Unlock Carrier Data Value
Without Losing Control

Scenario

A major retailer spends $14M annually on digital advertising to drive store traffic. GPS-based mobile attribution captures only 60% of visits — urban canyons, indoor signal loss, and app fragmentation create massive blind spots. Retailers cannot correlate ad exposure to actual visits, making media buying a guessing game. Worse: 40% of the ad spend—$5.6M—produces zero visible attribution because the link between exposure and store entry is invisible. The retailer cannot optimize media budgets or justify spend to leadership.

How It Works

Carrier uploads aggregated, anonymized location signals to Placino: daily store visit counts indexed by geographic cohort (census block group) and time windows, encrypted with envelope encryption (AES-256-GCM). Retailer imports ad impression logs indexed by the same geographic cohort and timestamp. Placino performs federated cohort-level matching: joining impressions to visit lifts without exposing individual device IDs or locations to either party. Propensity-score matching controls for selection bias (customers exposed to ads may differ from unexposed cohorts). Federated regression computes incremental store visit lift per ad channel with differential privacy (epsilon = 1.5). Results: "Cohorts exposed to Google Display had 18% higher store visit frequency than control; Facebook cohorts 12% higher." No individual device locations or customer paths are revealed.

Scenario

A mobile carrier and premium streaming service each serve overlapping 800K subscribers. Carrier predicts churn using billing friction signals (failed payments, plan downgrades, customer service calls). Streaming service predicts churn using engagement (login frequency, content consumption, watch time trends). Separately, each predicts churn at 68–70% accuracy. But the real churn drivers are invisible to each company: a subscriber with billing payment failures and low streaming engagement is 7.5x more likely to cancel than either signal alone suggests. Combined, they could save $8M in annual churn, but privacy restrictions prevent sharing raw customer records.

How It Works

Both companies encrypt churn-risk feature vectors to Placino: carrier uploads (hashed customer ID, failed-payment count, plan change frequency, support ticket volume); streamer uploads (hashed ID, login frequency, watch-time percentile, genre affinity). Placino uses Private Set Intersection to identify matched customers without exposing either company's customer roster. A federated machine learning model trains on combined signal sets using differential privacy (epsilon = 1.0): logistic regression on joint features to predict next-30-day churn. Trained model weights are shared; each company applies locally to their own customer base, seeing only cohort-level risk distributions, not individual predictions. Results: "High-risk cohorts (payment failures + low engagement) have 7.5x baseline churn rate." Each company routes these cohorts to targeted retention campaigns.

Scenario

Carriers possess the richest first-party behavioral data on Earth: real location, payment behavior, app usage, search queries. But direct monetization is impossible due to privacy law and buyer skepticism. Third-party data brokers claim to have "telco-quality" audiences but provide no validation. Advertisers distrust unverified segments, paying commodity prices. Carriers leave $18B in annual monetization upside on the table because they cannot prove segment quality or share detailed behavioral data.

How It Works

Carrier builds anonymized audience segments in Placino: cohorts defined by behavioral signals (location frequency, app category affinity, historical purchase patterns) using hashed pseudonymous IDs, not names. Segments encrypted with envelope encryption (AES-256-GCM). Advertisers query Placino's segment discovery interface: browsing cohort statistics (size, age distribution, propensity to purchase category X) and quality metrics (look-alike accuracy against past conversions). Placino computes segment metrics with differential privacy (epsilon = 2.0) to prevent reverse-engineering of individual attributes. Advertisers can purchase real-time activation APIs that return anonymized cohort membership (is this user in segment X? yes/no) without exposing the segment definition. Merkle-chain audit trails provide regulatory evidence of anonymization and privacy controls.

Scenario

Two regional carriers cover overlapping geographies. Carrier A sees cell-level signal quality (RSRP, latency, throughput) in its coverage areas but cannot see gaps where users roam onto Carrier B's network. Carrier B has the same blind spot in reverse. Users experience bad handoffs and dropped calls at coverage edges, but neither carrier can identify the root cause due to incomplete network visibility. Combined, they could optimize roaming and coverage planning, saving $4M+ in capex. But neither carrier shares network topology data due to competitive sensitivity and regulatory concerns.

How It Works

Both carriers upload anonymized, aggregated network signal data to Placino: signal strength (RSRP), latency, and throughput metrics per geographic cell (anonymized via census block grouping), encrypted with envelope encryption (AES-256-GCM). Placino performs federated gap analysis: comparing coverage quality at cell boundaries without exposing either carrier's proprietary cell locations, frequencies, or topology. Federated clustering identifies geographic zones where one carrier's signal is poor and the other's is strong — natural roaming optimization opportunities. Results are delivered as anonymized heatmaps ("Zone X: improve roaming agreement to reduce handoff failures") with no revealing of specific cell locations or frequencies. Differential privacy (epsilon = 2.0) prevents reverse-engineering of network topology.

Scenario

Connected TV (CTV) is now the fastest-growing ad channel, projected to reach $45B+ in 2026. But it is also the most unmeasured. Carriers control the set-top-box and see every ad served; advertisers see only aggregate performance. A major ecommerce brand runs $50M in annual CTV campaigns but cannot measure which ads drive purchases. Is CTV delivering 2x ROAS or 0.5x? Unknown. Advertisers demand proof before increasing budgets. Without deterministic attribution, $26B in CTV spend remains fundamentally invisible. Advertisers cap CTV budgets to 5% of mix despite believing true ROI is 3x higher.

How It Works

Carrier uploads anonymized CTV viewing events to Placino: household identifier (hashed cryptographically), program watched, ad exposure timestamp, and creative ID — encrypted with envelope encryption (AES-256-GCM). Advertiser imports purchase records: anonymized customer identifier (hashed in same cryptographic domain), transaction timestamp, product category purchased, and transaction amount. Placino performs deterministic cryptographic matching on household/customer identifiers to link ad exposures to purchases at the household level. Federated cohort analysis segments viewers by exposure intensity (saw ad 1x vs. 3x) and measures incremental purchase lift, controlling for selection bias using propensity-score matching. Results: "Households exposed to advertiser's ad had 2.8x higher purchase probability than unexposed cohort." Differential privacy (epsilon = 1.5) is applied to aggregate outputs so advertiser sees channel-level ROAS, not individual household journeys.