Lead-Gen Google Ads Audit: An Australian Coaching Business
Audit findings, plus an early read. The audit numbers below are Google Ads figures for the 30 days to early September 2026, where a lead is a Google Ads conversion. Our first changes went live on 10 October 2026. We’ll replace the early read with full results once there are 90 days of data under our management.
Early read: the first two weeks
These figures come from the client’s CRM attribution, not from Google Ads, so a lead is a new contact in the CRM and a customer is a closed sale. Google Ads only.
| Period | Ad spend | New leads | Cost per new lead | New customer acquisition cost |
|---|---|---|---|---|
| 8 weeks before (27 Jul to 20 Sep) | $18,722 | 181 | $103 | $1,560 |
| 3 weeks to 11 Oct (first changes on 10 Oct) | $6,044 | 23 | $263 | $3,022 |
| First 2 full weeks after (12 to 25 Oct) | $4,078 | 48 | $85 | Too early |
Cost per new lead in the first two full weeks was 18% below the eight-week baseline, and about a third of the three weeks just before the takeover, when results had dropped. No sales have closed yet from those two weeks’ leads, which is normal for a considered purchase, so new customer acquisition cost can’t be measured yet. Two weeks is too short to call this a result. Treat it as a direction, not proof.
Short answer: an Australian online coaching business was paying $31.70 per lead on its best campaign, and about $108 per lead on two others. The account’s foundation was sound: real booked sessions fed Smart Bidding. The problems were where the budget went: two funnel campaigns too small to learn, one keyword carrying more than half the spend, a brand campaign buying generic searches, and ad copy that attracted the wrong clicks.
The business and the starting point
The client sells a paid coaching program online across Australia. A lead is a booked session, and the account already sent two genuine conversion actions to Google Ads: the booking and a qualified lead from its CRM. That matters more than anything else in a lead-gen account, because Smart Bidding optimises toward whatever you tell it counts. Here it was optimising toward real prospects, not page views or button clicks.
The account at a glance
| Campaign type | Cost per lead | Leads (30 days) |
|---|---|---|
| Core search, Target CPA | $31.70 | 69.5 |
| High-intent search | $41.86 | 35 |
| Top-of-funnel search | $104.99 | 7 |
| Mid-funnel search | $110.58 | 6.5 |
$1,454 of the 30 days’ spend went into the two funnel campaigns. They produced 13.5 leads at a blended cost of about $108 each. The two proven campaigns produced 104.5 leads at $32 to $42.
Finding 1: the funnel campaigns couldn’t learn
Each funnel campaign had fewer than 10 conversions in 30 days. That is too little data for Smart Bidding to build a working model, so both were stuck in learning, and their cost per lead wasn’t going to come down.
The usual reason for funnel campaigns is that people need several touches over several days before they convert. The account’s own attribution data said otherwise: 79.4% of conversions happened within a day of the first click, after an average of 2.3 interactions. If the funnel campaigns were starting those journeys, attribution would have credited them. It didn’t.
Fix: move the funnel budget into the two proven campaigns, where the extra volume strengthens the bidding model instead of starving it.
Finding 2: one keyword was carrying the account
- 56% of spend went through a single broad match keyword.
- 9 of 12 ad groups were paused, and 165 of 182 keywords were paused or getting no traffic.
- The core campaign spent about $70 of its $75 daily budget, yet Google flagged it as limited by search volume, not budget.
So adding budget alone wouldn’t have grown it. The keywords had run out of searches to match. People researching this kind of help search in questions, phrased dozens of different ways, and an exact match list can’t cover them all. Because the account measured real conversions, wider matching was safe: Smart Bidding filters the quality.
One phrase match keyword had cost $180 for a single lead. Its search terms showed why: it matched “[service] + suburb” searches from people wanting an office they could visit nearby, which this online business doesn’t offer. The generic versions of the same searches did convert, so the fix was a narrow negative on the suburb pattern, not blocking the whole topic.
Fix: move research themes to phrase match, and add the budget from Finding 1 at the same time, so the extra reach has money behind it.
Finding 3: the brand campaign was buying generic searches
The brand campaign used exact match with Maximise Clicks, the standard brand setup for years. But exact match now includes close variants, and the search terms report showed generic, non-brand questions triggering brand ads. Across the account, 73% of visible spend came through close variants. Maximise Clicks bought those clicks at whatever they cost, with none of the qualification the brand campaign assumed.
Fix: negative keywords for the generic variants, brand list controls, and a different bid strategy.
Finding 4: GA4 recorded zero outcomes
Google Ads knew when a lead happened. GA4 didn’t: 745 active users and 5,500 events in a week, and zero key events. Without them there was no way to compare landing pages or channels, so every page-level decision was a guess.
Fix: mirror the existing conversions as GA4 key events. A small job with a large payoff, and the first thing to do.
Finding 5: high click-through rates hid low-quality clicks
The ads used keyword insertion, which repeats the searcher’s own words back to them. Every ad looks relevant, so click-through rate looks excellent. But copy that agrees with everyone stops filtering out people who won’t buy.
| Ad group (same week) | Click-through rate | Conversion rate |
|---|---|---|
| Ad group A | 9.73% | 2.50% |
| Ad group B | 4.30% | 25.0% |
The ad group with the best click-through rate converted worst.
Fix: replace keyword insertion with written copy that qualifies the click, trading some click-through rate for cheaper, better leads.
The plan, in priority order
- Set up GA4 key events so landing pages and channels can be measured.
- Consolidate the funnel budget and widen matching together in the proven campaigns.
- Move research themes to phrase match to reach question-style searches.
- Negatives, brand controls and safety exclusions, including crisis-support searches the business shouldn’t advertise on.
- Rewrite the ads to qualify clicks instead of mirroring every search.
- Test AI Max with its own budget and guardrails, once the fixes above are live.
What this means for other lead-gen businesses
- Check what counts as a conversion first. If Google Ads is counting page views or button clicks, nothing else in the account can be trusted.
- Small campaigns rarely earn their keep. Under about 10 conversions a month, Smart Bidding can’t learn. Fewer, better-fed campaigns usually beat a full funnel structure.
- “Limited by search volume” means widen, not spend more.
- Read the brand campaign’s search terms. Close variants mean exact match isn’t exact any more.
- Judge ads on cost per lead, not click-through rate.
Running Google Ads for lead-gen businesses is part of Google Ads specialist for small businesses at Adgenix. If you want the same audit on your account, the free Revenue Leak Report covers cost per lead, search terms and lead quality for lead-gen accounts.