Case study on promoting a driving school in Kyiv
How we set up Google Ads advertising, analytics, and switched to smart bidding
Audit
We received this ad account in an already working state. The client was dissatisfied with the number of monthly leads and asked us to audit the campaigns to get more conversions from the same budget.
During the audit, we noticed that most of the budget was being spent on search campaigns. This is generally fine: search generates “hot” leads since people are already looking for a specific service. However, in some niches search does not always work effectively — either the website is not very conversion-friendly, or the competition in a given GEO is too high.
Hypothesis testing
We decided to test display advertising. Initially, the idea was to combine search and banners in one campaign using Performance Max. However, the experiment failed: the campaign still leaned heavily toward search, and display placements didn’t get enough traction. As a result, the cost per lead slightly decreased, but the overall number of conversions did not grow.
After that, we tested pure display campaigns, but not on external websites — instead on Google’s own inventory: YouTube, Discovery, Gmail. Here, we were confident there would be no spam and no low-quality traffic.
The client offers several training programs (Category B, Category A, Category A1, special courses for women, and separate packages for KPI university students). At first, we launched a single campaign with ad groups for different directions. By the second month, the results were already better than Pmax. Search was kept only for branded queries, while all other placements were moved to Demand Gen.
We created one campaign with the goal “Maximize Conversions” and added ad groups with different banners per program. Then we faced a challenge: for example, Category A is less profitable than Category B, and we wanted to increase the budget specifically for B. Within one campaign, this is inconvenient to manage, especially with a small budget. Switching to “Target CPA” was not possible due to budget limitations. That’s why we decided to split one campaign into four according to the service categories.
Now we can:
1. Manage budgets separately for each program,
2. Analyze category effectiveness,
3. Quickly reallocate budgets depending on demand (sometimes Category B performs better, sometimes Category A generates cheaper leads).
As a result, these campaigns performed much better than our previous experiments: the number of leads increased by about 3 times.
Analytics
Ad Tests
Demand Gen
Smart
Bidding
After that, we worked more on creatives: testing ad copies, images, and combinations. We experimented with both real photos and AI-generated visuals.
Analytics was another important block. When we took over the project, conversions were being tracked through Google Analytics. Since Google now recommends switching to Google Tag, we set up conversion tracking directly in Google Ads. We now track:
- phone number clicks,
- messenger clicks,
- email copy actions with right button,
- website form submissions.
We also maintain a file with all leads, which sales managers update daily with lead quality feedback. The file includes the source and type of conversion (call or form submission). So far, we haven’t identified a clear winner — both calls and forms perform similarly. Therefore, we optimize campaigns for all types of conversions together.
Conclusion
Search campaigns are not always the most effective. In some niches, competition is too high, and CPCs make the cost per lead unprofitable. In such cases, testing other Google Ads formats can deliver better results. Google is not only about search.
