NOGL / Sources and measurement
Match the evidence to the question.
A catalog observation, a traffic estimate and a measured effect answer different questions. Choose the source before drawing a conclusion.
What do you need to know?
Choose a question to explore its evidence requirements. These guide examples do not run an analysis or connect an account.
- Suitable data
- A dated observation of the public catalog, with the URL, market, currency and collection method.
- What it can establish
- The products, displayed prices and availability visible in that observation.
- What remains uncertain
- Sales, launch dates, traffic and total demand remain unknown. A visible listing is not proof of a sale.
Four sources, four different promises.
This is a source-selection guide. Availability, licenses and account access must be confirmed for a connected pilot.
| Source | Coverage and permission | Limits to retain |
|---|---|---|
| Public catalog observation | Visible fields at a recorded time; check source terms and collection scope. | Partial catalog coverage, region and changing prices. No private sales or traffic. |
| Authorized first-party analytics | The customer’s agreed account, dates and measured events. | Consent and instrumentation gaps; definitions and attribution rules. |
| Licensed benchmark or estimate | The provider’s sample, market, methodology and permitted uses. | Sampling uncertainty and model assumptions. Estimates are not observed account totals. |
| Experiment or model | Authorized outcome data and a documented design or set of assumptions. | Experiments require valid comparisons; model estimates remain conditional unless causal identification is justified. |
A range question, followed to its limit.
Illustrative decision path · no live results
- 01
Question
Which visible zip layers should a buyer inspect?
- 02
Source choice
Record the selected public category and its observation time.
- 03
Observed fields
Keep supplied title, displayed price, variant and availability, where present.
- 04
Still unknown
Do not turn missing sales, traffic or launch dates into a demand score.
- 05
Review decision
Shortlist records for a buyer’s review; request appropriate evidence before making a purchase decision.
Check comparability before comparing.
- Use the same timeframe and record when each source was observed.
- Check market, currency, taxes and units before comparing prices or totals.
- Agree event, channel and product definitions with the source owner.
- Record missing coverage, sampling and measurement uncertainty.
- Name the reviewer and the additional evidence needed for the decision.
Take the next step for your question.
Is an example an observation?
No. An example explains a workflow. An observation records what a named source showed at a specific time.
Can a snapshot reveal a launch date?
A first observation only establishes when you saw the record. It does not establish when the product launched.
Is a model estimate a measured effect?
Not by itself. A scenario depends on inputs and assumptions. A causal claim needs an appropriate design, identification and uncertainty assessment.
Can a traffic estimate establish conversions?
No. Conversions require suitable outcome evidence and definitions. Estimated visits alone cannot establish purchases or revenue.