MachineRead terminal-window signal markA stylized terminal window with a green signal scan-line and two data marks, indicating machine-readable signal inspection.MachineReadAI & search readiness audit

Blog

Choose the checks that fit the site

Choose a MachineRead preset or refine supported check families with Custom scope while keeping the Essentials scoring frame consistent.

author
published
reading
4 min

A local repair shop should not need an agent-to-agent endpoint to explain its opening hours. An API product may need considerably more than a homepage and a contact link. Applying the same optional expectations to both would make an audit harder to act on.

MachineRead's presets let you declare the relevant site context before the scan. They change which supported sub-signals apply, while keeping the Essentials report at 14 included groups and 60 checked points. The selection is yours; MachineRead is not automatically classifying the business (MachineRead, "Methodology").

Presets are declared before the scan

MachineRead supports seven request identifiers: Blog/Content, Corporate/Brand, Services/Local, News/Publisher, SaaS/Product/API, Ecommerce/Catalog, and Custom/Power User. The frontend exposes those choices before submission, and the public request model accepts an optional preset plus optional custom_overrides.

That timing matters. Scope is not a post-hoc explanation for a disappointing report. It is part of the request. A user says what kind of site they want MachineRead to evaluate, and the resolver decides which supported families are applicable under that declared configuration.

The Essentials frame stays fixed

Preset selection changes applicability inside the audit. It does not change the main Essentials scoring frame.

Across every named preset and supported Custom configuration, MachineRead keeps 14 included report rows, eight locked rows, and a 60-point Essentials checked maximum. Scope changes which sub-signals are active inside rows, which strict agent-readiness probes are counted, which family copy appears in the report, and which benchmark context is fair to show.

That distinction keeps the score legible. If the Ecommerce preset is selected, commerce-related evidence can become applicable. If a general corporate site is selected, commerce expectations should not quietly become failures. In both cases, the Essentials report still uses the same row inventory and checked denominator.

A preset is therefore not a way to unlock paid rows, add arbitrary checks, or make the public free audit bigger. It is a way to say which supported free families should matter for this site type.

Custom starts from a known base

Custom/Power User is for explicit family control. It starts from the Blog/Content base, which includes Article/BlogPosting schema and feed discovery, then accepts supported overrides.

That gives Custom a safe default rather than an empty switchboard. A user can add OAuth/OIDC and auth.md, remove a Blog default, or choose a named preset and then make a material refinement. The response can then disclose which families were included, which were excluded, and which overrides were applied.

Custom is still bounded. Unknown keys are rejected. Locked and universal rows cannot be toggled. A disabled parent dimension cannot keep an enabled child. Commerce fields depend on Product/Offer schema. Protocol and account/auth dimensions need at least one matching child family if they are explicitly enabled.

Benchmarks depend on the resolved scope

Benchmarks also need the scope label attached. An unmodified named preset can use its same-preset peer cohort. A Blog/Content report can be compared with stored Blog/Content context, and an Ecommerce/Catalog report can be compared with stored Ecommerce/Catalog context.

Custom, legacy, and materially changed named-preset requests do not have a single matching named cohort. MachineRead labels their broader comparison context rather than presenting it as a like-for-like group.

For example, a custom scope that adds only two auth families differs from the full SaaS preset. An Ecommerce request that disables account/auth families also differs from its default cohort. Check the returned pool before comparing either report with an unmodified preset.

How to choose a scope

Pick the closest named preset when the site has a clear category:

  1. Use Blog/Content for blogs, newsletters, and content sites.
  2. Use Corporate/Brand for company, portfolio, agency, and brand sites.
  3. Use Services/Local for local businesses and service providers.
  4. Use News/Publisher for publishers and news sites.
  5. Use SaaS/Product/API for software products, APIs, and developer tools.
  6. Use Ecommerce/Catalog for stores, catalogs, and marketplaces.
  7. Use Custom/Power User when the named presets are close but not exact.

Keep the selected scope with the report, especially when sharing it or comparing repeat runs. It explains which families you asked MachineRead to inspect and prevents a change in scope from looking like a change in the site.

Works Cited

MachineRead. "Methodology." MachineRead, https://www.machineread.ai/docs/methodology. Accessed 10 Sept. 2026.

See also