How to switch agencies without your ROAS falling off a cliff

The fear is real; the risk is manageable

The single biggest reason brands stay with agencies they've stopped trusting is fear of the transition: the learning phases will reset, the numbers will crater, and three months of momentum will burn while the new team finds its feet. The fear isn't irrational — botched handovers do exactly that. But almost all transition damage comes from three avoidable mistakes: assets you didn't control, knowledge that left with the old agency, and a new team that rebuilt everything to look decisive.

All three have checklist-shaped solutions, which is what this post is. If you're still deciding whether to switch at all, start with when firing your agency is justified and why D2C brands actually switch — this guide assumes the decision is made and the job now is executing it without bleeding ROAS.

Before you give notice: secure your assets

Do this quietly, before any awkward conversation, because your leverage is highest while the relationship is officially fine. Verify — by logging in yourself, not by asking — that you have owner-level control of everything below. Budget one to two weeks for it; agencies rarely obstruct, but access untangling always turns up a surprise:

• Meta Business Manager, ad accounts, pixel/datasets and pages — owned by your BM, agency attached as partner.
• Google Ads account and its history (the account itself, not exported reports), GA4 and Tag Manager.
• Marketplace ad consoles — Amazon, Flipkart, quick-commerce seller panels.
• Creative source files, not just exported MP4s: working files, fonts, brand assets, UGC raw footage and usage rights.
• Custom and lookalike audiences, customer lists, product feeds and catalogue setups.
• The promo calendar, past performance reports and any naming-convention documentation.

If any of this sits in the agency's own accounts, migrating it is your first request — framed as routine housekeeping. An agency that resists moving assets you paid for is confirming your decision to leave.

The overlap month: run old and new in parallel

Use the notice period as an overlap month rather than a cliff. The old agency's brief shrinks to maintenance: keep campaigns running, no new structures, no experiments, flag anything unusual. The new agency's brief is to audit silently — read access to everything, no changes yet — and produce a handover assessment: what's working, what's fragile, what they'd change and in what order. That assessment doubles as transition insurance — if it comes back shallow, you've learned something important about the new agency while it's still cheap to act on.

Freeze major account changes for the whole overlap: no campaign restructures, no pixel or attribution changes, no landing-page migrations. Every frozen change is one less confound when you're trying to read the new team's first month. And insist on one structured knowledge-transfer call between both agencies with you in the room — awkward for twenty minutes, worth lakhs later. Most outgoing agencies behave professionally here; their reputation is the asset they're protecting.

The week-by-week checklist

A transition that protects ROAS looks roughly like this:

• Week 1: access audit complete (everything in the section above verified), baseline snapshot recorded — 30/90-day spend, revenue, MER, CAC, top campaigns and creatives — so "worse than before" has a number attached.
• Week 2: knowledge-transfer call done; naming conventions, audience logic and test history documented; new agency delivers its audit and a 90-day plan that explicitly lists what it will NOT touch.
• Week 3: new agency takes over live management. Winning campaigns continue untouched; only budget pacing and obvious waste get corrected.
• Week 4: first new tests launch alongside — not replacing — the incumbent winners, sized small enough that failure is information, not a crater.

The pattern underneath: control what changes, one variable at a time, with a baseline to compare against. Transitions fail when everything changes at once and nobody can say what broke. Print the checklist, literally — a transition run from a shared document survives people being busy; one run from memory doesn't.

The new agency's first 14 days: audit, don't rebuild

The most common transition killer is the ego rebuild: the incoming team torches the old account structure to demonstrate decisiveness, resets every learning the account had accumulated, and spends your next two months relearning what the last agency already knew. Campaign history, audience seasoning and creative learnings are assets; a rebuild liquidates them at zero.

What a professional first fortnight looks like: reconcile tracking against order data, fix genuine breakage (broken pixels, absurd frequency, budget bleeding into dead placements), document the account's current logic, and leave winners alone. Structural change comes later, deliberately, one hypothesis at a time. If your incoming agency's plan for week one is a full rebuild, that is itself a red flag worth reading about before you hand over the keys.

What's a normal dip — and what's an alarm

Expect some turbulence: a 10–20% efficiency dip for two to three weeks is normal as budgets re-pace and small edits ripple through learning phases. Judge the new team on trajectory and process during this window, not on absolute numbers — the realistic results timeline applies to transitions too. Hold the cadence tighter than usual through this window; fifteen minutes daily for the first week is not overkill on a meaningful account.

Alarms are different: conversion tracking that stops reconciling with orders, spend falling off a cliff because access wasn't sorted, winning campaigns paused without discussion, or a drop past 40% that's still there after week three. Each of those is a process failure, not transition physics — raise it the day you see it. And if you want a neutral baseline before you start any of this, a free Growth Audit documents your account's true pre-switch state, so both agencies are measured against the same facts.