Budget Optimizer
A solver that proposes a per-channel budget allocation maximizing predicted total ROAS subject to constraints.
Last updated
Definition
A typical implementation pairs SLSQP (Sequential Least Squares Programming) for the smooth case with Differential Evolution as the global fallback for non-convex landscapes. Inputs: per-channel posterior ROAS curves from MMM, total budget cap, per-channel min/max constraints, an exploration premium for under-sampled channels. Output: a vector of channel budgets with the expected lift over the current allocation.
How it applies in India
Around Diwali and other festive peaks demand shifts quickly; WatEase’s limits on how far budget moves at once, 20% a change and 10% of the total in a month by default, keep a short spike from swinging the whole plan.
An example
A brand spends ₹1 lakh a week across five campaigns. Two are still on the steep part of their response curves; three have flattened. Moving ₹15,000 from the flat campaigns to the steep ones can lift sales without spending a rupee more, which is exactly the move an optimiser looks for.
How WatEase uses it
WatEase’s budget planner in the Marketing add-on solves the split directly: it searches for the single level of marginal return at which every campaign’s last rupee earns the same, using each campaign’s saturation curve. Changes are bounded: no more than 20% up or down at a time (a business can allow up to 50%), no more than 10% of the total budget moved in a month, and never below a platform’s minimum. Recommendations from the marketing-mix model go to the Plan & Decisions queue to be approved and applied; in watch-only mode they stay advice. It is one of the jobs behind WatEase’s agentic marketing.
Frequently asked questions
Does WatEase’s optimiser move budget automatically?
Only within limits you can see. By default a change needs 85% confidence to apply on its own, and no single change exceeds 20%. Shifts suggested by the marketing-mix model wait for approval, and watch-only mode turns every change into advice.
Related terms
- Saturation CurveA transformation that models diminishing returns — doubling spend doesn't double conversions; the curve flattens.
- Thompson SamplingA Bayesian way to split traffic between options — creatives, templates, audiences — that sends most of it to the likely best one while still testing the rest.
- Marketing Mix Modeling (MMM)A statistical method that quantifies how each marketing channel contributes to a sales outcome over time, using historical spend + revenue + exogenous variables.
Try WatEase
Run your WhatsApp commerce on the platform built for India — Cloud API, GST invoices, UPI checkout, opt-in tracking, and 0% markup on WhatsApp messages (competitors add 20–25%). Pricing is on request — quoted by the WatEase team.
Get Started Today