See where every channel stops paying back.
A response curve for each channel, fitted on your own spend and sales. Where you sit on it today, where the returns flatten, and how much the next ten thousand still brings.
Invitation only while in beta. Tell us who you are and we get in touch.
Response curves, four channels
weekly spend against weekly sales, sample marketThe dot is where each channel spends today. An illustration on one of our sample markets: your file draws its own curves, each with its confidence band and its verdict.
One card per channel
The curve, the point you are on, and the three numbers that matter. Read in a glance, defended in a meeting.
The next ten thousand, channel by channel
The same curves, read at today’s spend. What one more ten thousand returns in each channel, ranked. The ranking is where the budget conversation starts, and the budget allocation is where it ends.
A channel at the top is still climbing its curve. A channel at the bottom is funding the flat part, and has been for a while.
Return on the next €10k a week
at today’s spendWhat the analysis hands back
A curve, not a rule of thumb
Hill or Michaelis-Menten, fitted on the weeks you actually ran, with the band around it. Not “TV saturates around a million” because someone said so in 2019.
Three numbers per channel
The half-saturation point, the ceiling, and the marginal return at today’s spend. Together they say whether the next euro is still working.
A verdict on the fit
A curve fitted on three spend levels is a guess, and TEA says so before the run. The number of distinct levels, not the number of rows, is what the curve stands on.
The same curves feed the budget
Nothing is re-estimated downstream. The curves you read here are the ones the allocation is computed on, so the recommendation and the diagnosis never disagree.
What your file needs
A date
One row per week. A year is enough for a channel that moved; two is better.
week date
The number the spend should explain
Sales, units, leads. One column.
sales units
Spend per channel
In euro, GRPs or impressions, one column each. The more distinct spend levels across the
weeks, the better the curve. tv_grp search_spend
A channel that spent the same every week has no curve to read, and the Data Scout tells you before you pay for the run.
How it works
-
Drop the file
CSV or Excel. The Data Scout maps date, KPI and the spend columns, and counts the spend levels behind each channel.
-
Choose the shape
Hill for channels that need a minimum presence before they move, Michaelis-Menten for the ones that answer from the first euro. The trade-off is explained before you choose.
-
Read the curves
One card per channel, the ranking of the next ten thousand, the export. Then the budget, on the same curves.
What a saturation curve is, in a paragraph
Every channel returns less per euro the more you put into it in a week: the first spots reach people who had not heard, the last ones reach the same people again. A saturation curve is that relationship, fitted on your own weeks of spend and sales. It rises steeply, then flattens, and where your current spend sits on it is the whole question: on the climb, the next euro is still working; on the plateau, it is funding reach you already have.
The curve has a half-saturation point, the spend at which the channel has given half of what it can, and a ceiling. Those two numbers, and the slope at today’s spend, are what a budget allocation is computed from. TEA fits them with the band around them and declares how many distinct spend levels the fit stands on, because that, and not the number of rows, is what makes a curve trustworthy.
What is a saturation curve?
The relationship between what you spend on a channel in a week and what it returns, which rises steeply at first and then flattens. The curve has a half-saturation point, where the channel has given half of what it can, and a ceiling. Where your current spend sits on it tells you whether the next euro is still working.
Hill or Michaelis-Menten?
Michaelis-Menten is the simpler shape, concave from the first euro. Hill adds an S, a slow start before the climb, which fits channels that need a minimum presence before they move anything. TEA offers both and explains the trade-off before you choose.
How many spend levels do I need?
Distinct levels, not rows. A curve fitted on two spend levels can return a saturation point four times the true one with a tight interval, and nothing inside the maths protests. TEA counts the levels in your file and says so before the run.
Does a channel need to go dark?
No. The curve is read from the variation in spend across weeks, and an always-on channel with a flighted budget carries enough of it. A channel that spent exactly the same every week carries none, and you are told.
What about TV in GRPs?
GRPs work as the spend axis. With a cost per GRP the curve is drawn in money, which is the axis a budget conversation is held on.
Where it goes next
Budget allocation
The curves, turned into the allocation that returns the most within your budget, with the reason for every move.
See the allocationMarketing mix modeling
What every channel, promotion and price move contributed to the year, week by week, with the baseline underneath.
See the modelStop funding the flat part of the curve.
Drop the file, read the curves, move the money. Free while in beta, by invitation.
Join the beta