A budget from the goal backwards

The most common way to set a marketing budget is to take last year's and adjust it: "220,000 plus ten percent, let's move on." It is quick, and it repeats last year whatever the team is trying to achieve. If the goal doubles, the budget doesn't notice; if the goal halves, the money is spent anyway.

A goal-based budget starts from the other end. First the goal: how many new customers do we need? Then walk backwards through the path, one step at a time, using rates from your own data: how many purchases does that take, how many enquiries, how many useful contacts, and what do those contacts cost?

The money falls out of the arithmetic. And so does the answer when it's too much: you can see which step to change, whether the goal, a conversion rate or the media price, instead of cutting a line at random.

Last year's number still has a use: as one of the checks the new budget must survive, not as its starting point.

Klasna's goal for the restart: 90 new families who keep their first package. New only: families coming back from spring aren't in it. The rates come from Klasna's own data:

• about 90% keep the first package (April);
• 20% of enquiries bought one (last September);
• 0.25% of useful contacts sent an enquiry (last September).

Walk backwards, dividing by each rate:

1. First packages: 90 ÷ 0.9 = 100.
2. Enquiries: 100 ÷ 0.2 = 500.
3. Useful contacts: 500 ÷ 0.0025 = 200,000.

Mind the percentage: 0.25% is 0.0025, not 0.25. Dividing by 0.25 instead would give 2,000 contacts and a budget a hundred times too small.

Each step answers "how many did the previous step need?", so a rate that is wrong at one step shows up in every number after it. That is why the rates must be your own, recent, and about the same kind of customer as the goal.

The last step turns contacts into money with the price from the media lesson: options B, C and E together cost 0.82 UAH per useful contact (126,000 ÷ 154,000).

Media budget: 200,000 × 0.82 = 164,000 UAH.

That is below Leo's 242,000 (220,000 + 10%). It is tempting to stop here and celebrate: the goal-based budget is cheaper and more precise. Don't stop yet.

A budget built from rates is only as good as the rates. Each was measured once, on a past campaign, possibly on different customers, and each multiplies into the result. A plan that looks much cheaper than anything the team has done before is more often a sign of an optimistic rate than of a brilliant plan.

So every goal-based budget gets the same next move: sanity checks, comparing it with history, cleaning the rates, testing payback and fitting it into the month. The next cards run them on the 164,000.

The first check is simple: what does one family cost under this plan, and how does that compare with what families have cost before?

164,000 ÷ 90 families = 1,822.22 UAH per kept family.

Klasna's history: in April 2026 the paid channels averaged 2,727.27 per new family, and October 2025 cost 4,000 per family. The new plan claims families a third cheaper than the best month the team has had.

That could be true; a better media mix (B, C and E instead of whatever ran before) does help. But a plan much cheaper than anything done before needs a reason, and the honest first suspect is a rate. One of the three rates came from last September's campaign, which included families who were coming back anyway. If they inflate the enquiry rate, the plan underestimates how many contacts new families need.

The check doesn't say the plan is wrong. It says where to look before anyone signs it.

Last September's 0.25% contact-to-enquiry rate counts every enquiry, including spring families who came back anyway. About a third of them were returning families. Those families return through emails to parents who opted in, not through paid media, so they don't belong in a funnel for new families.

For new families only, the rate is two thirds of 0.25%: about 0.1667%.

Redo the last step with it. Enquiries are still 500. Useful contacts: 500 ÷ (0.0025 × 2/3) = 300,000. Budget: 300,000 × 0.82 = 246,000 UAH, or 2,733.33 per kept family.

That is about April's 2,727.27. Now the plan agrees with history, and the agreement is a reason to trust it more, not less.

The general rule: a rate measured on a mix of new and returning customers flatters any plan that is about new customers. Before using a rate, ask who was in its denominator and numerator, and remove those your goal excludes.

Hanna's question: does 2,733 per family pay back? Compare it with what a family leaves.

A first package leaves 1,710 UAH of contribution. So on the first package alone, each family brought by the campaign loses about 1,000. The plan pays back only if many families buy a second package, which leaves 2,952. In April, 47% of search families did within 90 days.

That is not a reason to cancel the campaign. Most acquisition in a business with repeat purchases works this way. It is a reason to write the dependence into the plan: "this budget pays back if at least about half of new families buy a second package". Then the later review checks exactly that number, and the team knows what failure would look like before it happens.

A budget whose payback silently depends on a later purchase is a hidden bet. A budget that names the purchase it depends on is a plan someone can manage.

The last check is the calendar. September also carries the always-on search at 90,000, and Uliana wants 18,000 to test the school chats. Total: 246,000 + 90,000 + 18,000 = 354,000 in one month, more than any month in Hanna's forecast.

Two honest fixes: start the campaign in the last ten days of August, spreading the spend over two months, or trim the goal. Either is a real decision; quietly overspending September is not.

Two habits keep the budget readable:

• Keep always-on spend separate from the campaign, so each can be judged on its own job.
• Keep tests on their own small line with a stop rule, so a test can fail without sinking the plan.

So a goal-based budget passes four checks: cost per customer against history, rates cleaned of customers who return anyway, payback against contribution, and the total against the monthly line. "At least last year plus 10%" is not a check; it is the habit the method replaces.

For Dana's one-person practice the method runs on hours. She keeps 3 marketing hours a week, 24 hours to the end of December, and has room for three more students.

Goal: fill all three places. Work it backwards from her log: asking her students and friends once (2 hours) might bring one more student; talks would have to bring two, at 5 hours per student who stayed. Hours needed: 2 × 5 + 2 = 12 hours, half her 24.

Should she fill the other 12 with more channels? No. When the goal costs less than the budget, don't spend the budget. Book talks with two new communities, stop outreach when the slots fill, and keep the spare hours for teaching. Her real limits are finding communities that will host a talk, and her three free places.

How it fits together: start from the goal, walk back through the path with your own rates, turn contacts into money, then run four checks and spend no more than the goal needs. The next lesson, the case, assembles the whole September restart plan.