Research Guides
Inventory Search Appearance Before Breaking It Down
When a Search Console report lacks an expected search appearance, first check how the report was built. An export that started with a guessed feature name answers a different question from an inventory of the appearance types Google returned for the property and dates.
This worksheet uses fictional figures and does not query a real property. It checks the report's scope, not whether a site is eligible for a feature.
1. First record the appearance types actually returned
Google's Search Console API guide documents a two-step workflow. First request data grouped exclusively by the search appearance dimension. Then use a type found in that response as a filter for the breakdown you need, such as device or page. Save the exact returned identifier rather than translating a display label into a guessed API value.
Record the property, dates and search type with this inventory. A type observed for one property or period is not a promise that it will also appear in another. Treat the initial list as what the query returned, not as a registry of every feature Google supports.
2. Then answer a question for one observed type
Imagine a fictional editor asks, “Which devices account for impressions in this particular appearance type?” The inventory returned a relevant type. In the second request, keep the property, time window and search type fixed, filter by that observed identifier, and group by device.
Suppose the result contains 36 mobile and 14 desktop impressions, with no tablet row. The two returned device rows total 50 impressions. The defensible note describes those returned rows and flags that tablet was not returned. It does not state that nobody on a tablet saw such a result unless separate evidence establishes that conclusion.
If the editor instead needs a page breakdown, make a separately labeled filtered query. Do not splice its rows into the device table or assume a detail export must reproduce every count in a broader view.
3. Keep a receipt for both steps
- Inventory: property, search type, window and exact returned appearance identifiers.
- Detail: selected identifier, grouping dimensions and any additional filters.
- Coverage: response aggregation, row limits, pagination and omissions that could affect interpretation.
- Output: the actual rows used and the statement they support.
- Repeat: the same request definitions for the next comparison window.
Keep zero-valued rows, an empty returned result and failed requests distinguishable. A failed request did not measure anything. An empty result by itself does not explain whether the underlying appearance had no reportable observations or whether the comparison conditions changed.
4. Verify the intended comparison before publishing
When comparing two periods, repeat the inventory and keep each returned list. Confirm that the filtered type is the one you intended and that the dates, search type and grouping choices are comparable. If it is not returned in one period, label that condition rather than substituting a different feature to complete a chart.
Do not sum different appearance categories as a replacement for an independently scoped overall performance total. A feature-specific breakdown and a property-wide traffic statement answer different questions.
5. State which slice the numbers describe
A reproducible note for the fictional example reads: “After identifying the type in the appearance inventory, we filtered to that type and grouped by device for the same property, period and search type. The returned rows show 36 mobile and 14 desktop impressions; no tablet row was returned.” This confirms the report's construction, not that markup caused the observations or that the feature will appear again.