Żak schools: 82% more leads and a 17% lower cost per lead across two enrollment seasons

Żak is a nationwide network of schools for adults, offering part-time and online education in 80 cities across Poland: general secondary schools, post-secondary colleges, medical schools, vocational and specialist courses. The business has two enrollment seasons per year, the summer enrollment (May–October) and the winter enrollment (November–March), so advertising performance has to be evaluated in enrollment cycles, not calendar quarters. The partnership started with a website project for one sub-product, then the first TikTok campaigns, then full Meta Ads management with creative production and organic social media channels.

• Scale of the offer: dozens of programs, one team. The 'one campaign = one program' model meant dozens of separate structures and dozens of creatives, and creative production could not keep up.
• CPL too high. Previous campaigns generated too high a cost per lead, making the activity unprofitable. No creative testing, no testing of messaging angles; in early 2025 the account ran dozens of campaigns that did not match the client's real business goals.
• Insufficient campaign optimization. Campaigns often relied on a single graphic used for months, launched once, with no iteration and no follow-up analysis.
• Tangled account structure and fragmented analytics. Every landing page had its own pixel; manually set targeting for each program meant the algorithms never had enough data.

The client's priority: lower the CPL and generate more leads from the same budget.

Campaign restructuring

Campaigns were grouped by product category rather than individual programs, so each gathered more lead data, exited the learning phase faster and delivered better results on the same budget.

Unified analytics

One pixel tracking the same events across all landing pages, so data could be compared on a common basis.

Targeting with creatives instead of interest targeting

Previously every program had manual targeting (the floristry course at people interested in flowers, and so on), splitting the budget across dozens of tiny audiences. Instead, the creative itself says who it is for ("floristry course", "medical receptionist") and audience selection is left to Meta's algorithms.

A new system for working with creatives

Regular rotation and testing of category creatives (e.g. "medical programs at Żak"), program-specific creatives, and UGC-format creatives, which in many cases turned out to be the most effective. Organic posts served as an early test of messaging; the same designer worked on ads and social content.

Prioritization: leads over likes

Reach- and click-oriented campaigns were dropped; the budget focused exclusively on lead generation.

A three-layer reporting system

Overall level (total leads and CPL), product level (post-secondary colleges, one-year courses, medical schools, e-secondary school, online and offline courses) and program level (e.g. floristry, nail styling), later extended with destination type and age cohorts, so budget could move mid-season.

Audience segmentation and gradual scaling

A split between new and remarketing traffic, with messaging differentiated at category level. Scaling started with the most popular programs, then new programs and categories once results became repeatable. The budget managed in the peak month of the season is now 2.5 times bigger than before the partnership started, while the cost per lead has dropped by nearly a third.

Results are analyzed in enrollment cycles, like for like: Meta Ads vs Meta Ads, season vs the same season a year earlier.

Over the analyzed period the campaigns reached 7.97 million unique users and generated over 116 million impressions and nearly 570 thousand clicks. The source notes that campaign success is not measured by engagement: it is school sign-ups that decide whether a campaign pays off.

As stated on the source:

• Budget distribution within the season. For the first two seasons budgets were set monthly on the client's side and spread relatively evenly. Only in the third season, with two full cycles of own data, could a concrete recommendation be made: reduce spend in May and June and move the savings to July and August, the months with the highest sign-up potential.
• Preparing for the season change. The transitions between summer and winter enrollment were treated too loosely; the creative and budget plan for the switchover should be prepared in advance.
• No plan B for discontinued programs. When a program with a significant budget share was withdrawn from the offer, there was no creative backup ready, and several weeks of the season were lost catching up on volume. A reserve of tested backup creatives is now maintained for every category.

The partnership continues; the third summer enrollment (May–October 2026) is under way and the source says it is shaping up to be one of the best seasons in the account's history, with an update promised once it ends.