Step one · what we read
We read your file. Here is what we think it says.
We found 3,900 weekly rows, 3 Jan 2022 to 23 Dec 2024 across 25 regions, with 6 media channels and 5 other things that move sales. Check the three columns below. Everything after this screen is built on them being right.
Every column, and why we called it that
Your column
We think it is
Because
time
The week
156 of them, exactly 7 days apart, with nothing missing in between.
conversions
What you are trying to move
The series everything else is judged against. Runs $1.8M to $4.8M a week and never goes negative.
revenue_per_conversion
What one of those is worth
Turns the count above into money, so a return can be a multiple.
geo
The region
25 of them, each with a full 156-week history — so the same week can be compared across places that spent differently.
Channel0_spend
TV — Broadcast
$10.4M over the window, never negative, and zero in 92 of 156 weeks — a channel that was switched off, not a gap.
Channel0_impression
TV — Broadcast impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
Channel1_spend
Video — CTV
$5.8M over the window, never negative, and never zero — this one never went dark.
Channel1_impression
Video — CTV impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
Channel2_spend
Paid Social
$7.3M over the window, never negative, and never zero — this one never went dark.
Channel2_impression
Paid Social impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
Channel3_spend
Paid Search
$6.5M over the window, never negative, and never zero — this one never went dark.
Channel3_impression
Paid Search impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
Channel4_spend
Programmatic Display
$4.3M over the window, never negative, and never zero — this one never went dark.
Channel4_impression
Programmatic Display impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
Channel5_spend
Out of Home
$3.1M over the window, never negative, and zero in 96 of 156 weeks — a channel that was switched off, not a gap.
Channel5_impression
Out of Home impressions
Rises and falls with the spend above it, so we read it as the volume that money bought rather than as a second channel.
GQV
Category demand index
Moves on its own, not with your spending, so it can carry the demand that was going to be there anyway.
competitor_sales_control
Price index
Not media, and it moves independently of what you spent — so it gets to explain sales that no channel should be credited with.
Promo
Promotion
Not media, and it moves independently of what you spent — so it gets to explain sales that no channel should be credited with.
sentiment_score_control
Holiday
Not media, and it moves independently of what you spent — so it gets to explain sales that no channel should be credited with.
Organic_channel0_impression
Newsletter sends
Not media, and it moves independently of what you spent — so it gets to explain sales that no channel should be credited with.
When each channel was on air
TV — Broadcast
41% · bursty
Video — CTV
100% · continuous
Paid Social
100% · continuous
Paid Search
100% · continuous
Programmatic Display
100% · continuous
Out of Home
38% · bursty
2 channels go dark and come back — TV — Broadcast, Out of Home. Those weeks are the most valuable data in the file, because a channel that stops is a channel whose effect can be seen separately from everything that was still running. Channels that never stop are the ones we usually cannot put a number on.
Your 3 years, totalled
Revenue
$486.1M
Media spend
$37.5M
Spend as a share of it
7.7%
Complete rows
3,900