Google Ads conversion tracking checklist for Singapore campa

Finance pings you after a Tampines lead gen sprint: “We spent S$18,400 last month. How many of those ‘conversions’ became PayNow deposits or signed quotes?” You open Google Ads. Conversions look healthy. CRM shows half the volume, and a chunk of the rest never spoke to sales. That gap is not a bidding problem yet. It is a signal problem, and Smart Bidding will happily optimise toward whatever you labelled a conversion - including thank-you page reloads, bot form spam, and WhatsApp clicks that never became a conversation.

Singapore search CPCs keep climbing while mobile still drives most paid clicks. When the conversion tag is wrong, every SGD of spend teaches the algorithm the wrong lesson. A scored audit beats another tactics list because it forces you to prove each path before you raise budgets, flip to tCPA, or dump more into Performance Max. Failing here does not look like a low CTR. It looks like booked work that never shows up next to the invoice.

Use this checklist as a self-audit for marketing leaders who own the Google Ads number in Singapore - not as a hobbyist tag install guide. Score the account, rank the leaks, then repair in order.

What “conversion” must mean before you trust the dashboard

In a Singapore account, a conversion should map to a commercial action you can defend in a board pack: qualified form submit with phone or work email, WhatsApp enquiry that sales acknowledged, booked consult, demo request, or offline win imported after the deal closed. Page views, 30-second sessions, and “button click” events are diagnostics. They are not primary conversions for bidding.

Local reality makes this messier. Many service and B2B teams still close on WhatsApp after a form or a missed call. HDB, condo, and landed-home service businesses take deposits over PayNow. If your primary conversion is only a form thank-you and you never import offline outcomes, Google learns to buy cheap form-fillers, not people who pay. If you fire a conversion on every WhatsApp icon click with no quality filter, you train the account on curiosity, not intent.

Before the scorecard, write one sentence your CMO and sales lead both accept: “A counted conversion is X, valued at roughly Y SGD when it closes, and we reconcile it weekly against CRM stage Z.” If you cannot finish that sentence, stop adding campaigns. Fix definition first. Twenty to thirty percent of SEM budgets still leak through weak intent, thin negatives, and landing mismatch - and bad conversion labels make those leaks invisible because the dashboard keeps smiling.

When the post-click path is a landing page, treat proof and form friction as part of tracking health. Session behaviour often shows abandon on long forms, buried CTAs, or proof that never appears above the fold. Tools like HeyLead Insights help you see where qualified traffic stalls after the ad click so you are not only debugging tags in GTM while the page itself leaks demand.

If you want a specialist pass on tag design and bidding readiness rather than a DIY spiral, skim HeyLead’s SEM conversion tracking setup notes for Singapore accounts, then come back and score yourself honestly.

Audit scorecard: every Google Ads conversion path you should prove in SG

Work the list in order. For each item, open the live account, Tag Assistant or GTM preview, and one real lead from the last 14 days. Pass means you can show evidence in under ten minutes. Weak means partial coverage or delayed data. Fail means you are bidding on fiction.

1. 0 Consent, GST-era receipts, and thank-you copy do not break the fire If you use a consent banner, test accept vs deny paths. Confirm marketing tags behave as designed under PDPA-minded setups. Thank-you pages should not require a second navigation that users skip. Receipt or “we will WhatsApp you” pages still need a stable success condition. Fail cost: intermittent conversions that look like “Google is unstable” when the banner or SPA route is the real culprit.

2. 1 Performance Max and brand search are not laundering bad goals Many SG accounts now say “Performance Max drives the bulk.” PMax will chase whatever you mark primary. Audit asset group landing targets and conversion goals included in the campaign. Brand search often converts easily; if brand and non-brand share one soup of goals, non-brand CPA looks better than it is. Fail cost: budget gravity toward branded and remarketing-like demand while true prospecting stays dark.

3. 2 Value rules and SGD values match unit economics If you pass values, they should reflect contribution margin reality, not vanity. Include GST thinking the way finance does so reported ROAS is not a surprise in the P&L. Your only useful benchmark is gross margin divided by media cost - a 3× ROAS means nothing if contribution margin is 20%. Fail cost: tROAS targets that force the algorithm into discount-seeker traffic.

DIY tools for this audit

Free checks on HeyLead

• UTM link builder

• Core Web Vitals checker

• SERP snippet preview

• Meta description checker

• SEO competitor analysis (landing gaps)

Pass, weak, or fail: scoring rules that keep the priority honest

Give each of the twelve items one mark: Pass, Weak, or Fail. Pass needs live proof, not a slide from the agency kickoff. Weak is “works on desktop Chrome only,” “import ran once in March,” or “Enhanced Conversions toggled on but never validated.” Fail is missing, duplicated, or contradicted by CRM.

Weight fails in this order. First, anything that invents volume (duplicate fires, Every on lead forms, WhatsApp tap as primary). Second, anything that blinds you to revenue (no offline import when sales closes offline). Third, match and consent gaps that quietly drop signal. Fourth, landing and speed issues that cap conversion rate. Only then debate bid strategies.

Do not “fix” a Fail by raising budget. Automated bidding on junk is how accounts get the reputation that “Smart Bidding inflates spend.” The algorithm did what you asked. You asked for the wrong event.

A practical bar for Singapore growth leads: you need Pass on items 1-4 and 8 before you scale. You need a plan on 5-7 within thirty days if sales cycle or messaging apps matter in your funnel. PMax expansion waits until primary goals are clean. Otherwise you are buying a smoother dashboard, not cheaper booked work.

Reconcile weekly for two weeks after changes: Google Ads conversions vs CRM new qualified leads vs won revenue influenced by Google. Expect directionally similar shapes, not identical counts. Last-click across tools will disagree. Your job is a defendable cost per qualified lead and cost per won job in SGD, not perfect equality across every UI.

Thirty-day repair sequence after the scorecard

This is repair order, not a new media strategy. Keep campaign structure stable while signal stabilises so you do not reset learning on every tidy-up.

Steps overlap intentionally - parallel workstreams are fine once tag freeze is confirmed on day 1.

30-day fix plan

1. Days 1-3: freeze primary conversion changes mid-flight only long enough to document current actions, then implement the single primary definition with sales written approval. Move vanity events to secondary or remove them from bidding.

2. Days 2-5: repair Google tag, Conversion Linker, and thank-you fires on every ad landing template. Test mobile and desktop. Kill double tags from old hard-coded snippets fighting GTM.

3. Days 4-8: turn on and validate Enhanced Conversions; fix data layer fields for email and phone used in Singapore forms (include +65 handling consistently).

4. Days 6-12: separate call and WhatsApp actions; demote raw clicks; align sales SLA so WhatsApp enquiries get a same-day tag in CRM when they are real jobs.

5. Days 10-18: stand up offline conversion import or CRM stage upload with GCLID capture on all Google landers. Backfill 30-90 days if possible so bidding sees history.

6. Days 12-20: fix landing hosts, redirect chains, and Core Web Vitals on the top five spend URLs. Align headline proof to the ad promise so conversion rate can actually move.

7. Days 18-25: re-audit PMax and Search goal inclusion; split reporting on brand vs non-brand; only then revisit tCPA or tROAS targets using the cleaned primary.

8. Days 25-30: lock a weekly reconciliation ritual (not a theatre dashboard): spend, primary conversions, CRM qualified, won revenue, notes on tag changes. No silent edits to conversion actions without a changelog.

Teams that skip to day-25 bid changes usually relearn the same junk faster. Signal first. Bids second. Creative and keyword expansion third.

Two Singapore account fixes and what changed in bidding

One home-services account reduced counted conversions by 60% after demoting thank-you-page reloads - Smart Bidding CPA dropped 28% in the following four weeks.

FAQs

How often should we audit Google Ads conversion tracking in Singapore?

Run a full scorecard quarterly, and a lightweight check after any site release, consent banner change, new landing template, or CRM field change. If you ship web updates weekly, spot-check the top spend URLs every week. Tag breakage loves “small” releases.

Should WhatsApp clicks be a primary conversion?

Rarely in raw form. Prefer a qualified definition: click with pre-filled context plus sales acceptance in CRM, or a thank-you state after a short qualifier form that routes to WhatsApp. Raw icon clicks inflate volume and train bidding toward tire-kickers, especially on mobile.

Is Enhanced Conversions enough without offline import?

It helps recover browser-side loss, but it does not replace win/loss truth when deals close on sales calls, site visits, or long B2B cycles. Use both when offline closure is common in your model.

Why do Google Ads and GA4 conversion counts disagree?

The most common culprit is attribution window mismatch. GA4 often defaults to data-driven attribution across a broader path, while Google Ads frequently leans on last-click logic with a conversion window that does not match how long your Singapore sales cycle actually takes. Consent mode, filters, deduplication rules, and different session definitions widen the gap further.

Do not force the two UIs to match exactly. Use Google Ads primary actions for bidding, GA4 for path diagnostics, and CRM for money. That split keeps optimisation honest without wasting weeks on perfect count equality.

When is it safe to switch to tCPA or tROAS again?

After primary actions are stable, volume is meaningful for your niche, and two to four weeks of reconciled CRM data show the primary event predicts qualified pipeline. If you just rebuilt tags, stay on a simpler bid strategy until the new signal accumulates. Constant conversion edits plus aggressive automated bidding is how accounts thrash.

Putting it to work

This week

• Export every conversion action in the Singapore Google Ads account and label each Primary, Secondary, or Remove with sales in the room.

• Run three test conversions on mobile (form, call click if used, WhatsApp path if used) and record whether Ads and CRM each saw one clean event.

• Verify auto-tagging, GCLID landing on the thank-you or CRM hidden field, and UTMs on a sample of 20 leads.

• Score all twelve checklist rows Pass / Weak / Fail and pick the top three Fails only.

• Check Core Web Vitals and tag presence on the five URLs burning the most spend.

Next 30 days

• Ship Enhanced Conversions validation and offline import cadence.

• Remove double-counting across call, form, and chat.

• Rebuild PMax/Search goal inclusion off the cleaned primary.

• Reconcile spend to qualified pipeline in SGD weekly and only then retune bid targets.

Pull the last 30 days of Google Ads “conversions” beside CRM qualified stages and won revenue, and circle every action that cannot earn a seat at that table. If you want that conversion plumbing owned end to end in Singapore - tags, enhanced matching, offline import, and landing handoff so Smart Bidding trains on booked work rather than junk - HeyLead can take that signal layer off your plate. Martin works with SG clients remotely - reach him on WhatsApp.