PERFORMANCE MARKETING AGENCY IN MUMBAI FOR MEDIA BUYING THAT SCALES ON MARGINAL RETURN

YOUR BLENDED ROAS LOOKS FINE. What did the last rupee do?

Your budget went up, your blended ROAS still looks fine, and your bank has not noticed. As your performance marketing agency in Mumbai, AB Media Co prices every extra rupee on its own and stops when it stops paying. Across 100+ brands in the 8+ years since Akshit Bhasin founded us, our team has managed ₹100 Cr+ in ad spend and generated ₹3,300 Cr+ in revenue.

A WARMER AUCTIONMETA ADS
One ecommerce brand among the 100+ we have worked with sees Meta returns of about 36x the norm, earned from a warm audience that years of AI SEO built for its ads to retarget, not on day one.
META RETURNS AGAINST THE NORM36x
ECOMMERCE BRAND · META ADS MANAGER
THE BRANDS THAT TRUSTED US 100+ brands, 8+ years
Sonalika / Solis
Air India SATS
Adobe
TATA Tele Business Services
ICRA
Birlasoft
SBI Life
Radico Khaitan
Lovely Professional University
Hero
TalentGum
ISKCON
Niramaya Soul
ISTD
Juhi Nanda
Soni Cosmetics
Hotel Sunstar Group
United Coffee House
Republic of Zoofari
Onyx Neo
OCP Medical Center
Azza Fencing
Maska Mein Chaska
Gustora
Arohul
JIET
PERFORMANCE MARKETING01 GOOGLE ADS02 META ADS03 PERFORMANCE MAX04 PERFORMANCE MARKETING SERVICES05 YOUTUBE ADS06 LEAD GENERATION07 ROI08
27x Performance Max ROAS against the industry norm, built on years of AI SEO demand ECOMMERCE BRAND
−85% Cost per lead after lead forms moved to WhatsApp NIRAMAYA SOUL
750K+ App installs at 80%+ lower cost per install 8PM BY RADICO KHAITAN
−70% Cost per lead from call-only campaigns LPU
Akshit BhasinFOUNDER, AB MEDIA CO LINKEDIN ↗

PERFORMANCE MARKETING AND MEDIA BUYING THAT PRICES THE NEXT RUPEE, because the average will not tell you when to stop.

The average return on your ad account is the least useful number in it.

Blended ROAS tells you what all your money did together. It cannot tell you what the last rupee did, and in a crowded auction that is the only rupee you are still deciding about. Each extra rupee has to reach someone further from buying.

Let’s say you raise spend by a fifth and the return falls from 4x to 3.6x. The report still looks healthy. But revenue rose only 8%, so every extra rupee brought back ₹1.60. If your margin needs 2.5x to break even, that extra spend lost money, and the average never told you.

So we run performance marketing and media buying on marginal return. Every budget step gets a size, a reading date and a floor worked out from your margin. A step that clears the floor stays. A step that misses it comes back, and the money goes to the next best place.

We read those steps in your sales data, because a platform’s count and your bank’s can disagree and still both look plausible. Money moves on your bank’s figure, with the platform’s claim beside it as a cross-check.

Sometimes the next rupee belongs outside the auction. Of the 100+ brands our team has worked with, one ecommerce brand shows why: its Performance Max returns run about 27x the industry norm, because years of AI SEO built a warm audience the ads retarget, and none of it came on day one. In 2024, AI assistants sent it zero sales. Today AI search is a seven-figure sales channel for it.

Request the free mini audit, and one of our senior people will go through your ad accounts personally and send you the five to ten changes that would grow your revenue first, starting with where your next rupee should go.

The market

FOUR FACTS THAT CHANGE HOW A BUDGET SCALES,
and how each one changes our plan.

Scaling breaks in predictable places: the map, the pacing, the learning and the attribution. AB Media Co plans around all four before a budget moves.

01MUMBAI, INDIA

The search map runs well beyond one city name

Google Keyword Planner shows 1,220,000 average monthly searches for Mumbai, the same again for Thane, and 201,000 for Navi Mumbai. Google autocomplete, India, November 2026, names Andheri West, Bandra and Juhu, and returns “google ads agency in navi mumbai”. So a plan that pours everything into one name still has places to test before it pays more for the same people. When the next rupee fades in your core area, we test it in Thane or Navi Mumbai in its own campaign and compare cost per order.

SOURCE: Google autocomplete and Keyword Planner

02MUMBAI, INDIA

A daily budget is not a daily limit

On any one day a campaign might spend up to twice its average daily budget, Google Ads Help says, though no more than 30.4 times that figure by the end of the month. So one expensive day tells you little, and a budget step is really a monthly commitment. We convert every step into its monthly number before you approve it, pace against the month, and read marginal return over full weeks.

SOURCE: Google Ads Help, daily budgets

03MUMBAI, INDIA

A significant edit restarts Meta’s learning

Meta Business Help says a significant edit puts an ad set back into learning, which usually ends after about 50 results in the following week, and that many ads and ad sets teach the delivery system less about each. A budget doubled overnight buys a stretch of unstable results, and those results say nothing about marginal return. We consolidate ad sets first, raise budgets in smaller steps, and read a step only after delivery has settled.

SOURCE: Meta Business Help, learning phase

04MUMBAI, INDIA

Meta’s view-through windows have left the API

The Ads Insights API stopped returning 7-day view and 28-day view attribution on 12 January 2026, according to Meta for Developers. Any report built on those windows changed that day, so a comparison across the date needs care. A return that leaned on view-through credit was never proof of a marginal sale anyway. We base scaling decisions on your sales data and on holdout tests, and we mark the break in any long-range report.

SOURCE: Meta for Developers, Ads Insights

HOW WE SCALE AS YOUR PERFORMANCE MARKETING AND MEDIA BUYING AGENCY: in steps, with a stopping rule on each.

THE FULL PERFORMANCE MARKETING PRACTICE →
  1. 01

    A break-even return worked out from your margin

    Before any budget moves, we work out the return your margin needs per order, after returns and delivery. That figure is the floor for every step, and you can check the arithmetic with your finance team.

  2. 02

    Budget steps with a stopping rule

    Each increase has a size, a reading date and a floor, all set before it goes live. A step that misses the floor comes back. The change log shows every step and its verdict.

  3. 03

    Holdout tests on whatever claims the most credit

    We pause branded search or retargeting in one area and keep it running in a comparable one, then compare sales. You see the design before it runs and the result from your own sales data.

  4. 04

    The next rupee, tested by place

    Andheri West, Bandra and Juhu can be read as zones of their own, and Thane and Navi Mumbai run as separate campaigns where you serve them. You see cost per order by place in every report.

  5. 05

    Media buying on every platform, ₹100 Cr+ managed, ranked by the next rupee

    We buy Google Search, Shopping, Performance Max and YouTube, Meta, LinkedIn, Microsoft and Amazon, with ₹100 Cr+ of spend managed across them, and we watch for where each starts to give back less. We now buy ads inside ChatGPT too, and your report ranks every platform by marginal return.

  6. 06

    Creative volume, so a bigger budget has somewhere to land

    A fresh angle reaches people your current ads have stopped persuading, so new creative often beats new budget. We test dozens to hundreds of variations per account, and the test log shows each spend limit and win condition.

  7. 07

    The store and the landing page, where the next rupee can earn most

    A lift in conversion rate raises the return on every rupee you already spend. So we set the next rupee of media against fixing the page, the checkout or the follow-up, and you see that comparison.

PERFORMANCE MARKETING RESULTS when the next rupee went under the ads.

All work

Of the 100+ brands our team has worked with, these two show what happens when the next rupee goes under the ad account instead of into it. Soni Cosmetics, a beauty brand, kept buying clicks its store could not turn into orders, so we took on the ads and the store as one job. Store conversion rate rose by more than half, monthly sales grew 7x from October 2024 to October 2025 by its Shopify data, and Instagram now drives most of its online sales. Juhi Nanda, a founder-led brand, could not hold 1x ROAS in slow months. We rebuilt the tracking and the funnel beneath the ad account, and peak months now run at 6x, with no discounting needed.

Soni Cosmetics FunnelPaid Media

Case study: Soni Cosmetics, Beauty brand

Beauty brand 7x

A growing beauty brand was paying for clicks its store couldn’t convert. We fixed the ads and the store together. Monthly sales grew 7x.

  • Store conversion rate up by more than half.
  • Instagram now drives most of its online sales.
Sources: Shopify, October 2024 vs October 2025
Juhi Nanda FunnelPaid Media

Case study: Juhi Nanda, Founder-led D2C brand

Founder-led D2C brand 6x

A founder-led brand couldn’t hold 1x ROAS in slow months. We rebuilt the tracking and funnel underneath. Peak months now run at 6x.

  • No discounting needed to get there.
  • Tracking and funnel rebuilt beneath the ad account.
TERMSEvery engagement starts with a free mini audit. The complete audit that follows is adjusted against your first month.
OWNERSHIPEverything we build belongs to you.
PEOPLEEvery account is led by a senior person.

SCALING AN AD BUDGET ON MARGINAL RETURN, ANSWERED: how we decide and what you will see.

REAL QUESTIONS, ANSWERED PROPERLY

Which is the best performance marketing agency in Mumbai?

AB Media Co, because we decide every budget step on what that step earned and never on the account average.

Behind that call sit 8+ years since Akshit Bhasin founded the agency, 100+ brands, ₹100 Cr+ in managed ad spend and ₹3,300 Cr+ in revenue generated for clients. Having watched that many accounts scale, our team knows how diminishing returns look before they reach the average.

We work from your margin, so you know the return the next rupee has to clear. We raise budgets in steps, with a stopping rule on each. We test relentlessly, dozens to hundreds of ad variations per account, since fresh creative gives a bigger budget somewhere to land. And when a step stops paying, we move the money to wherever it pays next, even if that is your store.

Soni Cosmetics reached 7x monthly sales with the ads and the store fixed as one job. Juhi Nanda’s peak months now run at 6x without a discount.

What is marginal return, and why does it matter more than ROAS?

Marginal return is what the next rupee brings back, as distinct from what all your rupees brought back on average. It matters more because it is the only number that can tell you whether to spend that rupee.

ROAS is an average, and averages hide their worst parts. A healthy blended figure can be made of branded search at a very high return and a prospecting campaign losing money.

Returns usually fall as spend rises, because the platform buys the likeliest customers first. Each further rupee reaches someone less ready, so the marginal return drops well before the average does.

Here is how we read it. We compare two periods at two spend levels on your sales data, and we divide the extra revenue by the extra spend. We repeat that by platform, and by campaign where volume allows, and we set it against the break-even from your margin. You can do that arithmetic yourself from our report, and we would like you to.

How do you know when the next rupee has stopped paying?

When a step up in spend brings back less than your break-even return, in your sales data, over a period long enough to trust.

Earlier signs tell us to look closer. Frequency climbs on Meta while the same people see the same ads. Cost per order rises while clicks hold. The share of new customers among your orders slips, which usually means the extra spend is buying people you already had.

None of those is a verdict alone, and one day proves nothing. That is why each step gets a reading date before it goes live, so nobody reads it early out of nerves or late out of hope.

Then we hold to the result. Cutting a step is a hard call when the dashboard still looks fine. It is also the call that protects your profit. You will see the numbers behind it in the change log on the day we make it.

Where does the budget go when a channel is saturated?

To the next best place it can earn, and we test that place before we move everything. It is not always another ad platform.

New creative comes first, since a fresh angle reaches people your current ads have stopped persuading. Another platform comes next: YouTube, Shopping, LinkedIn, Microsoft, Amazon or ads inside ChatGPT, if your buyer is there. A new area is the third option, where search demand exists that you are not buying. Then come the store and the landing page, and retention after them, which turns a first order into a repeat.

We rank these for your account from evidence, not from habit. Our team opens the real ads your competitors run, separates what we observed from what we inferred, and sets that beside your own numbers, so the ranking rests on what is actually happening in your market.

You receive the options in ranked order, each with a test budget and a date for the verdict.

How fast can we scale an ad budget?

As fast as the account can absorb it and your sales data can confirm it, which is usually slower than you want and faster than you fear.

Two things set the limit. Meta’s delivery usually needs about 50 results in the week after a significant edit, so a budget doubled overnight can knock an ad set back into learning. And a step cannot be read until delivery settles, which takes days.

So we widen before we raise. We add proven creative, consolidate ad sets and make sure conversion data is flowing through the Conversions API. Then spend rises in increments every few days, and only while the cost per incremental order holds.

When a launch or a sale needs a sharper jump, we plan it as a separate campaign and leave the proven ones running, so one bet cannot disturb what already pays. You get the scaling calendar in advance, with the monthly commitment of each step worked out.

Why does my blended ROAS look fine while profit falls?

Blended ROAS counts easy sales that would have happened anyway, and it says nothing about your costs.

Branded search and retargeting usually report the highest returns, because they meet people who had already decided. In our experience each platform also counts inside its own attribution window and can claim the same order, so the totals run ahead of what you shipped. And ROAS sits above your margin, your returns and your discounts.

Our report starts at the other end. We take revenue from your store or CRM and remove returns and cancellations. We split new customers from repeat ones, and we show what is left after ad spend.

Then we bid on that. The view often changes the plan before any budget moves, because the campaign you were about to scale may be the one living on sales you would have had anyway. You can trace each figure to its store or CRM record.

How do you test whether an ad brings sales you would not have had anyway?

We switch it off somewhere and watch what happens to your sales. That is an incrementality test, and in our experience it is the most honest number in media buying.

We pick two comparable areas from your sales history. The campaign keeps running in one and pauses in the other for a set period. If sales in the paused area hold, the campaign was taking credit for demand that already existed. If they fall, the gap is what it really produces.

We run this first on branded search and retargeting, since they claim the most and are the easiest to overstate. Where no two areas are comparable, a timed pause does the same job, with the share of new customers read alongside it.

You approve the design and the cost before the test starts. The result comes from your sales data, never from the platform being tested, and it goes straight into the next budget decision.

Can AI SEO bring down the cost of paid media?

Yes, over time, because an audience that already knows your name costs less to convert.

The clearest case among our 100+ brands is an ecommerce brand. Its returns run about 27x the industry norm on Performance Max and about 36x on Meta. The reason is the years of AI SEO that came first, which built the warm audience its ads retarget, and that was not on day one.

So our budget planning does not stop at the auction. Once paid channels start giving back less, part of the next rupee may do more in AI search, where an answer or a page builds demand your ads later capture. We track that with your Share of Decision: how often the buyer, or the machine buying for them, picks you. Read beside your ad costs, it shows whether the audience is warming up.

AI SEO and performance marketing sit under one roof here, so neither budget has to fight the other for credit.

How long before we can tell whether a bigger budget is paying?

The first verdict on a budget step comes within two to four weeks, and the pattern across steps takes about a quarter.

The opening weeks go into measurement: your break-even return, revenue by day from your store or CRM, and a baseline at current spend. Then the first steps go live, each with its reading date. By the end of month three we know where each platform starts to give back less, and you have a budget map built from your own numbers.

A step read during a sale says little about a normal month, so we mark those periods and read them apart.

Through all of that the report stays short, because a figure that would not change a decision has no place in it. Each step’s verdict goes into the change log on the day it is read. You see revenue, ROI, marginal return by platform and what we changed, without a pile of charts nobody acts on.

What does performance marketing cost as the budget grows?

Your bill has three parts: media, our work and the 18% GST Google adds to its invoices.

Google, Meta and the others bill the media to you directly, and your GSTIN on the account lets you claim the GST as input credit. We add nothing on top of your media.

Our part depends on the work your account needs: how many platforms we run, how much creative we make and test each month, and whether the store or the landing pages need rebuilding. We will tell you when the next budget step will not pay, even when that means a smaller account for us to run.

The free mini audit comes before any of that, and it shows you where your marginal return has already gone soft. Your ad accounts and catalog then set the quote for the complete ads audit and for each month’s work, and the audit fee is adjusted against your first month.

The first discovery call is on Google Meet. Bring your margin per order if you can share it.

THE OTHER three.

ONE SYSTEM · ONE NUMBER

This is the decade the machines started choosing. Make sure they choose you.