PPC for eCommerce: A Complete Guide to Strategies That Scale

What PPC for eCommerce really means

PPC for eCommerce is paid advertising where an online store pays for each click on its ads across Google Search, Google Shopping, Performance Max, Meta, TikTok, Microsoft Ads and increasingly retail-media networks like Amazon Ads. The surfaces change every year. The job does not: turn paid attention into tracked, profitable orders.

The thing that separates eCommerce PPC from lead-gen PPC is the unit of success. A lead-gen campaign optimises towards a form submission and trusts sales to close the loop later. An eCommerce campaign has a real transaction at the end of the click, with a real order value, real cost of goods sold and a real margin. That changes everything — the tracking, the bidding, the creative, the landing page. Optimise to revenue alone and you will scale yourself into a loss; optimise to contribution margin and you will compound.

Everything below assumes you actually want to know if your spend is making money. If you only care about top-line ROAS, half of this guide is overkill. If you care about whether next quarter's P&L looks better than this one, keep reading.

Tracking: the foundation everything sits on

Nine out of ten eCommerce PPC accounts we audit are losing money in the same place: the conversion signal feeding the bidder is broken, duplicated or client-side only. Once iOS 14 and tightening browser privacy gutted pixel tracking, server-side conversion APIs stopped being a nice-to-have.

• Google Enhanced Conversions for Web with hashed customer data passed through GTM server-side, plus offline conversion uploads for refunds and cancellations so the bidder learns from true net revenue.
• Meta Conversions API deduplicated with the pixel via a stable event ID. Send Purchase with order value, currency, content IDs and category — not just an empty event.
• Consent mode v2 live and configured. In privacy-strict regions, missing consent signals quietly halve recorded conversions.
• One source of truth. Pick the platform you trust (usually Shopify or your backend) and reconcile ad-platform reported revenue against it weekly. A 10–25% gap is normal; a 50%+ gap means tracking is broken.
• Margin in the conversion value. Where the platform supports it, send contribution margin as the conversion value, not order value. Smart Bidding then optimises towards profit, not vanity revenue.

Fix this layer before you touch a single bid. Smart Bidding, Advantage+ and Performance Max are machine-learning bidders — they are exactly as good as the signal you feed them.

Account structure that scales

Bad account structure is the second most common reason an eCommerce PPC account stalls. The principles are simple: separate audiences with different intent, separate signals you want the algorithm to learn from each other, and give every campaign one job.

• Brand search, ring-fenced. A dedicated Brand campaign with exact-match brand terms. Cheap, near-100% intent, and the single best place to protect against competitors bidding on your name.
• Shopping / PMax for the catalogue. Segmented by margin tier or product category, not one giant feed dump. High-margin hero SKUs deserve their own asset group and ROAS target.
• Non-brand Search for high-intent generic terms. Exact and phrase, with negative keywords pulled from search-term reports weekly.
• Prospecting and retargeting kept apart on social. Top-of-funnel creative testing in one campaign; warm-audience retargeting and AOV-lift creatives in another.
• One conversion goal per campaign. Stop mixing "add to cart" and "purchase" goals on the same campaign and expecting the bidder to pick.

Google Shopping & Performance Max

For most eCommerce brands, Shopping and Performance Max are where the volume lives. Both run off the Merchant Center product feed, and feed quality is the lever that quietly drives 50–60% of performance.

• Titles do the keyword work. Lead with the most-searched attribute (brand, product type, key feature, size, colour). Shopping has no keywords — the title is the keyword.
• GTINs, MPNs and brand on every SKU. Missing identifiers drop you out of high-intent SERPs.
• Custom labels for ROAS tiering. Label SKUs by margin, price band, bestseller status and seasonality so you can set different ROAS targets per asset group.
• Brand exclusion lists on PMax. Without them, Performance Max eats your branded search traffic and takes credit for orders you would have got for free.
• Standard Shopping for control where it matters. A pure Shopping campaign for hero SKUs gives you per-product bid control PMax does not.
• Imagery that survives a crop. Square, white background, no text overlays — PMax recrops aggressively across surfaces.

Search: brand, generic and competitor

Search is the highest-intent paid surface in eCommerce — and the one most often run on autopilot. Three layers, each with a different job.

• Brand. Always-on, exact and phrase variants of brand and brand-plus-product searches. Land on the product or collection page, not the homepage.
• Generic high-intent. "buy", "best", "[product] online", size-, colour- or use-case-modified queries. Build out from your top-converting search terms — those are validated buyers, not guesses.
• Competitor (carefully). Only when your offer is genuinely comparable and your landing page does the comparison work. Otherwise you are paying to send shoppers to a brand they already know.
• Negative keyword hygiene. Weekly search-term review. Strip job-search terms, free-version terms, info-only modifiers ("how to", "what is") unless they convert in your data.

Meta, TikTok and paid social

Search and Shopping harvest existing demand. Paid social creates it. The two play very different roles in an eCommerce PPC mix, and the metrics you grade them on should differ too.

• Creative is the targeting. On Meta and TikTok, the algorithm finds the audience — your job is to ship enough creative variants for it to find them efficiently. Test 4–6 new concepts weekly, not one.
• Advantage+ Shopping campaigns for prospecting at scale, fed the full product catalogue and the same clean Purchase signal Meta CAPI sends.
• UGC and founder-led creative outperforms polished brand film in nearly every test we run, especially on TikTok. Authenticity reads as ad literacy, not laziness.
• Measure with blended ROAS plus incrementality. Platform-reported ROAS over-credits last-touch; blended ROAS (total revenue ÷ total ad spend) is the honest number. Run holdout tests quarterly to confirm incrementality.

Bidding, budgets and ROAS targets

The right bid strategy depends on the campaign's job, not on what is fashionable. A rough hierarchy:

• Maximise Conversion Value for new campaigns building data — no ROAS cap until you have at least 50 conversions over 30 days.
• Target ROAS once you have signal, set 10–15% below the observed steady-state ROAS to give the bidder headroom to scale.
• Manual CPC or Enhanced CPC only on tightly-controlled brand and competitor Search campaigns where you do not want machine learning redistributing your spend.
• Budgets sized to learn. A daily budget below 10x your target CPA per campaign starves the bidder of data. Consolidate before you cut.
• ROAS targets driven by margin. Required ROAS = 1 ÷ contribution-margin %. A 30%-margin product needs roughly 3.3x ROAS to break even, before considering returns and overheads.

Creative and landing pages

Even the cleanest PPC account loses to a bad landing experience. The post-click journey is part of the campaign, not a separate problem.

• Land on the product or filtered collection, not the homepage. A homepage is a navigation device; a PDP closes orders.
• Mobile-first PDPs: sub-2s LCP, sticky add-to-cart, reviews and trust signals above the fold, shipping and returns answered without a scroll-hunt.
• Message match between ad and PDP. If the ad sold the offer ("free shipping over £50"), the page must restate it within the first viewport.
• Checkout friction is PPC friction. Guest checkout, address autofill, Apple Pay and Shop Pay live on. Every recovered checkout abandon is cheaper than a new click.

We cover the page side of this in depth in our landing page best practices guide.

Scaling without losing profitability

Scaling eCommerce PPC is mostly the discipline of not breaking the things that are working. A simple cadence we use with clients:

1. 14–21 day window of steady, profitable ROAS on a campaign before any budget increase.
2. Increase budgets by 20–30%, not 100%. Big budget jumps reset the bidder's learning and tank performance for a week.
3. New audiences and creatives in parallel campaigns, not on top of proven ones. Protect the winner; iterate elsewhere.
4. Push ROAS targets down gradually to unlock volume only when margin and inventory can absorb it.
5. Watch contribution margin weekly, not daily ROAS. Daily numbers are noise; weekly numbers are signal.

Mistakes that quietly burn budget

• Client-only pixels in 2026. If you are still relying on a browser pixel alone, you are bidding on partial data.
• Optimising to revenue, not margin. Top-line ROAS hides unprofitable SKUs and discount-stacked orders.
• One mega-PMax campaign for everything. No control, no segmentation, no way to protect margin tiers.
• Ignoring the feed. Spending on bid strategy while titles, GTINs and images are broken.
• Sending paid traffic to the homepage. The single biggest conversion-rate killer in eCommerce PPC.
• Turning campaigns on and off weekly. Every restart resets the bidder's learning phase. Adjust, do not panic.
• No incrementality testing. Platform-reported ROAS always looks good; lift tests tell you what would have happened anyway.

Frequently asked questions