Strands of fiber-optic cable glowing blue in the dark
Technology

An exchange built to be audited.

Our own exchange and ad server, with HUMAN scanning in front of the auction and machine learning behind the yield. Here is what happens to every request.

The request path

Four stages, every impression.

  1. 01 · Receive

    Direct supply

    Requests arrive from publishers we contract with directly, through VAST, oRTB, our SDK or Prebid.

  2. 02 · Scan

    HUMAN pre-bid

    Each request is checked by HUMAN before any bidder sees it. Invalid traffic stops here.

  3. 03 · Optimize

    ML floors and routing

    Our ad server learns floors per placement and routes each impression to the demand most likely to pay for it.

  4. 04 · Report

    Shared numbers

    Revenue, fill and demand source, broken out by app, channel and placement, visible to both sides.

Supply chain

Read the proof in the request.

Every request we send carries a complete schain object with AdSphere360 as the selling node. Combined with our public sellers.json, buyers can trace the full path from their bid to the screen.

Example bid request 
{
  "imp": [{
    "video": {
      "mimes": ["video/mp4"],
      "maxduration": 30
    }
  }],
  "app": { "bundle": "example.fast.channel" },
  "device": { "devicetype": 3 },  // CTV
  "source": { "schain": {
    "complete": 1,
    "nodes": [{
      "asi": "adsphere360.com",
      "hp": 1
    }]
  }}
}
{
  "imp": [{
    "banner": { "w": 320, "h": 480 },
    "instl": 1
  }],
  "app": { "bundle": "com.example.puzzle" },
  "device": { "devicetype": 4 },  // phone
  "source": { "schain": {
    "complete": 1,
    "nodes": [{
      "asi": "adsphere360.com",
      "hp": 1
    }]
  }}
}
{
  "imp": [{
    "video": { "plcmt": 4 }
  }],
  "site": { "domain": "example-publisher.com" },
  "device": { "devicetype": 2 },  // desktop
  "source": { "schain": {
    "complete": 1,
    "nodes": [{
      "asi": "adsphere360.com",
      "hp": 1
    }]
  }}
}
Pre-bid IVT scanPassed
Supply chainComplete
Seller typeDirect
A bright data hall with rows of server cabinets
Yield

Floors that learn per placement.

A single floor across an app leaves money on the table. Our ad server learns what each placement clears at and adjusts continuously, so publishers earn more and buyers see fair prices.

  • Floors tuned per placement, not per account
  • Routing to the demand most likely to win and pay
  • Changes visible in reporting, never a black box
Talk to our team

Want the technical detail?

Our AdOps lead can walk you through endpoints, specs and the scan setup on a call.

Book a technical call