Case study · Education · Online Learning · Paid Social + PPC · February 2025 – January 2026
Flat budget, 25% more leads and a 21% lower cost per lead
Mac Lav delivers live online language lessons and relies on a high volume of student leads at a low cost. Across the year to January 2026, Mira held the Paid Social budget essentially flat and concentrated activity around the audiences and creative that converted, reducing cost per lead by 21% while lead volume rose by 25%, click-through rate by 38% and reach by 78%. Paid search was introduced in November 2025 as a second lead source.
Where it hurt.
Mac Lav delivers live online lessons across several language audiences, including English, Polish, Portuguese, Romanian, Hungarian and Ukrainian. Its commercial model depends on generating a high volume of student leads at a low cost, against some of the most demanding efficiency targets Mira runs.
A fixed Paid Social budget was being spread across audiences that did not contribute equally. As reach widened through the middle of the year, cost per lead drifted upwards and reached its highest point in late spring. The challenge was to increase lead volume without adding budget or allowing efficiency to decline.
What we saw.
Spreading a fixed budget evenly across every language and audience feels fair, but performs poorly. Mira saw that a small number of audiences carried the account while the rest diluted it, so the route to lower costs and higher volume was concentration: prove which audiences and creative converted, fund those, and keep everything else as a controlled test rather than standing spend. English would carry the account; the job was to find its most efficient shape and keep cost per lead falling as reach widened.
What we did, and why.
- 01 · Diagnose the dilution
A flat budget spread across several language audiences was masking which ones carried the results, while cost per lead had risen as reach expanded.
- 02 · Concentrate on proven audiences
Prioritise English, retain Polish as the efficient secondary audience, and keep weaker languages as small, controlled tests.
- 03 · Test and optimise continuously
Run weekly creative tests against a strict cost-per-lead target while holding the overall budget flat. Paid search was then introduced as a second lead source in November 2025.
The numbers, in plain language.
Across the 12 months from February 2025 to January 2026, Paid Social cost per lead fell by 21% and moved through its target, while the budget remained essentially flat. Lead volume increased by 25%, click-through rate rose by 38%, and reach grew by 78%, with none of those gains bought through increased spend.
The improvement was recovered from a mid-year trough. Cost per lead peaked in late spring before falling steadily through the second half of the period, reaching its best sustained level in January 2026.
By March 2026, English accounted for approximately 96% of Paid Social spend, with Polish retained as the efficient secondary audience and the remaining languages held as small tests. Paid search was introduced in November 2025 as an additional lead source, but its conversion results should be added only once the current tracking has been confirmed.
| Metric | Movement | Budget context | |
|---|---|---|---|
| Cost per lead | Lead volume | Click-through rate | Reach |
| −21% | +25% | +38% | +78% |
| Essentially flat spend | Essentially flat spend | Essentially flat spend | Essentially flat spend |
For brands like yours.
Any business running one budget across several audiences, languages, regions or product lines faces the same trap: even spending feels fair but quietly holds efficiency back. The audiences are not equal, and the budget should not be either. Hold the money flat, identify the audiences and creative that convert, fund those and test the rest cheaply, and the same budget can generate more leads at a lower cost over time. Efficiency at scale is a discipline, not a bigger cheque.
“We've reached a level of growth and performance we had never been able to achieve before.”
Sean Lavender · Business Director · Mac Lav
What we ran.
Want your own?
A 30-minute call. No deck. We listen, we tell you which tier fits, and we say whether we can help.