Client and goal
A construction company in Estonia: facade repairs, roofing, house repairs, stairwell and staircase renovation, joint sealing. Service area: Harju County (Tallinn and surroundings) and Narva.
- Goal: launch Google Ads for construction services and test demand by service and language, to learn which services and which language bring leads, and what to prioritise when scaling.
- Period: 17 March – 6 April (3 weeks).
- Budget: €318.
- Market specifics: a bilingual audience with a large Russian-speaking share, especially in Narva. Separate campaigns in Estonian (EE) and Russian (RU) were needed.
Starting point
Before I took over (6–16 March), the account ran a single Smart campaign for all services. It collected tens of thousands of impressions on generic searches but almost no clicks, and it was impossible to tell which service or language worked.
| Before: Smart campaign, 6–16 Mar | Test: Search, 17 Mar – 6 Apr | |
|---|---|---|
| Impressions | 29,539 | 1,117 |
| Clicks | 187 | 116 |
| CTR | 0.6% | 10.4% |
| Can you see which service and language work? | no | yes |
What I did
- A structure along three axes. Search campaigns by service (facades, roofing, house repairs, stairs, sealing) × language (EE and RU separately) × location.
- Language-specific campaigns. Each language has its own keywords and ads, and its performance is evaluated separately.
- Ad relevance. Each ad repeats what the customer searched for, service by service.
Key metrics (17 March – 6 April)
| Metric | Value |
|---|---|
| Spend | €318 |
| Impressions | 1,117 |
| Clicks | 116 |
| CTR | 10.4% |
| Average CPC | €2.74 (from €1.74 on Russian-language facades to €3.39 on roofing) |
| Leads | 4 |
| Cost per lead (CPL) | ~€80 |
Results by service
All campaigns with impressions. “Impression share” is the share of possible impressions on target searches that the ads actually received.
| Campaign | Language | Spend | Clicks | CTR | CPC | Impr. share |
|---|---|---|---|---|---|---|
| Facade work | EE | €171.66 | 58 | 10.4% | €2.96 | 80% |
| Roofing | EE | €74.57 | 22 | 7.4% | €3.39 | 75% |
| Facade work | RU | €29.65 | 17 | 16.8% | €1.74 | 84% |
| House repairs | EE | €22.38 | 8 | 7.7% | €2.80 | 77% |
| Joint sealing | EE | €12.26 | 6 | 25.0% | €2.04 | 96% |
| Joint sealing | RU | €3.80 | 2 | 15.4% | €1.90 | 100% |
| House repairs | RU | €2.43 | 1 | 12.5% | €2.43 | 89% |
| Staircase repairs | EE | €1.33 | 2 | 25.0% | €0.67 | 67% |
Analysis
Volume leaders. Facades and roofing drove the traffic: 84% of clicks and 87% of the budget.
Cost of acquisition. Four leads in three weeks at about €80 each is cost-effective for a local construction niche. 3.4% of the 116 clicks became leads, and that was at launch, before any landing page or bid optimisation.
Impression share. Even during the test, the ads received 75–100% of possible impressions, and 58–100% of top-of-page impressions (71% on the main Estonian facade campaign). Demand in the target area was almost fully covered, so growth has to come from new services and regions, not from a bigger budget on the same searches.
The language effect
Russian-language campaigns had almost twice the CTR and 40% cheaper clicks.
| Language | Spend | Impressions | Clicks | CTR | CPC |
|---|---|---|---|---|---|
| Estonian (EE) | €282 | 995 | 96 | 9.6% | €2.94 |
| Russian (RU) | €36 | 122 | 20 | 16.4% | €1.79 |
Russian-language campaigns reached a 12.5–16.8% CTR, well above the test average of 10.4%, because there is less competition for Russian-speaking customers. Estonian campaigns ranged from 7.4% to 25%, with the highest figures on niche services with few impressions.
Challenges and solutions
Problem: a bilingual market. One campaign for both languages doesn't work: audiences mix, ads in the wrong language lose clicks, and the difference in cost and return becomes invisible.
Solution: fully separate RU and EE campaigns with their own keywords, ads and performance analysis. That is how we found out that Russian-language clicks were 40% cheaper.
Problem: low volume in a local niche. Construction services in a small area generate few searches and few leads: in three weeks, all search campaigns together got just over a thousand impressions at a 75–100% impression share.
Solution: focus the test on learning about demand across services and languages instead of chasing volume. The result is not just leads but a demand map: which services people search for, in which language and at what cost per click.
Takeaways
- In a bilingual market, splitting campaigns by language is essential. The Russian-speaking segment was less competitive and more responsive (16.4% CTR vs 9.6%), but it gets lost in a combined campaign.
- A service × language structure shows what works. Facades and roofing bring 84% of clicks; sealing and stairs are niche services with a high CTR and low volume.
- Search beats Smart campaigns here. In construction, people search for a specific service.
- The market sets the ceiling. At a 75–100% impression share, growth comes from new services and regions, not higher bids.