Practical uses: customer segmentation for targeted marketing
Your blended Cost Per Unit to acquire a customer tells you what you spent on average. It does not tell you which customers are worth the spend - and that gap is where most Marketing Budgets get wasted.
Customer segmentation splits your Demand into distinct groups that behave differently - different Lifetime Value, different Churn Rate, different Cost Per Unit to acquire. Operators who segment make better Allocation decisions on Marketing Spend because they can see which groups of Buyers generate Profit and which groups destroy it.
Customer segmentation is the practice of dividing your total Demand into groups where members within each group behave more similarly to each other than to members of other groups.
The grouping variables depend on what you sell, but common ones include:
The key insight: your Demand is not one homogeneous pool. A $48/month user who found you through a blog post and a $48/month user who came through a paid ad slot are different animals - they Churn at different rates, they respond to different Upsell offers, and they cost different amounts to acquire.
Segmentation turns a blurry average into a set of distinct Unit Economics profiles.
Segmentation hits the P&L in three places:
1. Marketing Spend efficiency. If you know Segment A has an 8% Close Rate from a $2 ad slot and Segment B has a 3% Close Rate from a $6 ad slot, you can shift Budget toward A and immediately improve your Cost Per Unit of acquisition. This is marginal dollar allocation with real numbers attached.
2. Churn Rate reduction. Churn is rarely uniform. One segment might Churn at 4%/month while another Churns at 8%/month. If you treat them the same - same first-30-days experience, same Churn-reduction email sequences, same Service Recovery playbook - you are under-investing in the high-Churn group and over-investing in the group that would have stayed anyway. That is Value Leakage.
3. Lifetime Value accuracy. Lifetime Value calculated as a single average is a fiction. If your best segment has 2x the Lifetime Value of your worst, and you are setting Pricing and spending based on the blended number, you are subsidizing unprofitable customers with profitable ones. Your P&L looks fine until the mix shifts - then Revenue drops and you do not understand why.
Segmentation gives you the resolution to make Allocation decisions that actually match your Unit Economics.
Step 1: Pick your segmentation variables.
Start with data you already have. If you run a SaaS product, your database has: acquisition source, plan tier, feature usage, support tickets, renewal history. You do not need a Conjoint Analysis or a Scoring Model on day one - start with what the system already records.
Step 2: Calculate Unit Economics per segment.
For each segment, compute:
This is the gross Lifetime Value - the total Revenue a customer is expected to generate over the full relationship. It does not include acquisition cost. To get per-customer Profit, subtract Cost Per Unit to acquire separately:
Net customer value = Lifetime Value minus Cost Per Unit to acquire
Keeping these separate matters. The ratio of Lifetime Value to Cost Per Unit tells you how efficiently your Marketing Spend converts to Revenue. A ratio of 3:1 means every $1 spent acquiring a customer returns $3 in Lifetime Value. If you bake Cost Per Unit into the Lifetime Value definition, that ratio becomes meaningless - and you will be confused the first time you encounter a "3:1 Lifetime Value to Cost Per Unit ratio" in the wild.
Step 3: Rank and decide.
Sort segments by net customer value or by the ratio of Lifetime Value to Cost Per Unit. You now have a decision rule:
Step 4: Build targeted marketing.
Once you know which segments are valuable, tailor your positioning and ad slots to reach those specific Buyers. A generic message to your entire target audience wastes Budget on segments that will not convert or will not stay. Segment-specific messaging improves Close Rate because you are speaking to the specific problem each group has.
Segment when the Variance in customer behavior is large enough to act on. If every customer looks roughly the same - same Churn Rate, same Revenue, same Cost Per Unit - segmentation adds complexity without value. But this is rare.
Concrete triggers:
Do not segment when you have fewer than ~200 customers. Below that, your samples per segment are too small to distinguish real patterns from noise. Use your time on Execution instead.
You sell project management software at $48/month. Your Marketing Spend last quarter was $30,000 across two channels:
Compute Cost Per Unit to acquire. Channel A: $10,000 / 100 = $100 per customer. Channel B: $20,000 / 100 = $200 per customer.
Compute gross Lifetime Value. Channel A: $48 / 0.04 = $1,200. Channel B: $48 / 0.08 = $600.
Compute net customer value (Lifetime Value minus Cost Per Unit). Channel A: $1,200 - $100 = $1,100. Channel B: $600 - $200 = $400. Both channels are profitable.
Compute Lifetime Value to Cost Per Unit ratio. Channel A: $1,200 / $100 = 12:1. Channel B: $600 / $200 = 3:1. Each dollar spent on Channel A returns 4x more Lifetime Value than a dollar spent on Channel B.
Compute total net value under current Allocation. Channel A: 100 x $1,100 = $110,000. Channel B: 100 x $400 = $40,000. Combined: $150,000.
Compare to full reallocation. If you shift the entire $30,000 to Channel A (assuming similar Close Rate at scale): $30,000 / $100 = 300 customers x $1,100 = $330,000 in net value. That is $330,000 versus $150,000 - a 2.2x improvement from the same Budget.
Insight: Both channels are profitable, and the blended numbers look fine: $150 average Cost Per Unit, $900 average Lifetime Value, 6:1 ratio. But the per-segment view reveals Channel A produces $11 of net value per marketing dollar while Channel B produces $2. The average hid an opportunity cost of $180,000 in forgone value. Note the assumption: this analysis assumes Channel A's Close Rate holds at 3x the current spend. In practice, most channels hit diminishing returns at scale, so the real optimum is likely a heavier tilt toward A rather than full reallocation - but the direction of the decision is unambiguous.
You have 500 active customers on a $99/month plan. You want to launch a $199/month premium tier. You segment customers by feature usage into three groups:
Estimate Close Rate for the Upsell by segment. Power users already hit the ceiling of the current plan - estimated 25% Close Rate. Core users might close at 5%. Light users are unlikely to pay more for features they do not use - estimated 1%.
Compute expected Expansion Revenue per segment per year. Power: 80 x 0.25 x ($199 - $99) x 12 = $24,000. Core: 300 x 0.05 x $100 x 12 = $18,000. Light: 120 x 0.01 x $100 x 12 = $1,440.
Compute Implementation Cost per segment. Each segment gets a targeted email sequence plus a dedicated landing page. At $15 per customer - covering email platform fees, copywriting labor, and design work amortized across the segment - the costs are: Power: 80 x $15 = $1,200. Core: 300 x $15 = $4,500. Light: 120 x $15 = $1,800.
Compute ROI by segment. Power: $24,000 / $1,200 = 20x. Core: $18,000 / $4,500 = 4x. Light: $1,440 / $1,800 = 0.8x (loss).
Allocation decision. Run the Upsell campaign for Power and Core users. Skip Light users entirely - that $1,800 has better uses. Consider whether Light users' high support cost and low engagement signal a Churn risk worth addressing through Service Recovery instead.
Insight: Segmentation turned a single Upsell campaign into three different Allocation decisions. Without it, you would have spent $7,500 to reach all 500 customers and attributed the results to the campaign being acceptable - never realizing it was excellent for 80 Buyers, solid for 300, and negative ROI for 120.
Customer segmentation is a Unit Economics exercise, not a marketing exercise. Each segment has its own Cost Per Unit, Churn Rate, and Lifetime Value, which means each segment has its own P&L.
The blended average hides your best Allocation opportunities. Two segments can both be profitable while producing wildly different returns per marketing dollar. Without the per-segment view, you cannot optimize marginal dollar allocation.
Start with data you already have (acquisition channel, plan tier, usage patterns) and compute per-segment Unit Economics before investing in sophisticated Scoring Models or Conjoint Analysis.
Segmenting on demographics instead of behavior. Knowing a customer is a 35-year-old in Texas tells you almost nothing about their Churn Rate or Lifetime Value. Segment on what they do - how they buy, how they use the product, how they leave. Behavior predicts Unit Economics; demographics usually do not.
Creating too many segments too early. Five segments with 20 customers each gives you noisy estimates and no actionable signal. Start with 2-3 segments defined by your highest-Variance metric (usually Churn Rate or acquisition channel). You can refine later when you have more data and more customers.
You run a SaaS product at $60/month. Two segments from different acquisition channels:
Calculate Lifetime Value and net customer value for each segment. Then determine which segment should receive a $15,000 monthly Marketing Spend Budget - and explain why the obvious answer might be wrong.
Hint: Compute Lifetime Value as monthly Revenue divided by monthly Churn Rate (the gross figure). Then subtract Cost Per Unit to get net customer value. Finally, compute what happens if you allocate the full $15,000 to each segment separately.
Segment A: Lifetime Value = $60 / 0.03 = $2,000. Net customer value = $2,000 - $150 = $1,850. Segment B: Lifetime Value = $60 / 0.06 = $1,000. Net customer value = $1,000 - $50 = $950.
Segment A has nearly 2x the Lifetime Value and net customer value. The naive answer is to allocate toward A.
But compute net value per marketing dollar: Segment A = $1,850 / $150 = $12.33 per dollar. Segment B = $950 / $50 = $19.00 per dollar.
$15,000 allocated to A: $15,000 / $150 = 100 new customers x $1,850 = $185,000 net value. $15,000 allocated to B: $15,000 / $50 = 300 new customers x $950 = $285,000 net value.
Segment B generates $285,000 versus $185,000 for A - a 54% improvement - despite having lower Lifetime Value. The lower Cost Per Unit more than compensates for the higher Churn Rate. Lifetime Value alone does not determine Allocation; the marginal return per dollar of Marketing Spend does. One caveat: Segment A's Revenue accrues over a longer Time Horizon (~33 months versus ~17 months for B). If Cash Flow timing or Discount Rate matters to your business, that shifts the analysis - but the directional conclusion holds.
You have a SaaS product with 400 customers. You suspect there are meaningful segments but have not analyzed them yet. What three data points would you pull first, and what calculation would you run on each to decide if segmentation is worth pursuing?
Hint: Think about the three places segmentation hits the P&L: Cost Per Unit to acquire, Churn Rate, and Revenue per customer. What is the simplest split you can make?
Pull three things: (1) Acquisition channel per customer - then compute Cost Per Unit to acquire by channel. If the cheapest channel costs 2x less than the most expensive, there is a segment worth isolating. (2) Monthly Churn Rate by acquisition channel or by plan tier - if Churn varies by more than 2x across any grouping, you have a high-value segmentation axis. (3) Revenue per customer including any Expansion Revenue - if the top 20% of customers generate 50%+ of Revenue, that cluster defines your most valuable segment. If all three metrics show less than 30% Variance across any split, segmentation may not be worth the effort yet - focus on growing total Demand instead.
Customer segmentation refines your understanding of Demand - instead of treating Demand as a single number, you decompose it into sub-populations with distinct Unit Economics. Your prerequisite taught you that Demand sets the ceiling on Revenue; segmentation reveals that ceiling is actually multiple ceilings at different heights, and your job as an Operator is to figure out which ceiling is highest and allocate toward it. Downstream, segmentation feeds directly into targeted marketing (you cannot target what you have not defined), Lifetime Value calculations (the per-segment version replaces the misleading average), Churn Rate management (different segments need different Service Recovery and Churn-reduction strategies), and Pricing decisions (segments with different willingness to pay justify different Subscription Pricing tiers). It also connects to Allocation broadly - once you see that segments have different Unit Economics, every Budget decision becomes a question of which segment to invest in, which is the same marginal dollar allocation logic that runs through the entire P&L.
Disclaimer: This content is for educational and informational purposes only and does not constitute financial, investment, tax, or legal advice. It is not a recommendation to buy, sell, or hold any security or financial product. You should consult a qualified financial advisor, tax professional, or attorney before making financial decisions. Past performance is not indicative of future results. The author is not a registered investment advisor, broker-dealer, or financial planner.