Digital Measurement SEI Content Viewing

Engineering Portal / Digital / Digital Measurement SEI Content Viewing
This interface specification depicts the file schema required to supply Nielsen with response-level Viewing data in a “Server-to-Server” relationship for incorporation into the Nielsen One Content and/or Nielsen One Ads products.

Delivery Specifications

Files should be delivered at a fixed cadence in a dedicated S3 bucket following the outlined in this spec prefix and object naming conventions and data formats.

• All prefix labels and file names should be lowercase except for app ids.
• Supported format are:
  • Text files with utf-8 encoding in JSON Lines format;
  • Apache Parquet format with snappy compression;
• All data files have extensions to indicate the file format (.json | .parquet)
• Data file can be partitioned into multiple parts with min size of 256MB
• Data can be delivered separately in multiple splits, if needed due to organizational, technical or privacy requirements. This allows to permission s3 access separately, to process data independently and to persist data partitioned, within the limits of the SEI system;

S3 Bucket and Prefix Naming Convention

useast1-nlsn-w-dig-sei-<partnerid>-feeds-<env>/<filetype>/<split>/yyyy/mm/dd/hh/<object>

Name	Description
partnerid	Abbreviation provided by Nielsen for each provider or publisher
env	test or prod
filetype	exposure, dcr-exposure, dar-exposure, audio-exposure, etc., where: exposure is reserved for multi-product files <product>-exposure is for single product files, where <product> is [dar|dcr|dtvr|ctvc|audio]
split	a separate data split, can be by platform (ex.: ios, browser, android, ctv), by country (us, ca, jp, etc.), by publisher, by team, etc. or “all” (if data is provided in one split).
yyyy/mm/dd/hh	yyyy - year mm - month dd - date padded with 0 example 01, 02,..., 31 hh - hour padded with 0 example 00, 02,..., 23

S3 Bucket and Prefix Naming Convention

<partnerid>_<filetype>_<intid>_<appid>_<starttime>_<endtime>.[json|parquet]

Name	Description
partnerid	Abbreviation provided by Nielsen for each provider or publisher
filetype	exposure, dcr-exposure, dar-exposure, audio-exposure, etc., where: exposure is reserved for multi-product files <product>-exposure is for single product files, where <product> is [dar|dcr|dtvr|ctvc|audio]
initid	integration id: unique identifier provided by Nielsen
appid	Nielsen-provided server application identifier
starttime	start date and hour of the data in the file in UTC Example format: yyyymmddhh
endtime	end date and hour (not inclusive)  of the data in the file in UTC Example format: yyyymmddhh

Success File

Empty _SUCCESS file should be provided to indicate that data delivery for a particular hour and split is completed (even if there is no data for that particular hour and split).

Manifest File

A manifest should be provided which contains metadata related to the uploaded file(s). The manifest is a text file in .json format that implements the AWS unload manifest file format. It has the same name as the data file, but has the _manifest suffix.

Example:

 {
   "entries": [
     {"url":"s3://bucket/prefix/0000_object_00.snappy.parquet", "mandatory":true},
     {"url":"s3://bucket/prefix/0001_object_00.snappy.parquet", "mandatory":true},
     {"url":"s3://bucket/prefix/0002_object_00.snappy.parquet", "mandatory":true}
     {"url":"s3://bucket/prefix/0003_object_00.snappy.parquet", "mandatory":true}
   ],

   "meta": {
     "schema_version": "S2SV1.7.0",
     "accreditation_status": "0",
     "start_time": "1710154800",
     "end_time": "1710158399",
     "record_count": 31337
   }
 }

The manifest contains a complete list of all files provided for a given period and split, as well as a meta data (a.k.a. header) which should include the following attributes:

Parameter	Description	Required	Specified	Format / Example	Type
schema_version	Schema Version	Yes	Nielsen	S2SV1.7.0	String
accreditation_status	Accreditation Status	No	Client	MRC = 1	String
data_start_time	Data Start Time (min)	Yes	Client	Format:32-bit unsigned int Unix time in seconds Example: 1710154800	String
data_end_time	Data Start Time (max)	Yes	Client	Format:32-bit unsigned int Unix time in seconds Example: 1710158399	String
record_count	Number of records in data file	Yes	Client	Example: 31337	Number

Data delivery example for hour 11 (11:00:00 AM UTC to 11:59:59 PM UTC):

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/acme_dtvr-exposure_a9ddf15ea054ea415718767ea6_P47C2495B-BBBA-56DE-BE99-14758F92F034_2024031111_2024031112_0000.json

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/acme_dtvr-exposure_a9ddf15ea054ea415718767ea6_P47C2495B-BBBA-56DE-BE99-14758F92F034_2024031111_2024031112_0001.json

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/acme_dtvr-exposure_a9ddf15ea054ea415718767ea6_P47C2495B-BBBA-56DE-BE99-14758F92F034_2024031111_2024031112_0002.json

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/acme_dtvr-exposure_a9ddf15ea054ea415718767ea6_P47C2495B-BBBA-56DE-BE99-14758F92F034_2024031111_2024031112_0003.json

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/acme_dtvr-exposure_a9ddf15ea054ea415718767ea6_P47C2495B-BBBA-56DE-BE99-14758F92F034_2024031111_2024031112_manifest

useast1-nlsn-w-dig-sei-acme-feeds-prod/dar-exposure/ios/2022/05/25/11/_SUCCESS

SLA

The files must be delivered into the proper S3 bucket within 3 hours of the start of that hourly viewing file interval. For example, files from 1:00 AM to 2:00 AM must be delivered before 4 AM.

Accuracy of Measurement

The reported “wall clock” time of viewing needs to be within 25 seconds of the actual viewing time.

The reported content “reference” time needs to match the actual content that was played out with plus/minus 10 seconds (for clarity: start and end times can each be off by up to 10 seconds, so the combined under/over reporting for each individual viewing segment should be no greater than 20 seconds)

CTV, Mirroring, and Casting

For apps native to the OTT device (i.e. downloading and viewing a streaming app to an Apple TV), audit ping should fire from the OTT device, and Viewing data should reside in OTT Viewing file.

For mirroring, where video playback occurs on the mobile device and OTT device, only one Viewing file row is necessary where, if possible to determine, set: "secondscr":"MIR" and include in respective mobile platform Viewing file.

For a casting scenario where content is controlled via the mobile device, but displayed only on the OTT device such as an AppleTV or Chromecast, an audit ping must fire from the mobile device before the casting occurs and at the end of playback from the mobile device only. Viewing data resides in the respective mobile platform Viewing file.

{...
   "sessionid":"ABC",
   "streamid":"DEF",
   "position":[
      {
         "referencestart":"xxxxxxx123",
         "referenceend":"xxxxxxx183",
         "playheadstart":"0",
         "playheadend":"60"
      },
      {
         "referencestart":"xxxxxxx243",
         "referenceend":"xxxxxxx303",
         "playheadstart":"120",
         "playheadend":"180"
      }
   ]
}

{...
   "sessionid":"ABC",
   "streamid":"DEF",
   "position":[
      {
         "referencestart":"xxxxxxx183",
         "referenceend":"xxxxxxx243",
         "playheadstart":"60",
         "playheadend":"120"
      }
   ],
   "secondscr":"OTT"
}