In the last article, we already wrote about how Dentsu is engaged in forecasting the television viewing of advertisements on TV and what algorithms we use for this. In this article, we will tell you how this forecast is used to automatically place advertisements on TV, how the very idea of ββProgrammatic TV came about, what difficulties we encountered in its implementation and how it lives and works now, in a difficult from all points of view 2020.
A bit of history
Back in 2014, we started thinking about automating the processes that the agency's advertising group faces when purchasing advertising on TV.
, , () . , , -, .
2016 - () β (), - . API Vimb.
API ( -, ) . API .
, API . API, Programmatic TV .
, .
, , , . 2018 -.
, , , .
:
β , ( , - , , , . ., , - )
β , . ,
TVR β ,
GRP β , TVR . GRP
Affinity () β , , , (TVR / TVR )
CPP (cost per point) β . - ( ). CPP
- β 19:00 24:00 8:00 24:00
β -
β ( ),
β , .
β , .
, - . Vimb, , .
Vimb . () . Vimb , .
. . , . .
Vimb API , , , , , .
:
() β , .
() β , , . .
() β . . .
β Β« Β». .
Flex β , .
, , .
, , , ( , , , , ), . .
60-80% . , , .
Programmatic TV, .
, , 247 . . , .
,
, , , , ββ .
Data- , . , , TVR () β , GRP β , , , , β . , , , β .
, Β« Β» . GRP , , . , , , , ( ). , , , .
, , , . , ( - , 19 23) - , , , .
, : Β« , , Β»? .
. ?
β ( .)
, , β . .
, β . «» .
, (Customer Development). -, , , .
-, , , «» , , .
, .
, , , . :
, -
/
, , , ,
/
, . . ,
, .
. , . %
, , , - , . , , , .
, , . , , , . , :
1 β
.
2 β
( ) . .
3 β Β« Β»
,
4 β
, , , . , , ,
5 β
. ,
, .
,
, , , .
«».
β .
. . , , , .
, , , , .
, API β β, , , , . - , .
, : , . , , .
, , 247, .
?
, , Microsoft. Programmatic TV .
Programmatic TV , .
MS SQL.
gRPC.
WPF .
SSO.
, elastic, , .
:
API Vimb β β , , , , ,
Vimb. Vimb API Vimb
.
, . ClickHouse
. β , . . , , , J.
TV Programmatic
, |
249 |
|
24000 , 19000 , 265000 |
|
4732657 , Programmatic TV 3535984. 75% , |
TV Programmatic |
11- |
Programmatic TV , |
5 - 10% |
, Ptorgammatic TV |
40-50%
|