Victor DumeigeWhere $1,159 of Ad Spend Went
Eight months of Google Ads and GA4 for a Mercedes-Benz parts shop in Costa Rica — what the money bought, and what the two tools disagree about.
Highlights
- $1,159 of spend produced 1,781 WhatsApp conversations — about 65 cents each, on a capped budget of roughly $150 a month.
- Reported GA4 key events (4,990) overstate conversions by almost 3x. The figure that reconciles with Google Ads is the session key event rate: 1,677 sessions against Ads' 1,781.
- Exact match converted at 53% and cost $0.55 per conversation; broad match converted at 28% and cost $0.79.
- Paid search beat organic on quality as well as volume — 68.8% engagement against 57.6%, and 42.7% conversion against 31.8%.
- Method
- Campaign measurement — Google Ads and Google Analytics 4
- Year
- 2026
- Sample size
- 5,480 sessions · 4,285 ad clicks
- Data source
- Google Ads and GA4, client account (not public)
Repuestos Villalobos sells Mercedes-Benz parts in Alajuela, Costa Rica. I built the site and I run its search campaign, so this is my own work measured honestly rather than a client report written to look good. The budget is small — about $150 a month — which makes it a useful case: at this size, every wasted click shows.
From mid-January to the end of August 2026 the campaign drew 24,963 impressions, 4,285 clicks and 1,781 conversions for $1,159. A conversion is one specific action: a visitor tapping WhatsApp to message the shop. That definition sets the ceiling on what this can claim — it says nothing about how many conversations became sales.
The first thing to check is whether the two tools agree. They don't, and they shouldn't. Ads counted 4,285 clicks; GA4 counted 3,930 paid sessions — the normal 8% loss from people leaving before the page loads. On conversions the trap is bigger: GA4 reports 4,990 key events from paid traffic, which counts every WhatsApp tap including repeats. The figure that matches Ads is GA4's session key event rate — 42.7% of paid sessions, or 1,677 visits where someone opened WhatsApp at least once, against Ads' 1,781. Reporting the 4,990 as "conversions" would overstate results threefold, and that is the most common way small-business ad reporting goes wrong.
The second check is where the money went. Google names only 1,486 of the 4,285 clicks in the search terms report; the rest are aggregated for privacy, so two thirds of the spend can't be traced to a query. Within what is visible, exact-match terms converted at 53% and cost $0.55 per conversation, while broad match converted at 28% and cost $0.79 — and broad match produced 730 distinct search terms, 78% of which never earned a click. That is the concrete argument for tightening match types.
Across the same window the site took 5,480 sessions: 72% paid, 19% organic. Paid traffic also behaved better once it arrived — 68.8% engagement against organic's 57.6% — which follows from the targeting. Someone clicking an ad for "repuestos Mercedes Benz Costa Rica" has already said exactly what they want.
The read: the campaign works, cheaply, for the specific job of starting conversations. The obvious next moves are shifting budget from broad to exact match and testing whether a higher cap buys proportionally more. The honest caveat is that the site and the campaign launched together, so there is no baseline to compare against.
Campaign totals, January to August 2026
- Spend
- $1,159
- Clicks
- 4,285
- Conversations
- 1,781
- Cost per conversation
- $0.65
Figures as reported in Google Ads for the account I manage, 4 January – 30 August 2026. A conversion is one visitor opening WhatsApp from the site; it does not represent a sale.
Key findings
| Variable | Sessions that reached WhatsApp | Note |
|---|---|---|
| Paid Search | 42.7% | 3,930 sessions — 72% of all traffic, and the most engaged: 68.8% of these sessions were engaged, against a 57.6% site-wide average for organic. |
| Organic Search | 31.8% | 1,039 sessions, 19% of traffic, built up from nothing since the site launched in December 2025. |
| Direct | 10.9% | 377 sessions. The lowest conversion rate of any channel with meaningful volume — largely returning visitors and untagged links. |
| Cross-network | 58.2% | 67 sessions. Too little traffic to read anything into the rate. |
| Unassigned | 44.8% | 29 sessions GA4 could not attribute to a source. |
| AI Assistant | 70.8% | 24 sessions. The highest rate in the table on far too little traffic to mean anything yet — but a channel worth watching. |
| Referral | 42.9% | 14 sessions. Same caveat. |
Rate is GA4's session key event rate — the share of sessions in which the visitor opened WhatsApp at least once. Channels under 100 sessions are listed for completeness, not as findings.
Charts
Limitations
- A conversion is a WhatsApp button tap, not a sale. Nothing here says how many conversations became orders — that happens on the owner's phone.
- Google names only 1,486 of 4,285 clicks in the search terms report; the other 2,799 are aggregated for privacy. The match-type comparison therefore covers about a third of the spend.
- The site and the campaign launched at the same time, so there is no pre-campaign baseline. Organic growth over the period cannot be cleanly separated from the effect of the ads.
- Google reports modelled and fractional conversions at term level — one two-click term is credited with 2.95 conversions — so individual keyword figures are estimates, not counts.
- One account, one market, one product category, on a capped budget of roughly $150 a month. These rates are not a benchmark for anyone else's campaign.