How a Data-Backed Creative Audit Took Knya's ROAS from 3.42 to 5.64 in One Month
6.49
Best ROAS — rebranding statics
5.24
Best ROAS — statics with model (grey BG)
+65%
ROAS improvement January to February
₹247
Lowest CPA — rebranding statics
Context
Knya is a D2C medical apparel brand — scrubs, sets, and workwear built for healthcare professionals who spend long shifts on their feet. The brand had an established identity and was running Meta ads across multiple creative formats: static collages, flat lays, model-led statics, and video. The problem wasn’t spend — it was signal. No systematic analysis had been done across formats to understand what was actually driving performance and what was dragging it down.
Adbuffs ran a structured creative audit across static, flat lay, video, and rebranding ad formats to identify the highest-performing combinations — and cut spend on what wasn’t working.
Objectives
Identify which creative format, background, layout, and voiceover combinations produced the best ROAS and lowest CPA — and use those findings to restructure the creative mix going forward. The secondary goal was to test rebranded creatives gradually to understand whether a brand refresh could lift performance without disrupting the account.
What we did
- Static Ads — Model vs. No Model vs. Collage
Adbuffs tested four static ad variants head to head. Model-led statics on a grey background came out on top — 5.24 ROAS, 0.69% CTR, ₹329.13 CPA — making them the best-performing static format in the account. The same statics on a blue background delivered 4.21 ROAS. Collage ads landed in the middle at 3.35 ROAS on the highest spend (₹7.7L), indicating budget had been going to a mid-tier format at scale. Product-only statics on a pink background were the weakest, returning just 3.12 ROAS at the highest CPA of ₹426.40.
The insight was clear: the presence of a model, not the product shot, was driving conversion. Buyers responded to seeing scrubs worn — not folded or laid flat.
- Flat Lay Ads — Style × Format × Background
Three variables were tested across flat lay ads: product arrangement style, ad format, and background colour. Stacking products on top of each other (rather than distributing them) produced a 3.93 ROAS at 4.50% CR versus 3.61 ROAS for distributed layouts. Split-screen format (3.97 ROAS, 4.75% CR) outperformed fullscreen (3.51 ROAS, 3.50% CR) — the added context and messaging panel alongside the product drove measurably better conversion. On background colour, dark green (teal) returned 3.59 ROAS while light blue collapsed to 1.22 ROAS at ₹1,003 CPA — a cost difference that made continuing with light blue indefensible.
- Video Ads — Type × Voiceover
Model-wearing videos outperformed all other video types: 3.29 ROAS, 19.31% hook rate, 24.15% hold rate. Product-focused videos returned 2.62 ROAS with an 11.41% hook rate. Collage/split videos came in lowest at 2.02 ROAS. On voiceover, AI voice delivered the best numbers — 3.74 ROAS, 0.67% CTR, ₹362 CPA, 4.17% CR — ahead of natural voice (3.29 ROAS) and music-only (3.10 ROAS). The winning video formula: model wearing the product, AI voiceover.
- Rebranding Statics — Gradual Rollout
Rebranded creatives produced the highest ROAS in the account at 6.49 with a ₹247 CPA — but on only ₹21.7K spend. With limited data at that spend level, Adbuffs recommended a gradual rollout: swap logos on existing winning creatives first, then shift background colours, then update design elements. Scaling rebranded creatives overnight risked disrupting the account’s established learning before there was enough signal to trust the efficiency.
Results:Account ROAS improved 65% month on month — from 3.42 in January (CPA ₹464, CTR 0.37%) to 5.64 in February (CPA ₹276, CTR 0.34%). CPA dropped by ₹188 in a single month. The creative restructuring — shifting budget toward model-led statics on grey backgrounds, split-screen flat lays on dark green, and model + AI voice video — drove the improvement without increasing total spend.
Key Learnings
Model presence drives conversion in apparel — product-only creatives don’t close the gap
Across every format tested, creatives featuring a model wearing the product outperformed product-only alternatives. For a category built on fit, movement, and identity, this is structural — not a creative preference.
Background colour is not a cosmetic decision — it’s a performance variable
Light blue returned 1.22 ROAS versus 3.59 for dark green at nearly identical spend. That’s not a marginal difference. Background choices need to be tested and retired with the same rigour as any other creative element.
Split-screen adds context that fullscreen can’t — and the data confirms it
The messaging panel in split-screen flat lays drove 4.75% CR versus 3.50% for fullscreen. Adding copy alongside the product reduced friction and gave buyers a reason to act.
Rebranding at scale before the data is ready destroys efficiency
The highest ROAS in the account came from rebranded statics — but only at ₹21.7K spend. Scaling that prematurely would have reset learning phases and burned budget before the signal was reliable. Gradual rollout is the only safe path.
Creatives used for our Clients
