What is a customer?
A customer is a person or organization that buys a company’s goods or services. Customers generate the sales that let businesses operate and grow, so understanding who they are and what they want is central to most commercial activity.
Key definitions
– Customer: an individual or business that purchases products or services.
– Consumer: a person or business that uses or consumes a product or service. A customer can also be a consumer, but a customer might buy on behalf of others or for resale.
– External customer: someone outside the company who buys the finished goods or services.
– Internal customer: a person or group inside the organization that relies on another internal team for inputs or services (for example, HR serving employees).
– Churned customer: a former buyer who has stopped purchasing.
– Loyal customer: a repeat buyer who continues buying over time and often recommends the business.
Why customers matter
Customers are the primary source of revenue. Companies compete for customers by improving product quality, lowering prices, advertising, or creating better experiences. Satisfied customers tend to buy repeatedly, leave positive reviews, and refer others; dissatisfied customers can reduce sales and harm reputation.
Studying customers
Businesses segment customers to tailor marketing and product decisions. Common segmentation variables include age, income, location, and other demographics. Firms build “customer personas” — composite profiles that represent target buyers — to guide product design, pricing, and messaging. Academic courses in consumer behavior exist to analyze buying motives and decision processes.
Customer service and channels
Customer service is the set of activities that create positive purchase and post-sale experiences. It now includes real-time chat, texting, social media, and traditional phone and in-person support. In crowded markets, service quality often becomes the differentiator between otherwise similar offerings. Quality-improvement frameworks such as Six Sigma sometimes treat customer service as a key element of competitive advantage.
Types of customers (short list)
Common types include:
– Prospective: evaluating but not yet purchased.
– Window shoppers: browsing without intent to buy now.
– New: first-time buyers.
– Loyal: repeat buyers with ongoing value.
– Promotion-driven: buy when there’s a deal.
– Impulse: make unplanned purchases.
– Churned: have stopped buying.
– Angry: dissatisfied and vocal.
– Brand advocates: actively recommend your product.
– Referred: come through another customer.
– International: buyers outside the company’s home market.
What customers value most
Customers generally prioritize product/service quality, fair pricing, reliable customer service, and the ability to give feedback that the company addresses.
A short checklist for businesses handling customers
– Segment: identify customer groups (demographics, behavior, source).
– Profile: build customer personas for each major segment.
– Measure: track purchase frequency, revenue per customer, and churn.
– Serve: provide multi-channel customer support (chat, phone, email).
– Listen: collect feedback and act on it; acknowledge customer input.
– Reward loyalty: create incentives for repeat purchases and referrals.
– Iterate: update products, pricing, and service based on results.
Step-by-step: how to use this checklist (practical approach)
1. Gather sales and contact data for the last 12 months.
2. Segment by purchase frequency (one-time, occasional, repeat).
3. For each segment, compute average purchase value and count of purchases.
4. Prioritize segments that generate most revenue or have growth potential.
5. Design targeted retention tactics (e.g., loyalty offers for repeat buyers; onboarding and education for new customers).
6. Deploy multi-channel support and a feedback loop to log issues and fixes.
7. Reassess metrics quarterly and adjust tactics.
Worked numeric example: annual value of a loyal customer vs. a one-time buyer
Assumptions:
– Average order value = $40
– Loyal customer buys once per month (12 times/year)
– One-time buyer buys once
Annual revenue:
– Loyal customer: $40 × 12 = $480 per year
– One-time buyer: $40 × 1 = $40 per year
If a company’s gross margin on products is 30%:
– Loyal customer gross profit per year: $480 × 0.30 = $144
– One-time buyer gross profit: $40 × 0.30 = $12
Interpretation: under these assumptions, a single loyal customer delivers 12× the revenue and 12× the gross profit of a one-time buyer over a year. This simple arithmetic illustrates why many firms emphasize retention and loyalty programs.
The bottom line
Customers are the essential asset of any business. Classifying customers, studying their behavior, and delivering high-quality service are practical ways to increase repeat purchases, reduce churn, and grow revenue. Loyal customers typically provide the best long-term value because they buy repeatedly and recommend the business to others.
Sources
– Investopedia — Customer: https://www.investopedia.com/terms/c/customer.asp
– BMC Software — Internal vs. External Customers: https://www.bmc.com/blogs/internal-vs-external-customers/
– Semrush — The Marketing Funnel: What It Is & How It Works: https://www.semrush.com/blog/marketing-funnel/
– Zendesk — 12 Types Of Customers + How to Support Them: https://www.zendesk.com/blog/12-types-of-customers/
– The Council for Six Sigma Certification — What Is Six Sigma?: https://www.sixsigmacouncil.org/what-is-six-sigma/
Educational disclaimer
Educational disclaimer: This information is educational only and not individualized investment, legal, or business advice. Use it to inform decisions, not as a substitute for professional counsel tailored to your situation.
Practical checklist — measuring customer value and retention
– Define the measurement period (monthly, quarterly, yearly) and use the same period consistently.
– Track active customers at start and end of the period, new customers acquired during the period, and number of customers lost (churned).
– Collect revenue per customer, number of purchases per customer, and gross margin (revenue minus direct costs) to compute economic value.
– Compare average Customer Lifetime Value (CLV or LTV) to Customer Acquisition Cost (CAC). Aim for an LTV:CAC ratio meaningfully above 1 (many firms target ~3:1).
– Segment customers (by cohort, channel, product, or demographic) and compute metrics per segment for actionable insight.
Key formulas (simple, practical versions)
– Retention rate (period): Retention = (Customers at end of period − New customers during period) / Customers at start of period
– Churn rate (period): Churn = (Customers lost during period) / Customers at start of period
– Simple CLV (no discounting): CLV = Average purchase value × Purchases per period × Average number of periods a customer remains active
– CLV incorporating margin: CLV_margin = (Average purchase value × Purchases per period × Gross margin %) × Average lifespan (periods)
– LTV:CAC ratio = CLV / CAC
Worked numeric example
Assumptions:
– Average order value = $50
– Purchases per year = 4
– Gross margin = 60% (0.60)
– Average customer lifespan = 5 years
– CAC = $150
Step calculations:
1) Annual revenue per customer = $50 × 4 = $200
2) Annual gross profit per customer = $200 × 0.60 = $120
3) CLV_margin (5-year) = $120 × 5 = $600
4) LTV:CAC = $600 / $150 = 4.0
Interpretation: With these assumptions, each acquired customer yields $600 of gross profit over their lifetime versus $150 to acquire them — a healthy 4:1 ratio. If average lifespan increases to 6 years, CLV becomes $720 (20% higher), illustrating the leverage of retention.
Quick scenario: improving retention vs. acquisition
– If you reduce annual churn so average lifespan rises from 5 to 6 years, CLV increases 20% (as above). To achieve the same 20% revenue gain by acquiring new customers would require proportionally higher acquisition spend and creates more dependency on CAC efficiency.
Step-by-step process to improve customer value (operational)
1) Instrumentation: Ensure customer events (signup, purchase, renewal, support interactions) are tracked and stored by customer ID.
2) Baseline: Compute current retention, churn, CLV, CAC, and segment-level metrics.
3) Hypothesis: Pick one lever (onboarding, price, product quality, support response) and state expected metric change (e.g., “Improve 30-day retention by 5 percentage points”).
4) Experiment: Implement a controlled test (A/B test or cohort rollout). Track primary metric and secondary business metrics.
5) Analyze: Use cohort analysis to separate new
new-customer
customer effects from changes to existing cohorts. Compute lift by cohort (treatment minus control), test for statistical significance, and inspect secondary metrics (ARPU, support tickets, returns) for unintended trade-offs. Key checks:
– Ensure the test ran long enough to capture the metric (e.g., 30-day retention needs ≥30 days of post-signup data).
– Confirm treatment and control are balanced on acquisition source, geography, and initial spend.
– Look for selection bias (e.g., if the experiment changes who converts).
6) Rollout and scale: If the experiment shows a robust positive lift, define a staged rollout plan. Start with a percentage rollout (10% → 25% → 50% → 100%) and monitor leading indicators at each step. Maintain rollback criteria (e.g., >5% relative decline in gross margin or NPS).
7) Continuous monitoring and governance: Add key signals to dashboards and set alert thresholds (daily/weekly cadence depends on business velocity). Recommended metrics to monitor:
– Acquisition: CAC, conversion rate by channel.
– Engagement: 7/30/90-day retention, DAU/MAU where relevant.
– Financial: ARPU (average revenue per user), gross margin per user, CLV (customer lifetime value).
– Risk: refund rate, support volume, compliance flags.
8) Iterate: Treat CLV improvement as an ongoing program. Repeat the hypothesis → experiment → analyze cycle quarterly. Use learnings to refine segmentation (e.g., high-LTV vs low-LTV cohorts) and prioritize levers with highest ROI.
Worked numeric example (subscription business)
Assumptions:
– Monthly ARPA (average revenue per account) = $50
– Gross margin on subscription revenue = 60% → monthly gross margin per user = 50 * 0.60 = $30
– Baseline 30-day retention = 80% → monthly churn = 1 − 0.80 = 0.20
– Post-improvement retention = 85% → churn = 0.15
Simple steady-state CLV formula for subscription businesses (constant margin and churn, no discounting):
CLV = monthly gross margin / monthly churn
Baseline CLV = 30 / 0.20 = $150
Improved CLV = 30 / 0.15 = $200
Absolute increase = $50
Relative increase = $50 / $150 = 33.3%
Interpretation: A 5-percentage-point improvement in 30-day retention raised CLV by one third under these assumptions. That extra CLV can be reinvested in acquisition, used to justify premium pricing, or flow to the bottom line.
Checklist — experiment readiness
– Instrumentation: events tied to persistent customer ID (signup, subscription start, purchase, refund, support).
– Baseline computed for segments (retention curves, ARPU, CAC, margin).
– Hypothesis with quantified target and primary/secondary metrics.
– Experiment design: sample size, randomization method, duration.
– Analysis plan: cohort comparisons, significance test, lift and confidence intervals.
– Rollout plan with staged percentages and rollback rules.
– Post-rollout monitoring dashboard.
Instrumentation operational checklist
– Use a single canonical customer identifier across systems (auth ID, CRM ID).
– Track event timestamp, event type, product/plan, amount, channel, and campaign ID.
– Persist raw event stream for reprocessing (data warehouse or event store).
– Emit both raw events and derived user-state (e.g., active/subscribed) for fast queries.
– Maintain data lineage and schema versioning so experiments
…can be reproduced and historical metrics recalculated.
Data governance and privacy
– Map personally identifiable information (PII) fields and apply minimization: only emit user identifiers necessary for the experiment. Define retention periods and anonymization rules.
– Consent and legal checks: ensure tracking aligns with consent flags, cookie laws (GDPR/CCPA) and marketing preferences. Block events if consent is revoked.
– Access controls and audit logs: limit who can join experiment cohorts and who can modify experiment settings. Record changes (who/when/why) to experiment metadata.
Instrumentation QA checklist (pre-launch)
1. Canonical ID: verify a single canonical customer identifier is present and identical across frontend, backend, and analytics.
2. Event schema: check required fields (timestamp, event_type, product_id, amount, channel, campaign_id, experiment_id, cohort, user_id).
3. Volume sanity: baseline event volume vs production for the same time window (±10% expected). Investigate large deltas.
4. Duplicate suppression: ensure idempotency keys or deduplication logic for repeated events.
5. Timezone and latency: store UTC timestamps; measure max expected event delay and label late-arriving events.
6. Backfill capability: confirm raw stream is persisted and reprocessing jobs work on historical data.
7. End-to-end test users: create test accounts for each cohort and validate metrics from event through dashboard.
8. Schema versioning: register schema changes and migration plans before changing production event definitions.
Statistical controls and experiment hygiene
– Randomization integrity: randomize at the correct unit (user, session, account) and test for balance on key covariates (country, device, plan).
– Guard against peeking (optional stopping): repeated significance checks inflate false positives. Use predetermined sample sizes or sequential testing methods (alpha spending, Bayesian approaches).
– Multiple comparisons: if you test many metrics or variants, control for family-wise error (Bonferroni) or false discovery rate (Benjamini–Hochberg).
– Blocking and stratification: stratify randomization by important buckets (e.g., region, plan) if those influence the primary metric.
Sample-size calculation — worked numeric example
Goal: detect a lift in conversion rate from 10% to 11% (absolute difference 1%), α = 0.05 (two-sided), power = 80%.
Formula (two-proportion z-test approximate):
n_per_arm = [ Z_{α/2} * sqrt(2*p*(1-p)) + Z_{power} * sqrt(p1*(1-p1)+p2*(1-p2)) ]^2 / (p1 – p2)^2
where p = (p1 + p2) / 2.
Numbers:
– p1 = 0.10, p2 = 0.11, p = 0.105
– Z_{α/2} = 1.96, Z_{power} ≈ 0.842
Compute:
– Term1 = 1.96 * sqrt(2*0.105*0.895) ≈ 0.85
– Term2 = 0.842 * sqrt(0.1*0.9 + 0.11*0.89) ≈ 0.37
– Sum ≈ 1.215
– n
≈ (1.215)^2 / (0.01)^2 ≈ 1.476 / 0.0001 ≈ 14,762 per arm.
So you need about 14,762 users in the control arm and 14,762 in the treatment arm (total ≈ 29,524 users) to detect an absolute lift from 10% to 11% with two‑sided α = 0.05 and 80% power, using the approximate two‑proportion z‑test formula.
Practical implications and checks
– Sample definition: n refers to the number of independent user observations (not the number of conversions). Make sure your unit of analysis (user, session, cookie) matches how you compute conversion rates.
– Allocation: if you split traffic 50/50, each visitor is assigned to one arm; if you use different allocation ratios adjust n_per_arm accordingly.
– Attrition and eligibility: inflate n to cover ineligible users, bots, or expected dropouts. Common practice: target_sample = n_per_arm / expected_eligibility_rate. Example: if only 80% of visitors are eligible, divide by 0.8.
– Time to run the test: compute how many eligible users you get per day per arm. Example: if your site has 5,000 eligible users/day and you split 50/50 => 2,500 users/arm/day, so days needed ≈ 14,762 / 2,500 ≈ 5.9 days. Round up and run at least one full week to capture weekly cyclicality.
– Minimum duration: run for at least one full business cycle (often 1–2 weeks) even if you reach sample size earlier, to avoid seasonality bias.
Common variations and formulas
– One‑sided test: if you only care about an increase (not a decrease), use Z_{α} instead of Z_{α/2}; required n will be smaller. Be explicit and pre‑register this choice.
– Continuous metric (e.g., revenue per user): sample size per arm ≈ [ (Z_{α/2} + Z_{power})^2 * 2σ^2 ] / d^2, where σ is the metric standard deviation and d is the absolute difference you want to detect. Worked example: σ = 20 currency units, d = 5 units: Z sum ≈ 2.802, square ≈ 7.85, numerator = 7.85 * 2 * 400 = 6,280, n ≈ 6,280 / 25 ≈ 252 per arm.
– Minimal Detectable Effect (MDE): rearrange the formula to solve for d given n; useful when traffic is fixed.
Sequential testing and peeking
– Classic fixed‑sample calculations (like above) assume no interim looks. Repeatedly peeking