Case: a quarter's budget and what really moved

Leads are the easiest number to inflate: anyone who leaves a contact counts, including students, other schools and people collecting free pens. A channel that draws crowds can top the lead chart and bring no buyers.

So judge channels by opportunities: deals that passed qualification, where a real company with a real need is talking about buying. Then credit a channel only with opportunities above the baseline, what would have come anyway.

For Klasna's autumn the control is clear: the campaigns ran only in Kyiv and Lviv, and the other cities got none, only referrals and unpromoted articles. Whatever those cities did is the trend.

Leads aren't useless: they show reach, and a sudden drop is worth a look. But a budget decision based on "leads doubled" rewards whoever brings the biggest crowd, not whoever brings buyers.

Before crediting any campaign, measure what happened where there was no campaign. That's the trend, and it applies to the campaign cities too.

In the autumn data, the other cities (the control) had no events, no outbound, no partner and no paid promotion. Their opportunities went from 5 in autumn 2025 to 6 in autumn 2026, a growth of (6 − 5) ÷ 5 = 20%. Their leads grew by the same 20% (25 → 30).

Read it as: with nothing done, the market brought Klasna about a fifth more opportunities than a year ago. Any campaign has to beat that to earn credit.

Two habits help. Sum the same rows in both years (the same months and channels), so you compare like with like. And check that the control was really untouched: if a city got a breakfast invitation, it no longer shows the trend.

With the trend known, build the baseline for the campaign cities: last year × (1 + the control's growth).

Kyiv and Lviv had 15 opportunities in autumn 2025. Growing like the control, they'd have reached 15 × 1.2 = 18 without any campaign. They actually had 31, so the campaigns get credit for 31 − 18 = 13 opportunities.

Not 31 (all of them) and not 16 (31 − 15, last year alone). Comparing all of Klasna with last year (37 − 20 = 17) would also credit the campaigns with growth that happened where nothing ran.

The same method works per channel. Referrals in Kyiv and Lviv went from 10 to 12: baseline 10 × 1.2 = 12, so zero extra, and nothing was spent. Content went from 5 to 7: baseline 6, one extra. Channels that didn't exist in 2025 (events, outbound, partner) had a baseline of zero, so all their opportunities count as extra.

To compare channels in money, give every opportunity the same yardstick: its expected contribution.

Value of an opportunity = win rate × lessons a won client takes × contribution per lesson.

Marta's planning assumption from Klasna's history: 1 in 5 opportunities is won, and a won client takes about 480 lessons in its first year. Contribution is 339 UAH a lesson (800 − 430 − 31). So one opportunity is worth 0.2 × 480 = 96 expected lessons × 339 = 32,544 UAH.

It's rough, and it should say so: label it an assumption, not a promise, and keep it visible in the memo so anyone can challenge it. If the win rate turns out to be 1 in 4, every channel's value rises by the same factor.

The point isn't precision. It's that events, outbound and the partner are judged with one ruler, so the budget follows the comparison rather than whoever argues loudest.

Events were Marta's favourite: the October conference and two HR breakfasts brought 60 leads, the most of any channel. Per lead they looked cheap: 180,000 ÷ 60 = 3,000 UAH, against 3,333 for outbound.

But they produced only 3 opportunities, a 5% conversion. Events didn't exist last year, so all 3 count as extra: 3 × 32,544 = 97,632 UAH of expected contribution against 180,000 UAH spent. After spend: −82,368 UAH. ROMI = −82,368 ÷ 180,000 = −45.76%.

Per extra opportunity, events cost 60,000 UAH; outbound cost 10,000.

A negative ROMI means the channel lost money on this quarter's evidence. It doesn't prove events can never work: a smaller, better-targeted breakfast might. But "the most leads" was the wrong reason to keep it, and doubling it would double a loss.

The channels with the fewest leads carried Klasna's autumn.

• Outbound: 18 leads, 6 opportunities, 60,000 UAH spent. 6 × 32,544 = 195,264; after spend 135,264 UAH; ROMI 225.44%.
• The referral partner (from October): 6 leads, 3 opportunities, 40,000 UAH. 3 × 32,544 = 97,632; after spend 57,632 UAH; ROMI = 57,632 ÷ 40,000 = 144.08%.

Both reach the right people: outbound because it starts from companies that fit the profile, the partner because it introduces clients who already trust it.

Still, read them with care. Six or three opportunities can swing on one deal, so treat the ranking as a direction, not a final verdict. And neither channel simply scales: outbound grows only as fast as you can find the right companies and people lawfully, and a partner sends what its own network holds.

Put the channels side by side, all with one opportunity worth 32,544 UAH:

• Outbound: 6 extra, 135,264 UAH after its 60,000 spend: the best result.
• Partner: 3 extra, 57,632 UAH after 40,000.
• Referrals: 10 → 12 against a baseline of 12: zero extra, nothing spent. That growth would have come anyway.
• Paid content promotion: 5 → 7, baseline 6: one extra opportunity (32,544) for 45,000 UAH, a ROMI of −27.68%.
• Events: 3 extra, −82,368 UAH.

Careful with content. This table counts new opportunities, and paid promotion of general articles lost money, so cut that first. But the cost sheet and the data note do another job: they help deals that already exist to close. Measure those by use in live deals, not by this table.

Leads doubled, and the quarter as a whole earned something: 13 extra opportunities × 32,544 = 423,072 UAH against 325,000 spent, about 98,000 UAH after spend, a ROMI of 30.18%. So why not double the budget?

Because a total hides its parts. Most of the lead growth wasn't marketing's (the control grew 20%), and the money that worked sits in two small channels. Doubling every line would double the events loss too. And the winners can't simply double: outbound is limited by how fast you find the right companies lawfully, the partner by its own network.

The better move is to shift money step by step: less to events and paid promotion of general articles, more to outbound and the partner, and choose next quarter's control cities before it starts. Small numbers swing on one deal, so change in steps and check whether the shifts held.

How it fits together: opportunities, not leads; a control and a baseline; one yardstick in money; ROMI by channel; careful steps. Next module: the selling conversations themselves, starting with the discovery call.