For CMOs + customer-data + marketing analytics leadership
The customer is one person. Your systems see seven separate identities.
Anonymous web visitor, anonymous mobile-ad click, known email subscriber, anonymous foot traffic at Tampa, anonymous phone caller at Phoenix, known POS purchaser at Phoenix, known loyalty member. Adobe Customer Journey Analytics + Treasure Data + Tealium + Salesforce Data Cloud + Segment ship the CDP primitive. The per-location cross-touchpoint identity resolution across offline plus online signals is operator-side architecture.
What this gets you
- Single sequenced journey per customer across every touchpoint — web + email + paid ads + phone + walk-in + POS + loyalty + customer service + reviews. Cross- device, cross-channel, cross-location, cross-time.
- Deterministic-plus-probabilistic identity resolution with confidence-weighted scoring — deterministic on explicit identifier matches (email + phone + loyalty ID) at near-100 percent confidence; probabilistic on device fingerprint + behavioral pattern + location proximity + temporal sequence with confidence weighting.
- Position in the 4-skill Parallel-Inputs → Resolve → Forecast bundle — the Resolve stage stitches POS receipts + call tracking + walk-in foot traffic + behavioral enrichment into a single per-customer journey that feeds the MMM Forecast stage and the per-location attribution models.
- Multi-location identity (same customer at multiple stores) — the customer who visits Phoenix and Tampa and books online at Denver appears as one journey with three location interactions, not three separate identities. Per-location interaction history rolls up to the parent customer record.
- Per-jurisdiction PII policy gating— GDPR + CCPA + HIPAA + cannabis-state + per-state laws encoded as input filters on the identity-resolution stream. Right-to-deletion events propagate through the identity graph via the customer-data-orchestration emit-change pipeline.
Seven separate identities, one customer, no journey
A customer based in Phoenix opens Instagram during her lunch break and sees a paid ad for a local day-spa service. She taps through on mobile, lands on the operator landing page, browses for two minutes, and closes the tab without converting. The mobile session fingerprints to anonymous visitor ID A. Three days later she opens an email she received from the operator brand (she subscribed two years ago), clicks a promotional link from desktop, and lands on the same landing page. The desktop session fingerprints to anonymous visitor ID B. The email click attributes to known subscriber ID C.
She travels to Tampa for a long weekend. While shopping at the mall she sees a sign for the operator Tampa location and walks in for a five-minute browse. The mobile-data partner pings foot-traffic identity D. She does not purchase. She returns to Phoenix the following Monday and calls the Phoenix location to book the service she had researched. The call-tracking provider attributes call identity E to the Phoenix landing-page session. The booking goes into the CRM as new contact ID F because the receptionist did not ask for her email at the booking step.
Three weeks later she shows up at the Phoenix location for her booked appointment. She pays with the loyalty card she signed up for during her last visit two years ago, which the operator linked to the original email subscriber ID C. The POS transaction records purchase identity G.
The customer is one person. The systems see seven separate identities (A through G). Marketing attribution reports the Instagram ad as a no-conversion impression, the email as a no-conversion click, the Tampa foot-traffic as a no-conversion visit, the phone call as an inbound that nobody attributes back to the Instagram ad, the POS purchase as a loyalty-program retention event. The actual journey — ad discovery → consideration → cross-location window-shopping → booking → conversion at the original location — is invisible to the attribution model and to the marketing budget decisions that follow.
Cross-touchpoint identity resolution stitches the seven identities into one journey. Deterministic matching on the email-and-loyalty link confidently connects C and G. Probabilistic matching on the device fingerprint plus temporal pattern plus geo proximity links A to B at high confidence and B to C via the landing-page session. Foot-traffic ID D links to A through device fingerprint and to G through loyalty-membership-plus-geo. Call ID E links to A through the landing-page-session-plus-phone- number-extraction. The seven identities collapse to one journey with seven events. The Instagram ad attributes correctly. The Tampa visit registers as consideration. The MMM Forecast stage sees the accurate channel-contribution mix.
What is in market — and what each category leaves to you
The customer-journey-analytics platform layer is mature. The cross-touchpoint identity resolution across offline plus online plus multi-location at operator scale is architecture.
Enterprise CDP and customer-journey analytics — Adobe Experience Platform + CJA, Treasure Data, Tealium AudienceStream, Salesforce CDP / Data Cloud, Segment (Twilio), mParticle, Lytics, BlueConic
Excellent at event ingestion, identity-graph storage, journey-analytics dashboards. The offline POS plus call-tracking plus walk-in foot-traffic input stream wiring, the per-location identity stitching, the deterministic-plus-probabilistic resolution tuned per identifier mix, and the per-jurisdiction PII policy gating are operator-side architecture on top of the platform layer.
Identity-resolution specialists — LiveRamp, Acxiom, Throtle, Tapad (Experian), Neustar
Strong at third-party identity graphs that link online and offline identifiers across the broader consumer data ecosystem. The first-party cross-touchpoint resolution on the operator data estate (operator-owned POS + call + walk-in + behavioral) is a different layer; the third-party graphs enrich but do not replace first-party resolution.
Open-source CDP and event-streaming — RudderStack, Jitsu, PostHog (group-aware identity), Snowplow, Kafka identity-stream pipelines
Capable of identity-resolution pipelines for operators with in-house data-engineering capacity. The deterministic-plus-probabilistic confidence- weighted scoring, the per-jurisdiction PII policy engine, and the integration with the broader Parallel-Inputs → Resolve → Forecast architecture are the operator-side build.
Attribution and journey-overlap — Singular, AppsFlyer, Branch, Northbeam, Triple Whale, Rockerbox
Strong at attribution-modeling that depends on the identity-resolution substrate. The identity resolution feeds the attribution; the attribution consumes the journey rather than producing it. Operators that buy attribution without solving identity see inconsistent attribution outputs.
The seven CRM contact records for the same customer
The status quo at most multi-location operators. The Instagram lead, the email subscriber, the foot- traffic ping, the phone caller, the POS purchaser, and the loyalty member sit as seven CRM records that nobody has merged. The marketing director reports seven new contacts per month from this customer when reality is one customer who interacted through seven channels.
The pipeline, end to end
- Position in the 4-skill Parallel-Inputs → Resolve → Forecast bundle. Parallel-Input 1 (POS receipt integration) + Parallel-Input 2 (call tracking integration) + Parallel-Input 3 (walk-in foot-traffic integration) + Parallel-Input 4 (behavioral enrichment) feed the Resolve stage (this skill) which produces the resolved per-customer journey that feeds the Forecast stage (marketing mix modeling).
- Deterministic resolution layer.Match identities on explicit shared identifiers — same email, same phone, same loyalty ID, same hashed identifier with matching salt. Match confidence near- 100 percent when identifiers match. Deterministic resolution handles the high-confidence cases first.
- Probabilistic resolution layer.Match identities on inferred signals — device fingerprint (canvas + audio + WebGL plus user-agent plus IP), behavioral pattern (page-sequence + dwell time + interaction pattern), location proximity (geo + time window), temporal sequence (events that fit a plausible customer-journey timing). Confidence varies per signal mix.
- Confidence-weighted scoring. Each candidate match carries a confidence score. The resolution pipeline merges high-confidence matches into the resolved journey immediately. Borderline matches surface to an analyst review queue or to a threshold-tuned auto-merge with audit-trail logging. Low-confidence candidate matches stay separate until more signal arrives.
- Anonymous-to-known transition. When an anonymous visitor identifier (cookie + device fingerprint) later provides a known identifier (email form submit, phone call, loyalty signup), the resolution retroactively links the anonymous history to the known identity. The Instagram ad click from six weeks ago attributes correctly to the customer who just signed up.
- Multi-location identity (same customer at multiple stores). A customer who interacts with Phoenix and Tampa and Denver locations resolves to one parent customer record with three location-interaction histories. Per-location interactions roll up. Per-location attribution sees the customer as one journey crossing three locations rather than three separate journeys.
- Per-jurisdiction PII policy gating.The input stream filters on regulatory tags — GDPR consent state, CCPA opt-out, HIPAA BAA coverage, cannabis-state PHI restrictions, per-state laws. Healthcare PHI cannot match across non-PHI channels without explicit BAA-covered consent. The policy engine shares substrate with the marketing-compliance cross-agent overlay.
- Right-to-deletion propagation. When a customer exercises right-to-deletion under CCPA or GDPR, the deletion event propagates through the identity graph via the customer-data-orchestration emit-change pipeline. Affected identities tombstone; the resolved journey for that customer becomes unavailable to downstream consumers; audit trail captures the deletion.
- Identity decay handling.Customers who have not interacted for 6-plus months accumulate decay signal — phone number reassigned, email address abandoned, address moved, device replaced. Decay-aware resolution downweights stale signals. The decay model retrains per cycle against ground-truth re-engagement events.
- CRM deduplication output.The resolution output drives CRM deduplication — the seven CRM contact records for the same customer merge to one with three location-interaction histories. The CRM-record-creation skill on the operator stack ingests the resolved journey rather than creating duplicates.
- Match-rate, precision, and recall measurement. Match-rate (percent of events that resolved to a known identity), precision (percent of merges that were correct against ground-truth audit), recall (percent of true matches that the pipeline found) tracked per cycle. Signal feeds confidence-threshold tuning per signal mix.
- Integration with downstream Forecast and per- location attribution. The resolved per-customer journey feeds the MMM Forecast stage (cross-link to the marketing-mix- modeling pillar) and the per-location attribution models (cross-link to the per-location-attribution- models pillar). The same Resolve substrate feeds the customer-data-orchestration emit-change pipeline that broadcasts customer changes to every downstream consumer.
- ROI measurement. Cross-channel attribution accuracy pre vs post identity-resolution deployment. CRM-duplicate rate pre vs post. Marketing budget reallocation magnitude based on accurate channel attribution. Per-location customer lifetime value precision (a customer who interacts at three locations attributes to a parent LTV rather than three lower LTVs across locations).
Frequently asked
What is customer journey tracking?
Customer journey tracking is the practice of stitching every touchpoint a customer has with the operator brand into a single sequenced journey — web visits, email opens, paid-ad clicks, phone calls, walk-in visits to physical locations, POS purchases, loyalty events, customer-service interactions, review submissions. The journey crosses devices, channels, locations, and time. Enterprise customer-journey-analytics platforms (Adobe Customer Journey Analytics, Treasure Data, Tealium AudienceStream, Salesforce Data Cloud, Segment, mParticle) ship the underlying CDP primitive. The cross-touchpoint identity resolution that turns the raw events into a coherent per-customer journey across multi-location franchisees is operator-side architecture.
Why does single-channel customer journey tracking fail multi-location operators?
A multi-location operator runs a customer who searches online from Phoenix, clicks a paid ad on mobile, opens an email on desktop the next day, visits the Tampa location while traveling, calls the Phoenix location after returning home to book a follow-up service, and pays via POS at the Phoenix location three weeks later. The customer is one person; the systems see seven separate identities — anonymous web visitor, anonymous mobile-ad click, known email subscriber, anonymous foot-traffic at Tampa, anonymous phone caller at Phoenix, known POS purchaser at Phoenix. Single-channel tracking sees the seven separate identities. Cross-touchpoint resolution stitches them into the one customer journey.
How is this different from Adobe Customer Journey Analytics, Treasure Data, Tealium AudienceStream, Salesforce Data Cloud, Segment, or mParticle?
Those platforms ship the customer-journey-analytics primitive — event ingestion, identity-graph storage, journey-analytics dashboards. They are excellent at the platform layer. The per-location identity stitching across offline POS receipts plus call tracking plus walk-in foot traffic plus online behavioral signal, the deterministic-plus-probabilistic resolution tuned per identifier mix, the multi-location identity (the same customer at multiple stores), the post-cookie first-party-data strategy, and the integration with the Parallel-Inputs → Resolve → Forecast architecture on the offline-attribution-intelligence agent are operator-side wiring on top of the platform primitive.
What is deterministic vs probabilistic identity resolution?
Deterministic resolution matches identities on explicit shared identifiers — the same email address, the same phone number, the same loyalty ID. Match confidence is near-100 percent when identifiers match. Probabilistic resolution matches identities on inferred signals — device fingerprint, behavioral pattern, location proximity, temporal sequence. Match confidence varies. Real-world multi-location identity stitching needs both. Deterministic resolves the high-confidence cases when explicit identifiers are present. Probabilistic resolves the anonymous-to-known transition and the cross-device gaps. The architecture combines them with confidence-weighted scoring.
How does the Parallel-Inputs → Resolve → Forecast architecture work?
The offline-attribution-intelligence agent owns a 4-skill bundle. Parallel-Input 1 (POS receipt integration captures POS purchase events). Parallel-Input 2 (call tracking integration captures phone call events). Parallel-Input 3 (walk-in foot-traffic integration captures visit events). Parallel-Input 4 (behavioral enrichment captures online behavioral signal). Resolve (this skill — cross-touchpoint identity resolution stitches the parallel inputs into a single per-customer journey). Forecast (marketing mix modeling consumes the resolved journeys per market). The Resolve stage is the structural pivot from many-anonymous-events to single-journey-per-customer.
How do you handle PII in identity resolution under GDPR, CCPA, HIPAA, and cannabis-state law?
PII fields carry regulatory tags. The identity-resolution pipeline processes only the identifiers each regulatory regime permits — full email under GDPR consent, hashed email otherwise; full phone under appropriate consent, hashed phone otherwise; loyalty IDs as proxy when direct PII is not permitted. Right-to-deletion events propagate to the identity graph through the customer-data-orchestration emit-change pipeline, tombstoning the affected identity. Healthcare locations under HIPAA cannot match identity across non-PHI channels with PHI channels without explicit BAA-covered consent. The per-jurisdiction PII policy engine that gates the marketing-compliance overlay also gates the identity-resolution input stream.
Hire the agent that resolves identity across touchpoints
The offline-attribution-intelligence agent owns the 4-skill Parallel-Inputs → Resolve → Forecast bundle — POS + call + walk-in + behavioral as parallel inputs, cross-touchpoint identity resolution as the Resolve stage, marketing mix modeling as the Forecast stage — sitting on top of whichever CDP (Adobe Experience Platform, Treasure Data, Tealium AudienceStream, Salesforce Data Cloud, Segment, mParticle) you license downstream. Deterministic-plus- probabilistic resolution with confidence-weighted scoring + per-jurisdiction PII policy gating + per-location identity coverage measurement.
We scope on the call and send a private checkout link after.
Related reading: Per-location attribution · Per-market MMM · Customer data orchestration