01Why unit economics matters at all
A "unit" is one client. Unit economics answers one question: does EVERY SINGLE client bring the company profit, not the business as a whole. That's a fundamentally different view than "how much did we earn this month" — you can grow in revenue and go broke at the same time if every new client costs more than they bring in.
That's why a marketer thinks not "how much did we spend on ads" but "how much did one client cost, and how much money will they bring in." Everything else in this document is just a careful way to compute those two numbers and compare them.
02The funnel: why 4 steps, not one percentage
The junior approach is to take one number — "conversion from ad to sale." The problem: it mixes 4 different processes, each with its own cause and its own fix:
- Ad click — depends on the creative and targeting.
- Click → Direct — depends on how clear the offer is.
- Direct → booking — depends on HOW and WHEN the person responds (the most powerful lever, see §9).
- Booking → showed up and paid — depends on operations: reminders, ease of rescheduling.
Treat it as one number and it's impossible to know WHAT to fix when conversion drops. Splitting it out is a diagnostic tool, not complexity for its own sake.
03CPA — not just "spent ÷ bought"
CPA (Cost Per Acquisition) is the result of multiplying all 4 conversions along the chain. Budget 20,000 ₴, CPC 30 ₴ → clicks ≈ 667. Of those, some portion messages Direct (8%), then some portion of those books (15% or 42% — by scenario), and then some portion shows up and pays (85%). CPA is the budget divided by the final number of people who actually paid.
CPA isn't "the truth about the world," it's a function of assumptions: multiplying percentages produces a non-linear effect, so a small change at the input (response speed) produces a large change at the output.
04LTV — why "lifetime" doesn't mean "their whole life"
LTV (Lifetime Value) sounds like "how much the client will bring in over their whole life," but it's really a statistical expectation based on two things: how often the client comes back, and how long they stay active before they churn.
A client definitely brings in money on their first visit. After that, it's not a given they'll come back at all: only 35-50% make it to a second visit. Only that portion starts "accumulating" further value — through visit frequency (~1.1/month) and how long they stay active (~8 months — an estimate).
contribution from the first visit + (probability of becoming active × months active × visits/month × contribution per visit). This is a weighted average that accounts for the fact that not all clients are the same.
05Contribution margin vs. revenue vs. profit
This is where almost every beginner gets confused — let's go slowly:
| Revenue | the client's entire check, e.g. 540 ₴ |
| Contribution margin | left after direct costs (technician's commission, supplies), BEFORE fixed costs — ~52% of the check in the model, ~281 ₴ |
| Net profit | left after ALL of the business's costs (rent, salaries, taxes) — that's no longer the job of unit economics |
For marketing decisions, the right number to use is contribution margin — the only number that actually "grows" specifically because of this client. Calculating by revenue is claiming every hryvnia of the check is profit, which isn't true.
06LTV:CAC — the 3x rule, and why it's not a law of physics
The rule that "a healthy ratio is 3x and up" comes from the venture/SaaS world: investors need at least a 3x buffer to cover the costs NOT included in CAC — rent, salaries, support, taxes. This is an empirical rule, not a mathematical formula.
For a local service business (LUMÉ) the rule doesn't transfer perfectly, but the logic ("1x is breaking even, not turning a profit") still holds. The model shows a ratio of ~1.4x in the best-case scenario — an honest, unembellished result.
07The payback period — the second dimension people forget
LTV:CAC is about "is this profitable in the end." Payback is about "will there be enough cash to survive long enough to get there." A campaign can be super-profitable by LTV:CAC (5x) but take 20 months to pay back — if the business doesn't have 20 months of cash runway, it goes bankrupt BEFORE getting that payoff, even if everything looks great on paper.
08ROMI, ROI, ROAS — three terms everyone confuses
| ROI | a general term, "return on any investment," not just marketing |
| ROMI | the same idea, but specifically about marketing spend |
| ROAS | calculated on REVENUE, not profit — the simplest and most misleading metric |
The model uses ROMI on contribution margin — the honest version, showing real profitability, not just turnover.
09The model's most expensive insight: response speed
HOW FAST a person responds in Direct is one of the strongest levers in the entire model, stronger than the size of the ad budget. Per LeadResponse.co and Naiva.ai (who specialize specifically in DM bookings for beauty salons): the baseline "direct → booking" conversion is usually below 15%, rising to 35-50% with disciplined fast responses — most salons lose 30-40% of bookings to slow or missed replies.
These are vendor sources (they sell Direct-automation tools), not independent academic research — the same level of reliability as the other industry blogs cited in this project. The direction of the effect is further confirmed by an independent B2B study of 100,000+ leads (Oldroyd, MIT/InsideSales 2007; Harvard Business Review 2011): a 5-minute response instead of 30 gives a 100× higher chance of reaching the lead and a 21× higher chance of qualifying their interest — a different context, but the same effect. A person who just messaged Direct is at the moment of maximum interest, and without a fast response that window closes. An important caveat on sources: the decay of interest over time is, on its own, a universal psychological pattern that doesn't depend on country — but the exact percentages (15%/35-50%, 100×/21×) were measured on American beauty salons and B2B leads — there's no Ukrainian study with the same precise figure, so it's used as an order of magnitude, not a local coefficient. This ties directly into the future Project 4 (inquiry routing and SLA) — and there it's just as logical to count Telegram, not only Instagram Direct: time spent on Telegram in Ukraine grew 8× since the start of the full-scale war (Kantar Ukraine), and 81% of Ukrainians use it for messaging (USAID/Internews 2024).
10Analogy: the leaky bucket
A business is a bucket of water. Advertising is the tap pouring water (new clients) in from the top. But there are holes in the bottom — clients who churn out. If the holes are big (low retention), you can pour in as much water as you want — the level barely rises, because it all drains out the bottom. A junior marketer sees an empty bucket and says "we need a stronger tap" (a bigger budget). A senior marketer looks at the bottom first: patches the holes (improves retention), and only then increases the flow.
11Common junior mistakes
- Calculating LTV on revenue — it's the first number you see in a report, while margin has to be specifically computed.
- Judging a campaign by its first month — reports are monthly, and there's a temptation for a quick "did it work or not."
- Only looking at one scenario — building one is easier than building two.
- Believing budget growth linearly produces client growth — advertising has diminishing returns.
12How to work out "what if" yourself
If asked about a scenario that isn't in the model, reason through the logic out loud:
- Conversion drops somewhere → CPA goes up (fewer paying clients for the same budget).
- Retention goes up → LTV goes up, and LTV:CAC improves for two reasons at once.
- The average check goes up → contribution per visit and LTV both go up, but CPA doesn't change.
- Budget goes up → clients grow proportionally, CPA stays the same in the model (it doesn't account for audience burnout).
13The model's limitations — say this yourself, before you're asked
A good specialist is defined not by "a flawless model" but by an honest understanding of its boundaries:
- Doesn't account for CPA rising as budget increases (in reality it usually does).
- Some figures (15% no-show, 8% click→direct, 8-month lifetime) are reasoned estimates, not confirmed facts for the Ukrainian market.
- Treats an "average" client — in reality segments are very different. Same with the technician: Altegio (a Ukrainian-language CRM blog) warns that an averaged retention of ~65% can mask a large spread between the best and worst individual technician at a salon.
- Seasonality is now partly handled by the "Season" toggle (see section 14 below) — but the specific multipliers (peak ×1.4, trough ×0.8) are also an estimate, not a measured Ukrainian waxing-seasonality curve.
- The contribution margin (~52%) was sanity-checked against real Ukrainian beauty-retail margins (EVA — 3.46% net margin, MAKEUP — 2.62%, Brocard — ~6%, 2025), but that's retail selling physical goods, not a service business — direct transfer isn't valid, the figures are given only as an order-of-magnitude sanity check.
- The cohort retention curve (section 14) is a simplification: it assumes the same retention percentage repeats every month, though a real curve is usually steeper at the start.
- The blended CAC by channel (section 14) uses a single estimated figure for "alternative channels" (100₴) — no exact paid/free traffic split for the Ukrainian market was found.
Saying this yourself at an interview is a strong move: it shows you understand the model, rather than just believing the numbers on the screen.
14What was added beyond the base model — and why
After the first version, the limitations were honestly listed (section 13) — but some of them could be not just acknowledged, but partly closed. Four additions turn "we didn't account for this" into "here's how it can be accounted for, caveats and all."
Season — now a toggle, not just a caveat. A second toggle at the top of the page (trough/average month/peak) changes the client's visit frequency: peak — 1.4 visits/month instead of 1.1, trough — 0.8. There's no exact figure for Ukrainian waxing seasonality; this is an estimate of the direction of the effect.
Blended CAC — not the whole economics runs through advertising. A new block calculates not just the paid CAC from the funnel, but a blended figure that accounts for the share of clients coming from word of mouth and Google Maps/local search (20% by default, adjustable via a slider). The alternative-channel figure (100₴) is an estimate, not a found fact.
Cohort retention curve — not one number, but a trajectory. The model used to say "retention 35-50%" as a single figure. A new block shows a month-by-month curve: what percentage of the original client group remains active month after month, under the simplification that the same retention percentage repeats at every transition.
Model sensitivity — what to move first. A new block changes each of the 11 variables by ±20% in turn and shows how much LTV:CAC swings. It's honestly visible that the ad budget and CTR barely move the result in this model (that's the exact same point from the limitations about CPA rising with budget — now visible live), while the cost per click and the check economics move it the most.