E-commerce Google Ads audit checklist for Singapore

Open your Google Ads account on a weekday morning in Singapore and you will usually see the same pattern: spend ticking up on Shopping and Performance Max, a handful of branded Search terms looking healthy, and a ROAS number that looks fine until you fold in 9% GST, free-shipping thresholds, and the return rate on apparel or beauty. That is the real audit surface for PPC for e-commerce Singapore operators. Vanity CPL and last-click ROAS hide the leaks. This checklist is built so you can score the account yourself, rank the fixes, and move money back to queries and products that still clear contribution margin.

Singapore shoppers compare fast. They bounce between your site, Shopee, Lazada, and a WhatsApp enquiry to a reseller. High-intent paid search still books baskets, but only when the query, the product feed, the landing page, and the post-click proof line up. Use the sections below as a working audit, not a theory deck.

Score the account against contribution margin, not dashboard ROAS

Start every audit with unit economics, not campaign labels. Pull the last 60 to 90 days of Google Ads spend in SGD and map it to paid orders that actually settled, after cancellations and obvious fraud. Then apply your true variable cost stack: COGS, 9% GST treatment on the way you report, payment fees, pick-and-pack, last-mile, and expected returns for that category. A 4.2x platform ROAS can still lose money on low-AOV SKUs once free shipping at $39 or $49 kicks in.

Before touching bids, create a summary tab in your audit sheet with three columns: blended platform ROAS, contribution margin ROAS, and share of spend on below-breakeven SKUs. Plenty of Singapore stores discover 20 to 30% of SEM budget is propping up discovery terms, out-of-stock variants, or gift-set queries that never clear margin after returns. That is not a Smart Bidding failure. It is an audit that never left the Ads UI.

Segment the same view by device and by new versus returning customers. Mobile still drives the majority of paid search clicks, yet conversion often lags desktop by a wide gap on fashion, gadgets, and home. If mobile ROAS looks acceptable only because branded Search props it up, unbranded Shopping is probably leaking. Mark any campaign family that cannot show contribution margin by product category as a priority fix, not a “monitor” item.

Finally, check whether your conversion actions reward the right event. If the primary goal is “add to cart” or “begin checkout” while purchase is secondary, automated bidding will chase cheap mid-funnel events. For e-commerce in Singapore, purchase (or a tightly defined qualified purchase with value) should own bidding. Everything else is diagnostic. Score yourself 0 to 2 on each of: margin-aware ROAS reporting, SKU-level losers identified, correct primary conversion. Anything under 4 out of 6 means structure work comes after economics, not before.

• Margin-aware ROAS reporting (0-2): __

• SKU-level losers identified (0-2): __

• Correct primary conversion (0-2): __

• Pass threshold: 4 out of 6 or higher

Query, negative, and brand hygiene checklist for Search and Shopping

Open Search terms across Search, Shopping, and Performance Max search themes for the last 30 days. Export, sort by cost, and tag every row: buy-intent product, research, competitor, irrelevant, job-seeker, wholesale, or wrong geography. Singapore accounts routinely pay for “wholesale”, “supplier”, “PDF”, “salary”, and neighbouring-market spellings that never convert into a local retail order. Local junk patterns also show up heavily: “KL”, “JB”, “Malaysia”, “Carousell”, “lelong”, “lobang”, brand name + “review Reddit”, and Singlish suffixes like “lah” or “sia” on informational zero-intent searches. Build negative lists in layers: account-level junk, campaign-level category guards, and ad-group level precision. Do not rely on a single shared list you last touched in Q4.

Brand terms need their own line in the audit. If non-brand campaigns still capture your brand plus “promo code” or misspellings, you are inflating non-brand ROAS and teaching Smart Bidding the wrong lesson. Split brand Search, protect it with exact and phrase structures that match how locals type, and keep brand negatives on non-brand where the UI allows clean separation. Stockouts matter here too. If top spenders in the feed are low-stock or disabled on the site, pause or suppress them in Merchant Centre before you “optimise” bids. Paying for clicks into empty PDPs is pure waste.

Dayparting and capacity are quieter on pure e-commerce than on field services, but they still matter. If your CS team only handles WhatsApp and live chat heavy volume from 10am to 7pm SGT, and a large share of high-AOV orders need human confirmation (custom sets, corporate gifting, bulky furniture), align ad schedules or bid modifiers with response reality. Overnight clicks that abandon because chat is offline show up as weak CVR, then bidding pulls back on good terms. Audit chat and email response SLAs against hourly conversion data before you blame the algorithm.

Score this block honestly. If search terms debt is older than two weeks on a five-figure monthly spend, that alone can explain a soft ROAS month. When you want a second pair of eyes on structure without rebuilding everything cold, a focused SEM / Google Ads review on query hygiene and bid signals usually pays for itself faster than adding another campaign type.

• □ Negatives reviewed in the last 14 days - pass / fail

• □ Brand separation clean - pass / fail

• □ Out-of-stock feed hygiene - pass / fail

• □ Schedule aligned to support - pass / fail

Google Ads audit: Shopping and Performance Max feed checks for Singapore stores

Performance Max often “drives the bulk now,” which is fine if the asset groups, feed, and audience signals are disciplined. Audit PMax by listing every asset group’s product filter, audience signal quality, and whether brand exclusions or channel controls are doing real work. If one asset group covers the entire catalogue, Smart Bidding is free to over-serve cheap, low-margin SKUs and remarketing paths that look efficient in-platform. Split by margin tier or category where volume allows, and give each group creative that matches the promise on the PDP.

Feed quality is a Singapore-specific choke point when titles chase marketplace keyword stuffing. Google needs clean titles, correct GTIN where required, accurate availability, and SGD pricing that matches the landing page including how you display GST-inclusive pricing. Mismatches trigger disapprovals or soft trust hits that show up as lower CVR, not neat diagnostic labels. Check image backgrounds, size variants, and whether promo price annotations still match the site after a 11.11 or 12.12 hangover. Stale sale prices train users to bounce.

Compare Shopping-only or standard Shopping legacy setups against PMax overlap. Many accounts run both and double-count assisted conversions in narrative reports while fighting themselves on the same SKUs. Decide a primary acquisition surface per product class. Use experiments or clean geo or value rules only when you can read incrementality, not when you need a comforting dashboard. AI Max is Google’s 2025 search matching update within PMax that expands query reach beyond listed search themes; it requires tight negative lists and feed titles to avoid low-intent matches on price-sensitive Singapore queries. AI Max and broader query matching can cut CPCs in some accounts, but only after negatives and feed hygiene are in place. Broad match on messy feeds just buys cheaper wrong intent.

Fail two or more and pause expansion budgets until the feed and structure pass. Scaling spend on a dirty catalogue is how five-figure months disappear into “learning.”

• □ Merchant Centre errors under control - pass / fail

• □ Price and availability parity - pass / fail

• □ Asset groups not one giant bucket - pass / fail

• □ Clear brand treatment inside PMax - pass / fail

• □ No unexplained overlap war with Search - pass / fail

Landing page and tracking checks every Singapore PPC audit should include

Ad to page continuity is still where profitable PPC for e-commerce Singapore accounts separate from busy ones. If the Shopping ad shows a colour variant at $29 and the PDP defaults to a bundled $49 kit, you paid for a confused click. Match headline promise, price, and primary SKU. Above the fold, Singapore buyers look for shipping speed to their side of the island, return rules, and payment options they trust, including cards and PayNow where you offer it. Hide those under accordion noise and mobile CVR suffers.

Proof beats adjective-heavy copy. Show recent review volume, UGC on the exact product, and delivery expectations in working days, not vague “fast shipping.” Corporate gifting and bulk SKUs need a visible path that is not a generic contact form with no SLA. For subscription or replenishment products, state the cancel rule in plain language near the CTA. Run a short session-recording pass on your top five paid landing templates. Tools such as Microsoft Clarity or Hotjar (scroll depth, rage clicks, form abandon) show where size charts, shipping calculators, and GST-inclusive price blocks stall the session. Guesswork is slower and more expensive than watching ten real sessions on your highest-spend URLs.

Tracking audit, in order: Google tag firing once on purchase, Enhanced Conversions on, server-side or CAPI-style reinforcement if you have the stack, and order value passed net of the discounts you actually want bidding to learn. Deduplicate thank-you page reloads. Reconcile Google’s conversion count to Shopify, Magento, or your OMS paid orders for two full weeks. A 15 to 25% gap is common once ad blockers, ITP, and cross-device checkout enter the picture. Smart Bidding trained on inflated or deflated values will chase the wrong baskets.

Core Web Vitals still matter on PDPs and collection URLs that absorb ad traffic. If interaction delay or load pain is concentrated on mobile product galleries, fix that before you raise bids. A common pattern on Singapore fashion and beauty PDPs is mobile gallery lag that drives tab closes during commute browsing-fix image weight and interaction delay on your top paid URLs before you scale bids. If your pages need structural help, pair the ads audit with tighter E-Commerce marketing landing work so creative and page stop arguing with each other.

• □ Ad-to-PDP price, SKU, and promise match - pass / fail

• □ Shipping, returns, and PayNow/card proof above the fold on mobile - pass / fail

• □ Purchase tag, Enhanced Conversions, and value reconciliation for 14 days - pass / fail

• □ Mobile CWV acceptable on top paid PDPs and collection URLs - pass / fail

What marketing leaders are seeing

A beauty e-commerce brand in Orchard-area fulfilment ran a six-week Shopping audit on skincare serums (SGD 28k spend). Platform ROAS sat near 5.1x, but contribution margin ROAS after GST and returns was 1.6x on gift sets and travel minis. Suppressing below-breakeven SKUs in the feed and splitting asset groups by margin tier lifted contribution margin ROAS to 3.4x on the same category over the next 30 days without raising budget.

A home and living DTC brand shipping from a Jurong warehouse reviewed Search terms after an untouched six-week stretch on a mid-five-figure monthly spend. Roughly 24% of cost sat on wholesale, JB/KL, and Carousell-adjacent queries. Layered negatives and a clean brand split improved blended ROAS by 31% over the following month, before any bid strategy change.

FAQ on PPC audits for Singapore e-commerce

How often should we run a full Google Ads audit?

Do a deep pass every quarter, with a lighter search-terms, feed, and tracking check every two weeks when monthly spend is material. Big sale periods (11.11, year-end) deserve a pre-event and post-event audit so promo prices and negatives do not rot into January.

What ROAS should a Singapore store expect from Google Ads?

Cross-industry averages sit near the low-to-mid single digits on platform ROAS, but your number only matters relative to contribution margin after GST, shipping, and returns. Set break-even ROAS by category. A gadget line and a apparel line rarely share the same hurdle.

Is Performance Max enough on its own?

PMax can carry a large share of efficient volume when the feed and asset groups are tight. Most stores still need branded Search control, disciplined negatives where available, and non-brand experiments so you can see what PMax is actually buying. “PMax only” without feed hygiene is how catalogues get strip-mined for cheap clicks.

Should we pause Google Ads when Shopee or Lazada ads look cheaper?

Compare on contribution and on customers you own, not on marketplace dashboard CPS alone. Marketplace traffic and onsite Google traffic play different roles in repeat purchase and brand search. Audit cannibalisation, but do not assume the lowest media cost is the highest margin path.

What is the fastest fix if ROAS drops suddenly?

Check tracking first, then stock and price parity in the feed, then search terms for a new junk cluster, then landing page speed on the top URLs. Bid strategy panic is the slowest way to diagnose a broken purchase tag or an out-of-stock hero SKU.

Putting it to work

Next 30 days

1. 1Expand the fix to the next one or two money paths only after the first one shows cleaner bookings.
2. 2Align creative, keywords, or content with the same offer the page now states.
3. 3Review booked outcomes weekly; cut anything that still only produces unqualified volume.

Next 30 days

1. 1Expand the fix to the next one or two money paths only after the first one shows cleaner bookings.
2. 2Align creative, keywords, or content with the same offer the page now states.
3. 3Review booked outcomes weekly; cut anything that still only produces unqualified volume.