How the index is built

StreetProof publishes physical citations, not opinions. This page describes exactly how a photograph on a street becomes a number on a Presence Record, and the limits of that number. If you plan to cite our data, read the limitations section at the bottom first. It's the part most methodology pages leave out.

Collection

Manual capture fills deliberate coverage gaps. Drive Mode measures what naturally crosses a Scout's route or a safe parked view. Neither mode gets a shortcut: both must pass device, location, privacy, duplicate, brand-identity, and provenance checks. An accepted Drive Mode sighting from an unaffiliated Scout is real visibility and can count. Evidence from an owner, employee, sponsored campaign, or StreetProof-operated account stays labelled and contributes zero independent-observer credit.

An accepted licensed-sensor sighting receives the same base visibility weight as an accepted Scout sighting. Collection method does not change what happened at a particular place and time. Provenance does change what the record can prove: StreetProof-operated sensors cannot satisfy the independent-Scout threshold, and the entire sensor network receives the same daily and rolling contribution limits as one Scout. More sensors can improve geographic coverage; they cannot manufacture source diversity.

Privacy first

StreetProof processes eligible public-space imagery to detect visible business branding. Required privacy handling is applied before accepted evidence is stored, and captures that fail the privacy gate are not admitted to the evidence pipeline. We index businesses, not people. See privacy and removal.

Verification

Detected branding is machine-read, converted into an entity-match candidate, and checked against independent signals: OCR text, phone number, website/domain, known business records, location proximity, logo or visual similarity, and StreetProof's own observation history. A model can suggest a match or normalize messy text. It cannot publish identity by itself.

Weak or conflicting matches stay in the candidate queue. Reviewers can see the evidence source by source: what supported the match, what conflicted, and what the detector/OCR actually saw. Every minted observation enters the append-only, hash-chained ledger, but only corroborated or strictly verified observations enter public counts. One signed capture can be independently corroborated when its visual evidence and an outside business source agree. Weak or conflicting evidence stays provisional for review; nobody has to recreate the same angle. Additional independently corroborating source types can raise the record to verified. Ledger inclusion and checkpoint existence can be independently verified. Capture time is supported by device, server, provenance, and review evidence. External checkpoint anchoring began on the published activation date. The transparency page reports each checkpoint's actual external-proof status and distinguishes the retrospective public-beta baseline from externally timestamped history.

Machine matching is not perfect, and we do not pretend it is. The system is designed so uncertainty is inspectable instead of hidden in a confidence score. Every reviewer decision writes a human-review source and audit entry, so the correction trail is part of the product rather than a private cleanup job.

Physical citation taxonomy

Every public observation is a physical citation: an accepted record that a brand appeared in the physical world. It is the offline brand-mention equivalent of a backlink, not a literal web link or a ranking promise. The citation type controls what can safely be inferred from it.

Count taxonomy and reconciliation

Physical citations are accepted public, non-synthetic observations. Strictly verified observations are a subset of physical citations; corroborated observations also count as accepted evidence. Public synthetic observations are disclosed separately. Redacted, provisional, rejected, duplicate, and private QA rows do not enter the production citation count. The live reconciliation is published on methodology metrics.

Event time, acceptance time, and backfills

observed_at is when the physical capture occurred. accepted_at is when StreetProof resolved and admitted it, and published_at is when it became public. A later model may resolve a privacy-scrubbed archived capture without changing its original event time. Backfilled rows are marked derived_from_archive=true and retain the recognition model and methodology versions used for acceptance.

New observations store this classification as an immutable ledger fact. Public pages, evidence packets, and APIs expose it so consumers do not mistake an advertisement for a storefront. The full plain-language model is on the physical citations page.

How physical citations, PPS and PBI relate

The StreetProof Physical Brand Index™ (PBI) is the public index. A physical citation is one accepted evidence event. A Presence Record is the entity-level evidence history. Eligible records receive a Physical Presence Score (PPS), which orders them inside a defined city and industry cohort. Equal full-precision PPS values share a PBI rank.

Presence Records are city-local. When a separate identity review links records to the same brand, StreetProof may also show a network-wide brand total: distinct accepted, non-redacted citations across those reviewed records. That rollup does not replace the local count, change local PPS or PBI rank, or prove an office, franchise, or operating location.

Physical citations → Presence Record → independent-evidence eligibility → PPS → PBI position.

Industry cohort classification

Industry cohorts use a versioned StreetProof classification mapped to the United Nations ISIC Revision 5 taxonomy. Existing manual decisions are preserved. Automated labels require at least 95% confidence from a supported provider category, an exact reviewed brand entry, an unambiguous business-name rule, or agreement between those signals. Conflicting or unclear cases remain visibly unclassified until reviewed.

Industry classification changes only which peers form a cohort. It does not create a citation, change the evidence inputs to PPS, advance city coverage, affect Scout rewards, or imply that StreetProof endorses the business.

The Physical Presence Score (PPS)

PPS is a 0–100, cohort-relative brand-presence score (compared within city × industry), computed nightly from corroborated and strictly verified observations only, weighted by source trust. Each component answers a distinct question a customer might reasonably ask about a brand's observed physical footprint. Citation type remains visible, so promotional visibility is not presented as an operating address:

From street to score: a worked example

Say a scout photographs a wrapped van outside a hardware store on a Tuesday. The image is blurred for faces and plates, the wrap text is machine-read, and the phone number on the wrap matches an existing entity. Call it a Calgary electrician. The van's visual fingerprint doesn't match any vehicle we've seen for that entity, so a new asset is created. Because the scout is new, the observation sits as provisional; on Thursday another independent Scout observation Drive Mode sees the same van across town, the fingerprints match, and the observation is corroborated.

That night the scoring job reruns. The electrician's asset count ticks up by one vehicle, geographic spread gains a neighborhood cell it hadn't covered, and recency refreshes. None of these changes the score much on its own. PPS is cohort-relative, so the electrician's number moves only in comparison to other Calgary electricians, and single observations are deliberately small inputs. Sustained movement takes sustained presence. That is the design, not a limitation.

What gaming-resistance means here, concretely

“Can't be gamed” is a claim every ranking system makes and most can't back. Here is what ours rests on:

• Cohort-relative scoring. Your PPS is a comparison against businesses in your own city and industry, not an absolute count. Inflating raw observation numbers doesn't buy rank if the inflation is detectable. Volume spikes from small scout clusters are exactly what the anomaly detector watches for. Detected concentration creates a private review event; a confirmed integrity case can freeze the entity's rank pending human review and appeal. We do not publish an accusation from an automated signal alone.
• Repeated-source discounting. The same scout photographing the same business over and over hits same-cell, daily, and rolling 30-day contribution caps before any score component is calculated. A public rank requires at least three independent observers, five countable observations, and two observation days, plus an applied evidence-backed industry classification. Name-only guesses cannot create a comparison cohort or an official rank. One enthusiastic friend with a phone cannot move a score.
• Capture-mode neutrality. An accepted Drive Mode sighting is not discounted merely because the shutter was automatic. The system judges independence, location, time separation, duplication, and identity confidence. Passive measurement is useful precisely because the Scout was not hunting for a specific brand.
• Equal observation weight, separate independence. A reviewed licensed-sensor recognition and a reviewed Scout recognition have the same base visibility weight. Sensor networks are capped as a single platform-controlled source and receive no independent-observer credit. This measures real visibility without letting one automated network certify itself.
• Sponsor and affiliation firewall. Campaign-funded, owner-provided, employee, contractor, founder-connected, and platform-operated captures cannot satisfy the independent-observer gate. Sponsored and affiliated submissions remain score-neutral; approved platform sensors may contribute capped visibility only after independent Scouts open the public-ranking gate. Funding buys a defined collection project, never independence or a better position.
• Membership is display access, not evidence. Live Profile members can place a live map, Heartbeat, and badge on an authorized company domain. Membership cannot add observations, alter map areas, change a score, open the independent-observer gate, or buy PBI position. The widget reads the same public ledger as every other StreetProof surface.
• Mechanical rules, no human overrides. The score is a published formula over the ledger. Nobody at StreetProof, the founder included, can nudge a number; any change to an observation leaves an immutable audit-log entry, and the ledger itself is append-only. If we ever got a takedown demand from a business that disliked its score, there is literally no lever to pull.

What does not count

Public-feed data and attribution

Public records identify reviewed source-derived observations as reviewed sensor-derived evidence. StreetProof keeps exact sensor identifiers, source URLs, retained review crops, and operational selection logic private for security, privacy, and anti-gaming reasons. That does not erase provenance: every promoted observation retains its source class, frame hash, privacy audit, and reviewer decision in the private audit record.

Calgary sensor-derived observations contain information licensed under the Open Government Licence – City of Calgary. Contains information licensed under the Open Government Licence – City of Calgary. StreetProof publishes derived observations, not City camera images.

What PPS does not measure

Be clear-eyed about this before citing the score. PPS measures one thing: verifiable physical brand presence. It is not a quality rating or automatic proof of a local office. A brand with a high PPS has been observed repeatedly across real physical contexts. It could still do mediocre work, overcharge you, or answer the phone rudely. We measure none of that, on purpose, because presence is observable and quality is an opinion.

The inverse cuts too. A low or absent score is not proof a business is fake. Plenty of legitimate operations have thin physical footprints: a solo bookkeeper with no vehicle, a caterer working from a commissary kitchen, a new company whose first wrap is at the shop this week. PPS also can't see indoor presence, unbranded vehicles, or anything on streets we haven't covered. What we claim is narrow and defensible: this branding was observed, at these places, at these times, and here is the evidence. Anything beyond that is your inference, not our data.

Coverage confidence & the limits of absence

Coverage tiers describe the maturity of StreetProof's own dataset, not whether every business or area has been mapped. Seed means collection has begun but remains early. Building means breadth and repeat history are expanding. Dense means broad, recurring coverage across many city cells and sources. Seed and Building may use published fallback thresholds. Dense fails closed unless a reviewer approves an immutable eligible-cell manifest; at least 30% of those eligible cells must be observed, and the evidence must include three distinct ranking-eligible independent Scouts. Freshness is tier-aware. A Seed city is Current when its latest eligible citation is no more than 30 days old, Aging from 31–90 days, and Stale after 90 days. Building and Dense cities are Current only when evidence arrives on at least eight days in the latest 30-day window and at least half of observed cells have refreshed within 45 days. Unlaunched cities do not receive a freshness badge. Every API payload carries tier, freshness, confidence, methodology version, and a limitation. Absence of observations is never proof of absence.

Synthetic, rejected, duplicate, redacted, and private QA rows cannot advance a city. Membership cannot advance a tier. Source-family diversity improves city coverage quality, but does not turn a platform-operated observation into independent ranking evidence.

Recognition archive boundary

Faces and licence plates are blurred before any image becomes retained evidence, enters the recognition archive, or appears publicly. Unsanitized frames are processed transiently in the private privacy pipeline and are not retained as evidence. Privacy-scrubbed candidates remain in hot storage for up to 30 days. Material can remain for up to 365 days only when it is accepted, reviewed, training/reference-worthy, and both source rights and jurisdiction rules explicitly permit cold retention. Weak, rejected, duplicate, and unusable candidates do not enter cold storage. A documented legal or security hold—with an ID, reason, approver, and recorded time—is the only exception to ordinary deletion.

Methodology versions

Current public contracts: public-metrics-v3, city-coverage-v3, and observation-v2. Versioned configuration, immutable metric receipts, completed city-coverage runs, and coverage history preserve the rules used for each public label. Superseded versions remain in the methodology registry instead of being silently overwritten.

Error rates & corrections

Methodology metrics are published quarterly at /methodology/metrics, with a machine-readable copy at /api/v1/methodology/metrics. OCR precision/recall stay marked not-yet-reportable until a labeled holdout set exists. Entity-match review outcomes, redactions, dispute timing, and city coverage are published from live operating data. Material corrections are logged permanently at /corrections. We'd rather publish an unflattering error rate than an unmeasured one.

Visible text is not automatically a company name. Calls-to-action and promotional phrases such as “Rent Me,” “Call Now,” or “We Deliver” fail the automatic entity-name gate even when a phone number is also visible. The system must identify the complete operator name or keep the candidate private for review.

Conflict-of-interest disclosure

StreetProof was founded by the operator of Calgary Garage Door Fix, which is itself an indexed business. Scoring is cohort-relative and mechanical; no staff action can alter an observation without an immutable audit-log entry. The founder's company competes for its PPS under exactly the same formula as every other Calgary trades business. That constraint is the disclosure.

Status

The index is in public beta. Collection is open, but unverified sideload submissions remain non-anchor evidence until an outside information independently corroborates them. Demonstration data is synthetic and labeled as synthetic wherever it appears.