Когорты. Часть 3

Finance / Business-Performance Summary (Cohorts & “Economy” Model)

The presenter builds a cohort-based marketing accounting model across 12 cohorts / 12 months to show how marketing decisions—such as acquisition spend and retention/reactivation—affect:

• the current cohort immediately, and
• all past cohorts over time (via returning users),

which then determines an “economy” metric (cashflow/profit-like balance).

Disclosures: The subtitles include no explicit “not financial advice” disclaimer.

---

Instruments / Assets / Tickers / Sectors Mentioned

• None. The content is focused on operational/marketing analytics, and the only numeric values referenced are marketing costs in rubles.

---

Key Entities / Metrics Used

Cohort Grid Components (Table Tabs & What They Track)

• Marketing tab: audience acquisition costs (“costs for attracting an audience”).
• Traffic tab: number of incoming users per cohort per month.
• Returns / reactivations: returning users over time (“returns while slides”).
• Clients per cohort per month: number of clients each month for each cohort.
• Transactions per month: includes repeat purchases within cohorts.
• Conversion rate: conversion measured per month within a cohort (and “starting conversion” is mentioned).
• Average cost per click (CPC).
• Cost per visitor (CPV): derived from marketing spend / cohort traffic.
• Revenue / money turned over: revenue generated from visitors/transactions.
• “Economy” metric: revenue minus marketing spend (examples below).
• ARPU: discussed as growing/stabilizing.

---

Methodology / Step-by-Step Framework

1. Set up cohort accounting

   • Assume 12 cohorts formed over 12 months.
   • For each cohort, track month-by-month:
     • new-user inflow
     • returning/reactivated users
     • repeat purchases / transactions
     • conversion rate and transaction volume
     • acquisition cost drivers (e.g., CPC, cost per visitor)

2. Compute unit economics per cohort per month

   • Use acquisition costs and traffic to compute cost per visitor.
   • Multiply conversions by transaction frequency to compute revenue (“money we were able to turn over”).

3. Compute “economy” per cohort

   • Economy = revenue from cohort (including repeats) − marketing spend

4. Simulate scenario changes

   • Change acquisition spend month-to-month (e.g., keep at 75k vs increase).
   • Add retention/reactivation spend (e.g., “work … with … warmth of bees” to bring back old users).
   • Observe how outcomes propagate:
     • the current cohort reacts immediately
     • prior cohorts improve later due to returning users

5. Use the full cohort table

   • Extend/fill the table through the end of the timeline to see cumulative effects.
   • Compare how long it takes until each cohort’s monthly economy turns positive (“positivity timing”).

---

Key Numbers & Explicit Scenario Outcomes (Examples)

Values are in rubles. Figures are described with round numbers, and some subtitle-derived values appear partially garbled.

Baseline Example: Cohort 1

• Month 1 (January) marketing spend: 75,000 rubles
• Users/clients acquired from this spend: 15 users (wording suggests “use” = users)
• Transactions from those clients: 110 transactions (includes repeat/conditional repeats)
• Economy for cohort 1 after month 1: -25,500 rubles
  • Economy is negative because there is not enough return in the first period.

February: Increased Spend + Retention/Reactivation

• Marketing spend in February: 75,000 rubles
• Model adjustments described:
  • “price has increased” / fewer users for the same spend (e.g., from 15 down to about 14 / 70,000 referenced)
  • plus reactivation:
    • spend 5,000 rubles to bring back old users
    • returning users added: 5 (example)
• Economy improvement example (cohort 1):
  • economy improves from -25,500 to 24,650 rubles (becomes positive)
• Economy for cohort 2: described as still negative

March: Higher Acquisition Spend

• Planned March spend: 80,000 rubles
  • Subtitles also reference 75,000 and additional 10,000 (inconsistent due to garbling)
• Result described:
  • cohort economies shift; cohort 1 worsens slightly, 2 improves, and 3 becomes best
  • the presenter emphasizes changes propagate across all cohorts, not only the newest one.

Major Campaign Scenario

• Total “culture-based advertising campaign” budget: 100,000 rubles
• Budget distribution examples mentioned:
  • 85,000 into one channel
  • 7,000 into another
  • 8,000 left/other (plus small additional allocations like 6, 2,000, etc., shown as garbled)
• Claimed effect:
  • “conversion has grown sharply by month”
  • later cohorts reach near/positive states while early months for cohort 1 remain negative (first-month lag)

End-State / Filled-Table Example

• Total marketing expenses across cohorts in the model: 1,700,000 rubles
• Growth assumptions (partially garbled but key points):
  • traffic grows from 10,000 baseline to 10,000+ over time
  • returning users increase from around 500 in early months to higher levels later
• Outcome described:
  • by the end of the modeled history, all cohorts show positive dynamics
  • cohort 1 takes longer to become positive due to initial negative months.

Performance Comparison / Ranking

• Cohort 1 produces the most money in the model:
  • referenced value: 199,000 (described as highest)
• Even though early months were negative, growth occurred as:
  • conversion increased
  • ARPU stopped growing after learning improvements around March (explicitly mentioned)

---

Explicit Recommendations / Cautions

Recommendation

• Use a cohort-by-cohort table to understand:
  • accumulation timing, and
  • cross-cohort effects of marketing decisions.
• The presenter states they will make the table public and provide links (in the description) to prior videos covering related concepts.

Caution / Interpretation

• A single analytics number (e.g., cumulative revenue) can be misleading:
  • you don’t earn it immediately; it accumulates over time across cohorts.

Operational Conclusion

When increasing spend, account for:

• the immediate effect on the current cohort’s acquisition
• the delayed effects on prior cohorts via retention/reactivation
• the possibility that monthly conversion metrics may not fully reflect cumulative cohort impact

---

Presenter / Source

• Presenter: Daniil Konin
• He runs a blog about data-driven decision-making.