Choosing an agency at ₹25 lakh+ a month: the rules change

₹25 lakh a month changes what "good agency" means

At ₹3L/month of ad spend, a mediocre agency wastes lakhs. At ₹25L+, a 10% efficiency gap is ₹2.5L a month — ₹30L a year — compounding quietly while the dashboards look fine. The hiring question changes from "who can run Meta ads" to "who can prove incrementality, produce creative at volume, and coordinate five channels without the numbers double-counting each other." Most agencies that were excellent for you at ₹5L/month are structurally unable to answer that question, through no fault of their own.

You also stop being an easy account. At this spend you likely have some in-house capability, a finance team that questions attribution, marketplace revenue muddying measurement, and board or investor reporting on a fixed calendar. The agency has to slot into that machine and defend its numbers inside it — not sell you a prettier dashboard. Hire for that reality, not for the pitch meeting.

The four capabilities that actually matter now

Filter your shortlist on these four, in this order:

• Incrementality, not attribution. Platform-reported ROAS double-counts badly at scale. Ask how they run geo holdouts or spend-pause tests, and whether they report blended MER alongside platform numbers. An agency that has never run an incrementality test has never managed real scale.
• Creative production at volume. ₹25L+ of monthly spend burns creative fast. You need a system shipping 20–40 tested variants a month with a named creative strategist attached — not "our design team handles it".
• Multi-channel orchestration. Meta, Google, marketplaces and quick-commerce interact. Budgets should move between them monthly based on marginal returns, not sit siloed because two internal teams don't talk.
• Finance-grade reporting. Contribution after returns, RTO and shipping, reconciled with your order data — platform ROAS and bank-account reality diverge hard once COD enters the picture.

Everything else — awards, office, client logos, follower counts — is decoration. Any of the four missing will surface as a six-figure problem within two quarters.

Interrogate the team, not the agency

At this ticket size you are not buying the agency's brand; you are buying 30–60 hours a month of specific humans. Make the pitch team name them: who plans media, who buys daily, who owns creative strategy, and who picks up the phone when a launch breaks at 11pm during a sale event. Ask how many other accounts each of those people carries — a buyer running twelve accounts cannot think deeply about yours, whatever the agency's reputation.

Then pressure-test seniority. The classic pattern at larger agencies is a director-led pitch followed by an execution team two years out of college. Put protection in the contract: named key personnel, approval rights over replacements, and a minimum of senior hours per week. Agencies confident in their bench accept these clauses without flinching; the ones that resist are telling you something useful early.

Commercial models at this spend

The market structures you'll encounter: boutique specialists at roughly ₹75k–2.5L/month or 8–15% of spend; mid-size full-service at about ₹2L–6L/month; large networks at ₹8L+ on annual contracts. At ₹25L+ of monthly spend, percentage models produce large absolute fees, so most negotiations at this level land on hybrids — and the percentage itself typically tapers as spend grows.

Watch the incentives, not just the number. A pure percentage-of-spend fee rewards spending more, not spending better. A pure flat fee rewards inertia — the agency earns the same whether it pushes or coasts. The cleanest structures pair a base retainer with a bonus tied to an efficiency metric you both trust: blended MER or new-customer CAC, measured after returns, with definitions written down. And any agency that quotes a final number before understanding your margins, channels and creative needs is pricing a template rather than your business — scope-first pricing after a discovery call is the serious-agency norm at every level, and non-negotiable at this one.

Run a paid pilot, not a beauty parade

RFP decks reward writing, not media buying. A better process: shortlist two or three agencies against the four capabilities, then pay one for a bounded 30–45 day pilot — a defined channel slice, or a full-account audit with a 90-day re-forecast. Paid matters. Free pilots attract agencies planning to recover the cost later, and you want to see how they operate when they're already hired, not when they're still selling.

Before the pilot, run the data-room test: give read access to your ad accounts and analytics and ask for a teardown — what's broken, what they'd change in 90 days, and what they'd expect the change to be worth. The quality of the questions they ask about margins, returns and repeat rates tells you more than any case study. Keep reference calls to brands at comparable spend; a glowing reference from a ₹2L/month account says nothing about how the same agency behaves with ₹25L on the line.

Red flags specific to big accounts — and the next step

Three failure modes are unique to this level. Becoming the biggest client of a small shop: heroic effort, no bench, and one resignation wrecks your quarter. Becoming the smallest client of a network: you subsidise their marquee accounts and get the junior pod. And an agency with no opinion on incrementality at your spend is guessing with ₹3 Cr+ of your money a year — politely, but still guessing.

We've managed ₹150 Cr+ in ad spend across 160+ brands over six years, so these breakpoints are familiar from the inside. We've written up how we run spend at that scale, and a separate playbook on what changes past ₹50L/month — which is where you're headed next. If you want the data-room test run on your account before any agency conversation, that's what the free Growth Audit is — read what it covers, or book it directly.