Marketing mix modelling
What every channel, online and offline, adds to sales, after seasonality, festivals and discounts.
- Spend ceilings per channel
- Carry-over effects of YouTube, TV and print
- Refreshed monthly
Find out which ad spend makes money.
Mix modelling, incrementality tests and budget planning across Google, Meta, YouTube, marketplaces and offline, for brands anywhere. Prices in writing before you start, and our founder, Anil Bhandarkar, on the first call.
- Meta Ads says
- 420
- Google Ads says
- 380
- Together they claim
- 800
- You actually sold
- 600
Brands we work with
Fashion, food, education and lifestyle brands we’ve filmed and marketed.
Jayalaxmi Silks
Anagha Silks
Ladyhill School
Moto Clownz
Bento Cakery
Brownie Point
The Noodles Cafe
D'ill Coffee House
Why your dashboards disagree
Each platform grades its own homework. Add them up and they usually claim more sales than you made.
Every platform claims the same sale
A customer sees a YouTube ad, searches on Google, clicks a Meta retargeting ad and buys. All three report the sale as theirs.
Averages hide where spend stops paying off
Your overall return can look healthy while the spend you added most recently returns far less. The average doesn’t show the difference.
Offline and slow sales go missing
Store visits, phone orders, site visits for real estate and long decisions never reach the pixel, so the channels that drive them look weak.
Average ROAS says 2.1x. This week’s new spend returned 0.3x.
ROAS, or return on ad spend, is the revenue each unit of ad spend brings back. The overall average can look fine while your newest spend returns far less than it used to. Mix modelling tells you which number to act on.
- Average ROAS:
- blended return on all spend to date.
- Marginal ROAS:
- return on just this week’s new spend.
One measurement system, four parts
No single method is right on its own. Mix modelling sees the big picture, experiments prove the biggest calls, and attribution guides the day-to-day. Each part checks the others, so you plan with numbers finance can defend.
Incrementality and geo-lift tests
Switch spend on or off in chosen cities or audiences to prove what a channel really causes.
- Test design and city selection
- Results fed back into the model
- Clear go or no-go call
Budget planning and forecasts
Next quarter’s split by channel, with the expected range of sales for each scenario.
- What-if scenarios
- Caps and floors you set
- Plan vs actual every month
Attribution and repeat customers
Fair credit across the steps before a sale, and which channels bring customers who come back.
- Beyond last click
- Repeat rate by channel
- Marketplace and quick-commerce sales included
Two ways to start
Most brands start with the ad spend check. If the data supports it, we move to full measurement. We confirm the minimum spend and quote in your currency on the first call.
For brands spending a modest amount on ads each month
Ad spend check
A one-off project to find wasted spend and the quickest budget wins.
- Best for
- Two or three channels, mostly Google and Meta
- Data
- 6–12 months of ad spend and enquiries or sales
- Method
- Enquiry matching, holdout or before-and-after test
- Time
- About three weeks
- Price
- Fixed fee, quoted after a 30-minute call
For brands spending more, across several channels
Full measurement
Marketing mix modelling on your full history, refreshed every month and planned with you every quarter.
- Best for
- Several channels, online and offline, or many stores or cities
- Data
- 1–2 years of weekly spend and sales
- Method
- Mix modelling, geo-lift tests, attribution, retention
- Time
- First results in 4–6 weeks, then monthly
- Price
- Project or monthly retainer, quoted after a 30-minute call
Is analytics worth it for you yet?
Four questions. You’ll see which option fits, or whether to wait. No email needed.
Your result
Answer all four to see what fits
Takes about a minute.
How a project runs
Day 1
Scoping call
30 minutes on your channels, spend and the decision you need to make.
Week 1
Data access
You share access or exports of your ad spend and sales data.
Weeks 2–3
Model and check
We clean the data, build the model and test it on weeks it hasn’t seen. Full measurement takes a little longer.
Week 3
Walk-through
Anil and our analyst go through the findings with you and agree what to change. For full measurement, this is in weeks 4–6.
After that
Act and re-measure
Run the changes and the test plan, then measure again.
What we need from you
Messy spreadsheets are fine. Cleaning the data is part of the work.
- Ad spend by channel and week, for Google, Meta, YouTube and anything else you run.
- Sales, leads or enquiries by week, from Shopify, your CRM, POS or a spreadsheet.
- Offline marketing, like hoardings, print, radio, events or TV, with rough dates and costs.
- Prices, discounts and promotions, so sale periods aren’t mistaken for ad performance.
- Anything unusual, like stock-outs, new stores, launches or site downtime.
What a report shows
Three pages from an example report, built on made-up numbers for an imaginary brand, so you can see the format before sharing anything.
Page 1
What each channel added to sales
Base sales, no ads15% 12–18
Google Search31% 26–36
YouTube22% 17–27
Meta prospecting19% 15–23
Hoardings9% 5–13
Meta retargeting4% 1–7
Share of sales, with the likely range beside it. Adds to 100%.
Page 2
Where extra YouTube spend stops paying off
Page 3
Accuracy scorecard
We hide the last 8 weeks from the model, then ask it to predict them. You see how close it got before you trust any finding.
- Average weekly error
- 6%
- Weeks in range
- 7 of 8
- Verdict
- Safe to plan with
How we keep it honest
We’d rather show our work than ask for trust. Every finding points back to a method you can read, and where we don’t have proof yet, we say so instead of dressing up an example as a result.
Ranges, not magic numbers
Every result shows how sure the model is. Where data is too thin, we say so.
Checked before you see it
The model must predict weeks of sales it hasn’t seen before we share any finding.
Confirmed with real tests
Big findings get a simple experiment, like pausing a channel in a few cities.
Method in writing
Data used, assumptions and limits, shared with every report.
Your data, deleted on request
Kept for one year after the contract, or deleted within 30 days if you email us.
Bad news included
If a channel, or our own campaign, isn’t working, the report says so first.
Who does the work
“Our clients kept asking which part of their ad spend was actually working. Analytics is how we answer that with evidence, not opinions.”
Has run Plotme since 2024. Before that, two years leading delivery for 25+ enterprise accounts at AVRL. Anil works with a data analyst on every project: the analyst builds and checks the models, using our own proprietary model alongside open-source mix modelling tools, and Anil reviews every finding and walks you through it. We don’t publish our analysts’ names; you meet yours when the project starts.
Analytics questions
Something else? WhatsApp Anil.
Will this just tell me what Google and Meta already report?
No. Those reports count every sale an ad touched, including sales that would have happened anyway, and each platform gives itself the credit. We start from your actual sales and work backwards, so the number is usually lower than the platform shows, and more useful.
How long does it take?
An ad spend check takes about three weeks from getting your data. Full measurement gives first results in four to six weeks, then a refresh every month.
How much do we need to spend on ads?
Enough to get a reliable read: a modest monthly budget is enough for an ad spend check, and a larger budget across several channels supports full mix modelling. We confirm whether your spend is enough on the first call, based on your channels and how much data you have. If your spend is too small for a reliable answer, we’ll tell you so.
Who will I talk to?
Anil is on the first call. After that, Anil and the analyst on your project go through every report with you. We don’t publish our analysts’ names; you meet yours when the project starts.
See where your spend stops paying off
Thirty minutes with Anil. You’ll leave knowing whether analytics is worth it for you, which option fits and what it would cost.
- 1. You send this form. Anil replies within one working day.
- 2. A 30-minute call on your channels, spend and the decision you face.
- 3. A written scope and price. No obligation.
- Call or WhatsApp
- +91 78925 26156
- anil@plotme.io