Why Running Two Products in One Ad Account Was Capping ROAS — And How Separating Them Hit 9x

A travel accessories brand was running its hero towel and premium backpack campaigns inside the same Meta ad account. Blended ROAS sat stubbornly around 1.1–1.3x. A dedicated ad account was created solely for the backpack, with its own budget, its own pixel learning, and its own audience. Within two weeks, the backpack account was running at 9x ROAS — while the main account blended at 2.4x. The two products had always been strong. They just needed separate spaces to prove it.

March 4th daily report. Main account: 2.4x ROAS on ₹13,607 spend. Solid, but it had been sitting in that range for weeks — towel campaigns carrying it, backpack campaigns dragging the average down. Then one line in the Key Wins section: "The backpack campaigns being at 9x ROAS yesterday."

Not 2x. Not 3x. Nine times.

The backpack hadn't changed. The creative hadn't changed. The product was the same compression backpack that had been struggling to pull its weight in a shared account for two months. What had changed: seventeen days earlier, the team had moved it to its own dedicated ad account — separated from the towels entirely, with its own budget, its own pixel, and its own room to find the right buyer.

BRAND SNAPSHOT

• Industry: D2C Travel Accessories

• Category: Ultra Nanofiber Towels, Compression Backpacks, Travel Organizers, Compression Pockets

• Geography: India

• Stage: ₹0 → ₹10L+/month in 3 months

• Services: Meta Ads (Scientific Media Buying), Creative Strategy, Multi-Account Campaign Architecture

Why a Brand with Two Strong Products Still Had Mediocre Blended ROAS

This brand had two genuinely good products. The Ultra Nanofiber Towel had been running at 2.5–3x ROAS on winning creatives since January. The compression backpack had real product differentiation — premium fabric, proper compression, and a market full of plastic bag buyers who'd never experienced the upgrade.

But for the first eight weeks of the engagement, both products lived in the same Meta ad account. The account's blended ROAS reflected the average of both — towel campaigns performing well, backpack campaigns bringing it down, everything smoothed into a number that made neither product look as strong as it actually was.

The backpack wasn't failing because of poor creative or wrong targeting. It was failing because it was competing for algorithm attention with a product that had a completely different buyer profile.

The Root Cause: Different Price Points Mean Different Buyers — and Meta Doesn't Know That

The founder identified it on February 5th, weeks before the fix: "I feel target audience for backpacks and towels could be different due to price point."

It was a quiet but precise observation. A ₹999–₹1,599 nanofiber towel is an impulse-to-consideration purchase. The buyer is a casual traveler, a gym regular, someone who sees a towel that packs into a palm and thinks: why not, it's cheap and solves a real problem. They decide fast. The purchase window is short.

A ₹4,999–₹6,000 compression backpack is a different decision entirely. The buyer is planning a weekend trip, evaluating carry-on luggage options, comparing against plastic compression bags they've used before. They research. They compare. They may visit the product page twice before buying. The purchase window is longer. The audience profile is entirely different.

When both products run in the same ad account, Meta's algorithm is forced to find one type of buyer it thinks will convert — and it optimises toward the product generating the most purchase signals, which was always the lower-priced towel. The backpack's audience — frequent flyers, weekend trip planners, buyers researching carry-on solutions — was being systematically underweighted because their behaviour didn't match the algorithm's learned model of "someone who buys from this account."

Every backpack click that didn't immediately convert was training Meta away from the exact buyer who would eventually buy a backpack with slightly more consideration time.

How a Dedicated Account Gave the Backpack Room to Find Its Own Buyer

Sentinel (Scientific Media Buying): On February 17th, the team created a new Meta ad account dedicated exclusively to the backpack. No towels. No compression pockets. No other products sharing budget, audience data, or pixel signals. Starting budget: ₹4,000/day — enough for meaningful daily purchase signals without overcommitting before the account had found its footing.

Within the dedicated account, the algorithm was learning from backpack buyers only. Every add-to-cart, every product-page visit, every purchase was a signal about the type of person who buys a ₹5,000 compression backpack for a weekend trip. No towel-buyer noise. No blended conversion data pulling the model toward a different audience profile. Pure backpack signal.

Simultaneously, the main account lost the drag of underperforming backpack campaigns. The towel, compression pocket, and accessories campaigns could now optimise against their own buyer profiles — driving the main account's ROAS from a blended 1.1–1.3x toward the 2.4x it hit by early March.

Mythos (Creative Advantage): The separation also clarified the creative strategy for backpacks. With a dedicated account came a dedicated creative brief. The founder had already articulated the positioning: "Great for 'carry on luggage' and for small trips." The compression backpack wasn't a general travel bag — it was specifically for people who travel light, hate checking luggage, and want premium compression over the plastic bags they'd been using. The creative team built UGC-style content around that exact buyer: weekend trip packing, carry-on comparisons, the visible quality difference over plastic compression alternatives.

The creative could now be tested cleanly — no cross-contamination with towel audiences, no blended performance masking which backpack angles actually resonated.

Vault (Brand Value Engine): Dual-account reporting was introduced immediately. Daily performance updates tracked both accounts separately before showing a combined total. The format became a fixed part of the daily report: "Backpack account: ₹2,970 / Other account: ₹14,780 / Total spend: ₹17,750." This wasn't just accounting — it was diagnostic. The founder could see exactly how each product category was performing without one masking the other.

9x ROAS on Backpacks. 2.4x Blended. The Numbers.

• Pre-separation baseline: Backpack campaigns running in shared account — dragging blended ROAS to 1.1–1.3x over multiple weeks

• Dedicated backpack account live: February 17, ₹4K/day starting budget

• March 1: Dual-account tracking introduced — "Backpack account: ₹2,970 / Other account: ₹14,780 / Total spend: ₹17,750"

• March 3–4: Main account hits 2.4x ROAS on ₹13,607 daily spend — towel + accessories performing cleanly

• March 4 daily report: "The backpack campaigns being at 9x ROAS yesterday" — backpack-dedicated account

• 9x vs 2.4x: Same brand. Same products. Two accounts. The separation revealed the performance that was always there — buried under blended averages.

What Every D2C Brand with Multiple Products Can Learn from This

• Blended ROAS hides the truth about each product. When a ₹999 product and a ₹5,000 product share one ad account, their audience profiles blur and their individual ROAS numbers disappear into an average. A 1.3x blended ROAS might mean your towel is at 3x and your backpack is at 0.4x — and you'd never know unless you separated them. The number that matters is per-product ROAS, not the account average.

• Price point is a proxy for buyer psychology, which is a proxy for audience signal. If two products in your catalog appeal to buyers with different purchase timelines, different intent levels, or different demographic profiles — they should have separate pixel learning environments. Mixing them is asking Meta to find one audience that fits two completely different buying decisions. It can't.

• A dedicated account is the highest-signal test you can run. Every other optimisation — creative refresh, audience targeting, budget scaling — operates within the assumptions the algorithm has already formed. A dedicated account resets those assumptions entirely and gives you a clean answer to the question: "If this product had its own audience, what would it actually do?"

What Made This Harder Than Expected

• Two months of mixed-account data had to be left behind. The backpack's dedicated account started from zero purchase history. All the learning the shared account had accumulated — audience models, creative performance data, purchase signals — was irrelevant in the new environment. The first 1–2 weeks of the dedicated account were a cold start, requiring patience before the algorithm could form its own backpack-buyer model.

• Budget allocation became a weekly negotiation. With two accounts running simultaneously, the daily budget conversation now happened twice: how much for towels, how much for backpacks. When the backpack account hit 9x, the question became how aggressively to scale it without disrupting the learning phase it had just found. Scaling too fast after a ROAS spike can reset algorithm confidence — the team had to balance ambition with the discipline to let the account compound at its own pace.

What the Brand Got Wrong Before Working With Arlox

• "One account is easier to manage." The instinct is operational simplicity: one dashboard, one budget, one set of reports. But simplicity at the account level creates complexity at the performance level — you lose visibility into how each product actually performs, and you hand Meta an impossible optimization problem. Two accounts with clean data each are easier to understand than one account with blended noise.

• "If the backpack isn't working, the creative needs fixing." For two months, the backpack's underperformance was interpreted as a creative or targeting problem. Multiple creative variations were tested. Audience segments were adjusted. Nothing moved it past marginal improvement. The actual problem was structural — the backpack was fighting for pixel attention with a product that had a faster-converting buyer profile. No creative fix can solve an architecture problem.