Method

How Jacobi turns price checks into reviewable evidence.

Jacobi compares the prices returned to configured request profiles during one audit window — and keeps the observations, the extraction context, and the limits of the result together so your team can judge what happened.

Method v0.9Last reviewed Jun 26, 2026
01 ]The methodevidence, not a story
The short version

Compare what a page returns, not a story about what it might do.

A Jacobi audit starts with a public URL that an authenticated workspace is allowed to submit. The service asks for that URL through a configured set of request profiles within a short window, extracts usable prices, and shows the observations side by side.

The result is evidence for review. It can show a price spread, a MAP-floor undercut, a partial response set, or a blocked target. It does not turn one set of requests into a legal conclusion about a seller's intent or a site's pricing policy.

The pipeline

How an audit becomes evidence.

Profiles

A configured set of request contexts — country, device, session, network tier.

Requests

The same URL is asked through every profile inside one short audit window.

Extraction

Usable prices are pulled from each response with its source context kept.

Comparison

Observations are placed side by side; coverage gaps stay visible.

Evidence

A coverage-aware finding or MAP result, timestamped and reviewable.

Configured profiles

What changes between observations.

A profile is a configured request context, not a real person. Depending on the audit mode, Jacobi can compare selected country, device headers, session and referrer state, network labels, and a controlled language pair. The exact context that is available is retained with the evidence.

DimensionWhat Jacobi comparesImportant boundary
Request localeA selected country and a controlled language pair where supported.Not a promise of city, income, or personal targeting.
Device presentationConfigured desktop or mobile-style request headers.A request profile, not a physical device measurement.
Session contextFresh or returning session labels and referral context.Only configured factors are compared; the rest stays unknown.

Why this matters. Jacobi treats price variation, attributable variation, and a MAP-floor comparison as different outputs. A difference is recorded; it is not automatically assigned a cause.

Decision rules

How Jacobi stays restrained.

01

Coverage before confidence

Blocked, missing, and unusable responses stay visible. Thin coverage limits a comparison's confidence instead of being rounded into a clean answer.

02

MAP is a separate policy check

A MAP finding needs a configured floor, a recorded observed low price below that floor, and enough usable coverage for the rule to run.

03

No automatic legal conclusion

Findings describe what was observed at a point in time. They support your review; they are not an adjudication, guarantee, or legal advice.

Evidence lifecycle

What a reviewer can trace.

01

Watchlist

Product, seller, MAP floor, and target URL.

02

Scan job

Role, target limit, budget, rate, and status controls.

03

Evidence rows

Observed prices with source and extraction context.

04

Finding

Coverage-aware comparison or MAP policy result.

05

Export or share

Server-generated PDF or JSON, checksum, and controlled access.

Workspace exports are generated on the server and recorded. A public share can be redacted, time-limited, and revoked; redacted packets remove internal workspace identifiers, probe session identifiers, extraction metadata, and URL paths while keeping the useful evidence context.

Limits and safeguards

A clear result is never more certain than its inputs.

  • Third-party sites can rate limit, block, change, or require a login. Jacobi shows the resulting limitation rather than promising a complete read.
  • Managed provider access can improve reach. Guarded direct requests can be a fallback, but they do not carry the same configured proxy context.
  • Smart-24 names a configured profile matrix. Matching scout observations can be inferred to avoid needless identical external requests.
  • A captured price is a timestamped observation, not continuous monitoring and not a guarantee of a future offer.
  • Demo results are clearly cached samples. They are never represented as a live capture.

For access, data handling, and acceptable use, see the Privacy Policy and Terms of Service.