, but must not contain any differing or exclusively invisible claims.
Are content containing evidence blocks cited more frequently by AI systems?
There is no guarantee of this. Clear statements and unambiguously attributed sources make it easier to verify the content. Whether a system cites a website depends on its own procedures and the available database.
How do I deal with outdated documents?
Mark the evidence block as requiring review and remove or qualify the statement if its recency is crucial. An older date can remain visible as long as the historical context is clearly stated and no current validity is suggested.
Conclusion: Evidence needs structure and accountability
Evidence Blocks transform sweeping claims into verifiable statements. For SMEs, the benefit lies in a clear process: select a statement, verify the source, document the context and measurement method, define the threshold, obtain expert approval, and monitor regularly.
Start with five to ten statements that are relevant to potential customers' decision-making. Well-documented and maintained statements form a reliable knowledge base for people, search engines, and digital assistance systems.
