Cohort Revenue Projection from Meta Acquisition — Forecasting D2C LTV by Month

A customer you acquire on Meta this month is not worth their first-order contribution. They are worth the stream of revenue they generate across months 2, 3, 6, 12.

Indian D2C founders who don't project cohort revenue forward end up under-investing in acquisition. They set CAC tolerance against first-order math when they should set it against 12-month cohort math.

Why Cohort Projection Matters

Three reasons:

• You spend acquisition cash today. Revenue comes back over 12-24 months.

• LTV is the only number that defends CAC tolerance. First-order CAC is myopic.

• Cohort quality varies by acquisition month. January cohort ≠ June cohort.

The Building Blocks

To project cohort revenue, you need:

1. Month-1 average order value (M1 AOV) from new customers

2. Repeat-purchase probability by month (M2, M3, M6, M12)

3. Average order value of repeat orders (usually 15-30% higher than first orders)

4. Discount rate for time value of money — typically 10-12% annually for Indian D2C

5. Acquisition cohort size (number of new customers per month)

Repeat Probability by Category

Realistic repeat-purchase probabilities for Indian D2C, month-on-month:

Apparel

M2: 12-18%, M3: 8-14%, M6: 22-32%, M12: 38-52%. Strong seasonal patterns.

Beauty / Skincare

M2: 20-32%, M3: 18-28%, M6: 45-65%, M12: 65-85%. Highest repeat rates in D2C.

Jewelry

M2: 4-8%, M3: 6-12%, M6: 14-22%, M12: 28-42%. Low frequency, high AOV.

Home / Furniture

M2: 3-6%, M3: 5-9%, M6: 11-18%, M12: 22-35%. Lowest repeat, longest cycles.

Food / F&B

M2: 25-40%, M3: 25-40%, M6: 55-75%, M12: 75-90%. Subscription-like.

Health / Supplements

M2: 30-45%, M3: 28-42%, M6: 60-80%, M12: 78-92%. Subscription-favourable.

Worked Example: Apparel Brand

Inputs:

• M1 AOV: ₹1,800

• M2 repeat probability: 15%, AOV ₹2,100

• M3 repeat probability: 10%, AOV ₹2,100

• M6 repeat probability: 28%, AOV ₹2,200

• M12 repeat probability: 45%, AOV ₹2,400

Per customer, undiscounted revenue across 12 months: ₹1,800 + (₹2,100 × 0.15) + (₹2,100 × 0.10) + (₹2,200 × 0.28) + (₹2,400 × 0.45) = ₹1,800 + ₹315 + ₹210 + ₹616 + ₹1,080 = ₹4,021.

At 40% contribution margin: ₹1,608 in contribution per acquired customer over 12 months. That's your real LTV input to CAC tolerance.

Why Most Projections Are Wrong

• Using gross margin instead of contribution margin. Ignores shipping, returns, discounts.

• Projecting from one cohort. Quality varies — use 6-month rolling average.

• Not discounting for time value of money. ₹2,000 in month 12 is not worth ₹2,000 today.

• Including refunded orders. Net of returns, always.

• Ignoring acquisition-channel difference. Meta-acquired vs Google-acquired vs organic — different LTV.

How to Build a Projection Model

1. Pull 12-month order history segmented by acquisition month

2. Calculate cumulative purchase rate for each cohort at M1, M2, M3, M6, M12

3. Average across at least 6 cohorts to smooth out seasonality

4. Apply discount rate (10-12% annual)

5. Refresh monthly as new cohorts mature

Using Projections to Set CAC Tolerance

Once you have projected LTV per acquired customer:

• CAC tolerance = LTV × 0.33 for healthy 3x LTV/CAC ratio

• For aggressive growth: LTV × 0.40-0.50 (LTV/CAC of 2-2.5x)

• For defensive cash management: LTV × 0.20-0.25 (LTV/CAC of 4-5x)

See [Revenue tier vs CAC tolerance](https://www.wittelsbach.ai/post/revenue-tier-vs-cac-tolerance-math-d2c-meta-ads-india-2026-playbook) for the full math.

Cohort Quality Differences by Acquisition Channel

Important: not all acquired customers are equal.

• Meta-acquired (broad targeting): Baseline LTV.

• Meta-acquired (interest-stacked): 10-15% higher LTV.

• Meta-acquired (lookalike of top customers): 25-40% higher LTV.

• Google Shopping: 20-30% higher LTV (higher intent).

• Organic / direct: 50-80% higher LTV (highest intent).

• Marketplaces: 20-30% lower LTV (price-sensitive, less loyal).

How Wittelsbach AI Projects Cohort Revenue

Bach AI ingests order history, segments cohorts by acquisition channel and creative, projects 12-month revenue per cohort, and surfaces which Meta campaigns are actually building long-term value vs short-term revenue. The CAC tolerance number stops being a guess. Connect your Meta account at [app.wittelsbach.ai](https://app.wittelsbach.ai) for a free audit.

Frequently Asked Questions

How long do I need to wait before projecting cohort LTV?

Minimum 90 days of cohort data, ideally 180. Projecting 12-month LTV from 30-day data systematically overestimates by 25-40% because early cohorts show repeat behaviour that doesn't generalise. Most Indian D2C brands set CAC tolerance using 30-60 day cohort data and over-spend as a result.

Should I project LTV per SKU or per customer?

Per customer, segmented by first-order SKU. Customers acquired via best-seller SKU often show different LTV than customers acquired via niche SKU. Brands that ignore this end up over-investing acquisition spend in low-LTV-acquired customers.

How do I handle seasonal cohorts?

Average across at least 12 months of cohorts. October-November cohorts in India often show different patterns from January-February cohorts due to festive shopping. Single-cohort projections trained on October will overestimate January acquisition value.

What discount rate should I use for cohort projection?

10-12% annually for most Indian D2C. This represents cost of capital + uncertainty. For VC-funded brands burning cash, use 15-18% (higher cost of capital). For mature, profitable brands, 8-10% is defensible. The number matters more than people realise — small changes compound.

How often should I refresh my cohort projection model?

Monthly minimum, quarterly at most. Cohort quality drifts as you scale acquisition — the customer you acquired at ₹50L/month spend is different from the one at ₹3Cr/month spend. Brands that refresh quarterly catch the drift; brands that refresh annually consistently over-spend during scale-up periods.