The brief: scale e-commerce revenue for a helmet brand with an existing catalogue
Vega Auto Accessories is a well-known Indian brand in the helmet and auto accessories category, selling through its own e-commerce store at vegaauto.com. When I took over the paid media, the brand already had strong product lines, decent organic traffic, and a functioning Shopify store. What it did not have was a paid acquisition system designed to scale.
The brief was straightforward: drive significantly more revenue through the website using Meta and Google Ads, without letting cost of acquisition eat into margins. Vega's catalogue runs deep: full-face helmets, half-face helmets, open-face models, riding gloves, riding jackets, helmet accessories, and several SKUs within each line. That depth is both an advantage (more products to advertise, more price points to match to audiences) and a problem (more complexity in the catalogue feed, more creative needed, more campaign structure to manage).
The target was aggressive quarterly revenue growth while holding ROAS above 5x. Anything below that and the margins on mid-range helmets (which make up the bulk of volume) stop making sense after factoring in fulfilment and returns. That 5x floor shaped the entire account.
- Brand: Vega Auto Accessories (vegaauto.com), helmets and riding gear
- Goal: aggressive e-commerce revenue scaling via Meta and Google Ads
- Constraint: hold blended ROAS above 5x to protect margins on mid-range helmet SKUs
- Starting point: strong brand, deep product catalogue, but no structured paid acquisition system
How I structured the ad account for scaling
The first thing I did was separate the account into three layers: prospecting, retargeting, and catalogue (dynamic product ads). Most e-commerce accounts I audit have one campaign doing all three jobs at once, which means Meta's algorithm optimises for the cheapest conversion (usually a retargeting purchase from someone who was going to buy anyway) and prospecting quietly dies. You cannot scale on retargeting alone because the retargeting pool is a function of how many new people you bring in at the top.
On Meta, prospecting campaigns ran broad and interest-based audiences with static creative (images, short-form video, carousel). Retargeting campaigns hit website visitors, add-to-cart users, and video viewers in 7-day and 14-day windows. Catalogue campaigns used Vega's product feed to serve dynamic product ads (DPAs) showing the exact helmet or accessory a visitor had browsed. Each layer had its own budget, its own cost-per-purchase target, and its own creative rotation schedule.
On Google, I ran a split between Performance Max campaigns (which pull from the product feed and serve across Search, Shopping, Display, YouTube, and Discover), branded search campaigns to capture bottom-funnel intent, and standard Shopping for specific high-margin SKUs where I wanted manual bidding control. Performance Max handled the scale; branded search protected the brand terms from competitors bidding on "Vega helmet"; standard Shopping gave me a lever to push individual products when stock or margin justified it.
This three-layer split on both platforms is what let me increase budget without cannibalising returns. When I added ₹1L to prospecting, it fed more people into the retargeting and DPA pools, which then converted at a higher ROAS. The system compounds instead of collapsing.
- Meta: three campaign layers (prospecting, retargeting, catalogue/DPA), each with its own budget and ROAS target
- Google: Performance Max for scale, branded search for defence, standard Shopping for high-margin SKU control
- Retargeting windows: 7-day add-to-cart, 14-day site visitors, video viewers
- The structure creates a compounding loop: more prospecting spend feeds the retargeting and DPA pools
Creative testing: what worked for helmets and riding gear
Creative is where most e-commerce scaling efforts fail. You can have the perfect account structure, but if the ads are stale, CPMs climb and click-through rates fall. For Vega Auto, I ran a structured creative testing framework: every two weeks, I launched a batch of 3–5 new ad variants into a dedicated testing campaign with a fixed budget. Winners (defined as ads that hit the target CPA within ₹3,000 of spend) graduated into the main prospecting campaigns. Losers were killed.
The formats that consistently won for Vega were product-focused short videos (10–15 seconds) showing the helmet from multiple angles, with the price overlaid on the thumbnail. Carousel ads also performed well, especially when each card showed a different SKU within the same category (for example, four full-face helmets at four price points). Static single-image ads worked for retargeting but rarely won in prospecting against video.
One finding that surprised me: lifestyle imagery (rider on a bike, scenic road) underperformed product-on-white by a wide margin in Meta prospecting. The hypothesis is that Vega's buyers are searching for a specific helmet, not aspirational content. They want to see the product, the price, and the trust signals (ISI certification, brand name). So I leaned into that: clean product shots, price on the creative, and the Vega logo visible. It is not glamorous creative, but it converted.
On Google, creative testing meant feed optimisation more than ad design. I tested product titles (leading with "Vega" vs. leading with the helmet type), product images (lifestyle vs. product-on-white), and custom labels to segment the feed by margin tier. The title change alone moved click-through rate on Shopping ads by roughly 15–20%, which compounds into significantly more traffic at the same spend.
- Bi-weekly creative testing batches: 3–5 variants per cycle, winners graduate at ₹3,000 spend threshold
- Top Meta formats: 10–15 second product videos, multi-SKU carousels with prices
- Product-on-white outperformed lifestyle imagery in prospecting by a wide margin
- Google feed optimisation: product title structure and image testing moved Shopping CTR by 15–20%
- Creative refresh every 2–3 weeks prevented CPM creep on Meta
Catalogue ads and dynamic product ads: the quiet revenue engine
Catalogue ads (also called dynamic product ads, or DPAs) were the single most efficient revenue channel in the Vega account. These are ads that pull directly from the product feed and show each user the specific product they browsed, added to cart, or viewed a similar product to. For an e-commerce store with 200+ SKUs, this is the format that scales without needing new creative for every product.
Setting up catalogue ads properly for Vega meant getting the product feed right first. I worked on the Shopify feed to ensure every product had clean titles, accurate pricing, high-resolution images, correct availability status, and structured product types. A messy feed means messy ads: wrong prices, out-of-stock products showing up, or images that look broken on mobile. I also set up product sets within the catalogue so I could run separate DPA campaigns for helmets, accessories, and riding gear, each with its own budget.
The retargeting DPAs (showing people the exact product they viewed) ran at a ROAS consistently above 8x. That number looks incredible until you remember these are warm audiences who already showed purchase intent. The real work is in the broad DPAs: catalogue ads served to prospecting audiences who have never visited the site. Meta uses signals from the pixel and the catalogue to find new buyers likely to purchase from the feed. Broad DPAs for Vega ran at roughly 4–5x ROAS, which is lower than retargeting but still profitable and, critically, scalable because you are not limited by the size of your retargeting pool.
I allocated roughly 30–35% of Meta budget to catalogue campaigns by the end of the quarter. That share grew over time as the pixel matured and Meta's algorithm got better at matching products to prospecting audiences.
- Retargeting DPAs: consistently above 8x ROAS on warm audiences
- Broad (prospecting) DPAs: 4–5x ROAS, scalable because not capped by retargeting pool size
- Product feed hygiene: clean titles, accurate pricing, high-res images, correct availability
- Product sets split by category (helmets, accessories, riding gear) for separate budget control
- Catalogue campaigns grew to 30–35% of total Meta budget as pixel data matured
Budget pacing: how I scaled from conservative spend to ₹25L+ monthly
Scaling ad spend without destroying ROAS is the hardest part of e-commerce performance marketing. The playbook I used for Vega was incremental: I never increased any single campaign's daily budget by more than 20% in one move. Jumps larger than that reset Meta's learning phase and cause CPM and CPA spikes that can take days to recover from. Patience is the strategy.
The pacing worked in weekly cycles. Each Monday I reviewed the prior week's performance by campaign layer (prospecting, retargeting, DPA) and by platform (Meta, Google). If a campaign hit its ROAS target for the week, it got a 15–20% budget increase. If it missed, I held the budget and diagnosed: was it creative fatigue, audience saturation, or a platform-wide CPM spike? Only after identifying the cause did I act, usually by refreshing creative or tightening the audience rather than cutting budget.
This discipline is what took the account from an initial conservative test budget to ₹25L+ in total monthly spend across clients, with Vega being one of the largest individual accounts. The quarterly revenue of ₹80L+ came from compounding these small weekly increases over 12–13 weeks. There was no single moment where I "turned on the budget." It was a steady ramp, checked weekly, with creative refreshed every cycle to keep the engine from stalling.
On Google, budget scaling followed a similar logic but with a different lever. Performance Max campaigns respond well to target ROAS bidding: I started with a conservative tROAS, let the campaign stabilise for 2–3 weeks, then gradually lowered the target (allowing the algorithm to spend more aggressively) while monitoring actual ROAS. If actual ROAS held above 5x as I lowered the target, I continued. The moment it dipped, I pulled the target back up and waited.
- Never more than 20% budget increase per campaign per move to avoid learning-phase resets
- Weekly review cycles: Monday performance audit by campaign layer and platform
- Scaling from test budget to ₹25L+ monthly spend took 12–13 weeks of compounding increases
- Google Performance Max: gradual tROAS lowering to unlock more spend while monitoring actual returns
- Creative refresh every 2–3 weeks was mandatory to sustain ROAS through each budget increase
The results: ₹80L+ revenue at 5.58x blended ROAS
Over the quarter, the Vega Auto Accessories account generated ₹80L+ in tracked e-commerce revenue at a blended ROAS of 5.58x across Meta and Google Ads combined. That means for every rupee spent on ads, the store saw ₹5.58 in revenue. On a mid-range helmet catalogue where average order values sit in the ₹1,500–₹4,000 range, that return clears the margin threshold comfortably.
Breaking it down by platform: Meta contributed the majority of revenue, driven by the combination of prospecting video ads, retargeting, and catalogue DPAs. Google's contribution came primarily through Performance Max and branded search, with standard Shopping adding incremental volume on high-margin SKUs. The blended number matters more than individual platform ROAS because the customer journey crosses platforms: someone sees a Meta video ad, searches "Vega helmet" on Google the next day, and buys through a Shopping click. Attribution on either platform alone underreports.
The trajectory across the quarter was not flat. Month one was largely setup, creative testing, and learning-phase investment: ROAS was closer to 4x as the pixel collected data and I identified winning creative. Month two saw ROAS climb above 5x as winners scaled and catalogue campaigns kicked in. Month three, with mature pixel data, a library of proven creative, and warmed-up audiences, ROAS stabilised above 5.5x at the highest budget level of the quarter. That curve is what a well-paced scaling effort looks like.
- ₹80L+ e-commerce revenue in a single quarter
- 5.58x blended ROAS across Meta and Google Ads
- Average order value in the ₹1,500–₹4,000 range (helmet-heavy catalogue)
- Month 1: ~4x ROAS during setup and learning; Month 2: above 5x as winners scaled; Month 3: 5.5x+ at peak spend
- Cross-platform attribution: Meta prospecting fed Google branded search conversions
What tracking and attribution looked like
Accurate tracking is the foundation of everything above. For Vega Auto, I set up Meta Pixel with the Conversions API (CAPI) running server-side through Shopify's native integration. This dual setup (browser pixel plus server-side events) meant purchase events were deduplicated and Meta's delivery algorithm optimised on real transactions, not inflated or missing data. Without CAPI, I have seen Meta underreport purchases by 20–30% post-iOS 14.5, which makes scaling decisions based on in-platform ROAS unreliable.
On Google, enhanced conversions were configured to pass hashed first-party data (email and phone from the checkout) back to Google Ads, improving conversion matching and giving Performance Max better signals to optimise against. Google Analytics 4 ran alongside for a platform-independent view of revenue, though I used it for directional validation rather than as the primary reporting source. The primary source was always in-platform data (Meta Ads Manager, Google Ads) cross-referenced with Shopify's order dashboard.
I also set up UTM tagging across every ad, ad set, and campaign so that Shopify's analytics could attribute orders to specific campaigns and creatives. This was how I validated which creative tests actually drove revenue, not just clicks. A video ad might have a great click-through rate but terrible conversion rate on the site; the UTM data catches that gap, and I used it to kill underperforming creative even when the in-platform metrics looked healthy.
- Meta Pixel + Conversions API (server-side via Shopify) for deduplicated purchase tracking
- Google enhanced conversions with hashed first-party checkout data
- GA4 for directional cross-platform validation; in-platform data as primary reporting source
- Full UTM tagging across every ad for Shopify-level creative attribution
- Without CAPI, Meta purchase tracking can underreport by 20–30% post-iOS 14.5
Lessons for other e-commerce brands scaling paid ads
The first lesson is structural: separate your campaigns by function (prospecting, retargeting, catalogue), not by product or audience whim. If you let Meta optimise one campaign that does everything, it will find the cheapest conversion and ignore the rest. Structure gives you visibility and control, which is what you need when you are spending ₹25L+ a month.
The second lesson is about creative. It is not optional, it is the fuel. Every two to three weeks, you need fresh creative in your prospecting campaigns or CPMs will climb and ROAS will fall. For Vega, the winning formula was simple: product-focused, price-visible, brand-visible. Do not over-invest in aspirational lifestyle content if your buyer is searching for a specific product at a specific price. Test fast, kill fast, scale winners.
The third lesson is about patience in scaling. The instinct when something works is to double the budget overnight. Resist it. A 20% increase per move, reviewed weekly, with creative refreshed alongside the budget increase, is what gets you from ₹5L to ₹25L+ without ROAS collapse. The quarter is 13 weeks. Compound 15–20% weekly increases over 13 weeks and you will be surprised how large the budget gets without any single dramatic jump.
Finally, tracking is not a setup-and-forget task. If your Conversions API goes down, if your product feed breaks, if your UTM structure gets inconsistent, you lose the data that every scaling decision depends on. I checked the feed, the pixel health, and the CAPI match rate weekly. Boring work, but it is the difference between making decisions on real data and making decisions on guesses.