Smart Screen Advertising Platform

Every screen filled.
Every play proven.

One platform that pools digital screens at any number of locations, fills them with ads, and proves what each ad achieved. Software and hardware integration, delivered as one system.

Audience

Measured.
Not modelled.

Radar and on-device camera analysis report how many people stood in front of each screen, for how long, during each ad.

Response

Tied to the
exact moment.

A fresh QR code on every ad play ties each scan to one screen, one creative and one second in time.

What it does

One system. Four jobs.
All of them provable.

We design and deliver the complete advertising system for physical digital screens: the player on the screen, the sensors beside it, the ad server behind it and the reporting on top. Take it as a ready solution, or have us deploy it around your network.

The gap

Plenty of screens.
Not enough proof.

Advertisers still have no accepted way to verify what a screen delivered. These are the five questions they ask, what the market answers today, and what this system answers instead.

The market today

This system

How the system works

One clock. One record.
Every play.

Each screen carries a player and its sensors. They send signed play logs and audience counts to the central platform, which sends back ads and schedules. The platform joins every play with its audience, context and scans into one record.

Screen Player + radar + camera Runs offline. Signs every play. signed play logs audience counts ads + schedules Platform Ad server Screen registry Play record store Redirect + attribution Experiments AdvertisersDashboards, reports, API ExchangesOpenRTB 2.6 programmatic demand Screen ownersUptime, plays, earnings
01

Capability one

Pool every screen.
Never show a blank one.

Every screen in the network runs the same player and takes its ads from one central ad server, so a campaign can be booked across hundreds of locations in one step.

  • A screen registry holds each screen's location, venue type, size, orientation, operating hours and excluded ad categories.
  • Venues are classified with the OpenOOH venue taxonomy, so buyers' tools recognise them.
  • Screens owned by different parties sit in one pool. Each owner sees their own screens, uptime and earnings.
  • Owner payouts are calculated from verified uptime and verified plays, not from installed screen count.
  • Downloads and caches ads in advance, so playback continues through network and power interruptions.
  • Restarts itself after a power cut or crash, with a hardware watchdog and a real-time clock.
  • Sends a heartbeat with app version, connectivity, storage and temperature.
  • Updates remotely. No site visit is needed for a software release.
Share of loopPer-play bookingFrequency capsCompetitor separationDaypartingBudget pacingCreative approvalPrice floors per venue

The system speaks OpenRTB 2.6, the industry standard with dedicated fields for physical screens. It connects to existing exchanges so agency budgets can reach the network without you building an exchange of your own.

Fill ladder click a rung to remove that demand
    Direct campaign Monsoon Sale Slot 0412

    The highest-value available ad always wins the slot. House ads sit at the bottom, so the screen is never blank.

    Proof-of-play, signed on the devicescreen-on evidence ✓
    02

    Capability two

    Count the people.
    Not the multipliers.

    Live sensor view simulated
    Now playing creative v3
    In zone0
    Entries0
    Avg dwell0.0s
    Closest0.0m

    Two sensors, each doing the job it is good at. Radar on every screen for count and dwell. Cameras on a sample of screens for attention. Each reading is matched to the ad playing at that second.

    Radar, on every screen

    A small module in or beside the screen tracks people as moving points. No image and no device identifier is ever captured.

    • Entries to the viewing zone, dwell, closest approach, slowed or stopped
    • 24 GHz module: up to 3 people, 6 m. Lifts, lobbies, counters
    • 60 GHz module: 6 to 14 m across 120°. Busier indoor sites

    Camera, on a calibration sample

    Analysis runs entirely on the device. Frames are analysed in memory and discarded. Nothing is stored or sent.

    • People present, people facing the screen, dwell and attention time
    • Working range roughly 6 to 8 m indoors
    • Attention means face direction, not eye tracking
    Why both, and why a sample

    Cameras on a sample of screens per venue type give a dwell-to-attention ratio. That ratio is applied to radar-only screens of the same type, so the whole network gets attention estimates at radar cost, with cameras only where the venue permits them.

    A stricter crediting rule

    An ad is credited with a viewer only when that person faced the screen for at least one to two seconds inside the ad's slot. The common industry rule credits every ad that overlapped a person's presence.

    Other sensors we integrate

    Overhead time-of-flight counter

    Venue entrance footfall, so screen audience can be compared against total venue traffic.

    Existing venue CCTV

    Body counts where a venue already has cameras. Good for counts, weaker for attention.

    2D LiDAR

    Tracking in wide, open zones out to 12 m.

    Vehicle-counting camera

    Roadside screens. Counts vehicles and applies an occupancy factor.

    03

    Capability three

    A new code on every play.
    Every scan names its moment.

    Every response to an ad is tied to the exact screen, creative and moment that produced it, because the screen draws a fresh QR code for each play.

    • Each code carries a short signed token standing for the screen, the creative and that play.
    • A scan passes through the redirect service, which logs it and forwards the person in a fraction of a second.
    • Tokens expire. A scan of a photographed code days later is recognised as a late scan and reported separately.
    • No app, no consent prompt, no personal identity. A scan yields the token, the time, a coarse city and the phone's operating system.
    • Codes use a branded short domain with a visible, readable address, which guards against fake-QR fraud.
    Per-play QR the code is real, try your phone
    Creative v3 Fresh brew,
    2 minutes away.
    go.brand.in/k7Qx2m
    Token
    k7Qx2m
    Screen
    BLR-CAF-0412
    Creative
    cr_219 · v3
    Play started
    14:02:07
    Scan events0 scans

    No scans yet. Press the button, or point a phone at the code.

    Conversion endpoints

    Any fixed response channel on a screen, such as a promo code or phone number, is attributed to the creative playing at that second by matching against the play log. Advertisers can send back orders, leads or redemptions through an API or a file upload, matched to the original scan token.

    04

    Capability four

    The right ad for this screen,
    this hour, this weather.

    Live triggers change the conditions
    Evening daypart Good afternoon, Mumbai.

    Cold brew and a quiet table, till 6 pm.

    20 s creative

    Targeting

    • Location. City, pin code, radius around a store or landmark.
    • Venue type. Café, gym, salon, clinic, residential lobby, mall, office.
    • Time. Hour, day of week, festival and holiday calendars.
    • Measured audience. Screens and hours ranked by actual count and dwell.
    • Lookalike screens. More screens like your best performers.

    Dynamic creative

    One HTML5 template with swappable parts, such as price, city name, countdown or offer, produces many versions of an ad without new design work. For low-powered players the server renders the finished image, so the same campaign runs on every screen type.

    Optimisation

    • Forecasting. Expected count per screen per hour, from the network's own history.
    • Pacing. Budget steered toward the screens and hours that perform.
    • Creative rotation. Delivery shifts to the version earning more dwell, attention or scans.
    • Price floors. Set per venue type and daypart.

    Controlled experiments

    What did the campaign cause?

    Because the system controls delivery, it can randomly hold a campaign back from 5 to 10% of eligible plays, or from a matched set of screens. Comparing response with and without the ad gives the lift the campaign caused, not just a raw count. It needs no personal data and no consent.

    Illustrative numbers on 400,000 eligible plays. A larger holdout costs more plays and tightens the answer.

    Scan rate, ad shown
    Scan rate, held back
    Measured lift +38% ± 9 pts

    Pricing models the data supports

    Per play or share of loopPer thousand measured impressionsPer attentive secondBase price plus a bonus per scan

    The data

    Every play becomes one record.

    What played, who was there, what was happening and what followed, joined on a single clock. Pick a layer to see what it holds and how trustworthy it is.

    play_8f3a21c9 Exact

    Play

    Source: Player and ad server

    Dashboards

    For the network operator, each screen owner and each advertiser, with their own access levels.

    Reports and exports

    Scheduled reports and CSV exports.

    API

    For agencies and advertisers who want the data in their own tools.

    Method on every number

    A campaign report that states the method behind each figure, ready for third-party audit.

    Design principle

    The data we deliberately leave out.

    We collect what is clean, cheap and provable. These signals are possible in a lab, or were common in the past. We leave them out because they are inaccurate in the field, legally exposed, or both.

    Screens and hardware integration

    Runs on almost any screen.
    We integrate the hardware too.

    The intelligence sits in our software, and we specify, integrate and quality-check the hardware it runs on: player boxes, radar, cameras and LED controllers. Everything is commercially available, so no part of the system depends on a single hardware vendor.

    A

    Android player box on HDMI

    A small box we specify plugs into any screen with an HDMI port.

    Ordinary TVs and most existing screens
    B

    App on a commercial display

    Our player installs directly on signage-grade panels and starts on boot.

    Samsung, LG, Sony and Philips commercial displays
    C

    App on an LED controller

    Our player runs on the Android-based controller that drives the LED wall.

    Indoor and outdoor LED walls and billboards
    D

    Web player inside another system

    Our player runs as a web page inside a network's existing signage software.

    Established networks that keep their own software

    Screen types

    Also supported: video walls, interactive touch kiosks with touch events, and shelf-edge and stretched bar displays in retail aisles.

    Player box, radar module, camera, light sensor and edge computer

    System configurations

    Start with the base.
    Add measurement screen by screen.

    The system is modular. Nothing already installed has to be replaced when a screen moves up a tier.

    A typical network mix

    Most networks do not need the Attention configuration everywhere. A sound design puts Measured on every indoor screen and Attention on a sample of screens in each venue type, enough to calibrate the rest.

    Privacy by design

    Sensor in. Numbers out.

    One rule runs through the whole system: only counts and durations ever leave the screen.

    • All sensor analysis runs on the device at the screen.
    • No face template is created, so no person can be recognised or matched.
    • No person is linked across screens or across days.
    • Very small counts are suppressed so no individual can be singled out.
    • Every screen with a sensor carries a visible notice.
    • Consent is asked only when a person chooses to scan and share their details, on the advertiser's landing page.
    • The under-18 age estimate is never used to choose an ad.

    What the system does not claim

    • It does not identify any individual.
    • It does not track eye movement.
    • It does not count people through walls or by WiFi.
    • It does not follow people from one screen to another, or from a screen to their phone.

    Built for India's data-protection regime now, and mirroring the conditions European regulators set for anonymous audience measurement on advertising screens.

    How to work with us

    A ready solution,
    or built around your network.

    Stackbinary is a software development company. This platform is something we have designed end to end, and we deploy it the way your business needs it.

    01

    Deploy the full platform

    Player, ad server, screen registry, owner portal, QR attribution, dashboards and billing, hosted and supported by us. Start with the Core configuration and add measurement as you grow.

    02

    Add it to an existing network

    Where you already run signage software, our web player and ad server sit inside it. You keep your screens and content system and gain ad filling, proof-of-play, QR attribution and experiments. Sensors can be added later.

    03

    Custom development

    New screen types, new sensors, a specific exchange, your CRM or till system, a custom advertiser portal. We build the integration, on fixed scope and with weekly working demos.

    What we provide

    Software

    Design, build, hosting and support for the full system.

    Hardware integration

    Hardware specification, integration and quality control for players and sensors.

    Rollout

    Installation guides and remote device management.

    Measurement study

    A pilot that validates sensor counts against manual counts, with the method published.

    Next step

    Show advertisers proof
    your competitors cannot.

    A thirty-minute call on your screens, your venues and your advertisers. A written fixed price within days, and an honest no if it is not worth building.

    Book a Discovery Call →