Technology, Media & Telecom — Business Models, Metrics & Diligence
Volume I Scope: Core TMT business models — B2B SaaS revenue mechanics, unit economics, forecasting. Cloud infrastructure, embedded finance, consumer subscriptions, adtech/martech, vertical SaaS. Key metrics tables and model comparison frameworks.
Volume II Scope: Advanced dynamics — network effects and non-contractual retention. Marketplace economics (buyer/seller dynamics, take rates, disintermediation). Consumer platforms, superapps, creator economies. Structural weaknesses of niche models. Diligence red flags. LTV derivations and exit patterns.
Design: This chapter merges both volumes into one comprehensive reference. Each beat is dense — treat it as a full page of a technical reference manual. Designed for investment banking interview prep and due diligence work.
Subscription Types: Monthly vs annual vs multi-year contracts. Annual upfront payments improve cash flow (deferred revenue increases) but signal customer confidence. Multi-year deals reduce visibility into churn (can't observe it for 24+ months).
Revenue Components:
Billing vs. Recognition (ASC 606): GAAP revenue is recognized ratably over the contract period even if payment is annual. Deferred revenue (liability) is a leading indicator — growing deferred revenue suggests a healthy pipeline ahead.
Cohort Analysis: Track each customer cohort's (e.g., those acquired in Q1 2024) revenue contribution over time. Healthy SaaS shows expanding cohorts — each vintage contributes more over time due to expansion revenue overcoming churn.
CAC Deep Dive: Fully-loaded CAC includes: sales rep salary + commission, marketing spend, tools, onboarding support, customer success allocated cost. Many companies understate CAC by 2-3x by excluding indirect costs.
CAC Payback Period: Months to recover the acquisition cost from gross margin. Enterprise SaaS: 12-18 months acceptable (longer contracts justify longer payback). SMB SaaS: under 6 months preferred (short contract terms mean rapid return required).
LTV / CAC Ratio: 3-5x target. Worked example: $5000 CAC, 80% gross margin, $500 ARPU means payback in 12-15 months. LTV = $500 / 0.05 (5% monthly churn) = $10,000. LTV/CAC = 2.0x. To achieve 3.0x, need to either reduce CAC, increase ARPU, or improve churn.
Gross Margin: Software gross margin should be 70-85%. Below 70% signals heavy professional services costs, infrastructure costs, or commodity hosting. As SaaS scales, gross margin typically expands due to fixed platform costs being spread across more customers.
Net Dollar Retention: The single best predictor of SaaS health. >130% is elite. 100-120% is very strong. 100-110% concerning for a growth-stage company. Below 100% = leakage, a fundamental problem even if new customer acquisition is strong.
Customer Concentration Risk: Top 10 customers as % of revenue. Above 20% is high concentration risk. Above 30% is dangerous. A single customer departure could crater revenue or force conservative guidance.
EV/ARR Multiples: Range from 5x (slow-growth, mature, low-margin) to 40x+ (hypergrowth, >100% NRR, high-margin). Rule of 40 is a key determinant — companies above the line trade at premium multiples.
Public Comps Methodology: Select SaaS companies with similar growth/margin profile. Adjust multiples for NRR (higher NRR = higher multiple), gross margin (70%+ commands premium), and TAM (large TAM = higher multiple). Build a regression model if you have 10+ comps.
DCF for SaaS: Project revenue using cohort-based model: (# New Customers × ACV) + Expansion − Churn. Assume NRR converges to steady state. Terminal value using perpetuity growth (assume 3-5% steady state growth) or exit multiple.
Key Valuation Drivers: Growth rate (most important), NRR (expansion story), gross margin (profitability path), LTV/CAC (sustainability), TAM (ceiling), competitive moat (durability).
SaaS multiples are highly sensitive to interest rates. In rising rate environments, high-growth but unprofitable SaaS compresses dramatically (growth 80% + −20% margin valued at ~10-15x ARR instead of 40x). Conversely, in low-rate environments, growth is privileged and margin compression is tolerated.
Consumption-Based Pricing: Pay for compute, storage, bandwidth, API calls. Revenue scales with customer usage, not seats. High variance — a few large customers can drive disproportionate revenue.
Capital Dynamics: Massive capex for data centers, networking, hardware. AWS, Azure, Google Cloud dominate. Barriers to entry are enormous (capex, regulatory compliance, talent). Margins improve dramatically at scale.
Margin Structure: Gross margins 60-65% (lower than pure SaaS due to infrastructure costs). Operating leverage improves at scale — fixed capex spread across growing revenue base. AWS estimated at 30%+ operating margins at scale.
Vendor Lock-In: Proprietary APIs, data gravity (exporting data is expensive and slow), trained engineering teams. Switching costs are extremely high. A $10M/year AWS customer faces $2M+ switching costs in engineering time and downtime.
Key Metrics: Consumption growth (sequentially), NRR often >130% (heavy expansion), customer count by spend tier (concentration risk), remaining performance obligations (RPO) as leading revenue indicator.
Winner-Take-Most Dynamics: Network effects through developer ecosystem, marketplace of integrations, community knowledge. The three major providers (AWS, Azure, GCP) collectively hold 70%+ market share. New entrants face impossible uphill battle.
Definition: Financial services (payments, lending, insurance, banking) embedded directly into non-financial platforms. Examples: Shopify Payments, Uber's driver payouts, DoorDash Dasher Card.
BaaS Model: Regulated bank provides the license and compliance infrastructure. Fintech/platform provides the user interface and product. Revenue split: bank takes interest margin/fees, platform takes transaction fee (0.5-2%) or monthly subscription.
Revenue Streams: Interchange fees (~1-3% on card transactions), lending spread (borrow at 3%, lend at 8%), subscription fees for premium financial features, float income (holding customer deposits).
Key Risks: Regulatory scrutiny intensifying (OCC, FDIC, state regulators), partner bank concentration (losing partner bank = losing business), fraud and credit losses on lending products, compliance costs scaling faster than revenue.
Valuation Consideration: Embedded finance often valued as a revenue multiplier on the core business rather than standalone. Example: if core SaaS is worth 15x ARR, embedded finance revenue (5-10% of total) might get only 5x multiple due to different risk profile.
Diligence Focus: Partner bank relationships (permanence), regulatory environment trends, credit loss rates on lending, and whether the embedded financial product increases core product stickiness or is easily commoditized.
Revenue Mechanics: Monthly/annual subscription. Freemium conversion rates typically 2-5% (Spotify ~3-4%). Premium ARPU vs blended ARPU (including free tier). Annual customers churn at 1/5th the rate of monthly (stickier due to sunk cost).
Content Costs: Streaming services face massive content acquisition/licensing costs. Netflix spends $17B/year on content. Content amortization schedules (spreading costs across years) critically affect reported margins and GAAP accounting.
Churn Dynamics: Voluntary churn (user cancels) vs involuntary churn (card fails). Involuntary churn can be 30-50% of total churn — retention efforts should focus here. Monthly plans churn 2-3x faster than annual.
Key Metrics: Subscriber count, ARPU, churn rate (by cohort), subscriber acquisition cost (often through discounts or free trial), content cost per subscriber (should trend down as subscriber base grows), engagement minutes.
Competitive Moats: Exclusive content (Netflix's strategy — costly and hard to replicate), catalog depth, recommendation algorithms (personalization), bundling (Apple One, Amazon Prime ecosystem).
High subscriber growth can mask deteriorating unit economics if content costs rise faster than ARPU. A company growing 50% but with content costs growing 75% is heading for margin compression. Watch cost per subscriber trend closely.
Programmatic Ecosystem: DSPs (demand-side, buyers: The Trade Desk), SSPs (supply-side, publishers: Magnite), ad exchanges (connect buyers/sellers), data management platforms (DMPs, manage audience data).
Pricing Models: CPM (cost per 1000 impressions, standard for brand), CPC (cost per click, performance-focused), CPA (cost per action/conversion, most efficient but hardest to scale). Average CPM ranges $2-15 depending on vertical and audience quality.
Revenue Model: Platforms take percentage of ad spend passing through. DSPs/SSPs typically 10-20% take rate. Higher for niche platforms with valuable audiences. Lower for commodified inventory.
Attribution Challenge: Multi-touch attribution (giving credit to all touchpoints), last-click attribution (credits only last touchpoint, often inflates performance), incrementality testing (did the ad actually drive the purchase?). Privacy changes (iOS ATT, cookie deprecation) are disrupting attribution.
Key Risks: Dependence on third-party cookies (dying, regulatory pressure), platform concentration (Google/Meta duopoly control 60%+ of US digital ad spend), brand safety concerns (ads next to objectionable content), ad fraud (fake impressions, bot traffic).
Margin Structure: MarTech (software-like, 70%+ gross margin) vs AdTech (lower margin due to media pass-through, 40-60% gross margin). AdTech is more leveraged on top-line growth for margin expansion.
Definition: SaaS purpose-built for a specific industry (healthcare, construction, restaurants, legal, dental). Examples: Toast (restaurants), Procore (construction), Veeva (life sciences), ServiceTitan (home services).
Advantages Over Horizontal SaaS: Deeper workflow integration (product is mission-critical), higher NRR (not just retention but expansion through complementary features), lower churn (customer has no alternative), ability to embed payments and financial services.
Embedded Payments Strategy: Many vertical SaaS companies process payments within their platform, adding 0.5-2% take rate on transaction volume as incremental revenue. Toast's payment product is 30%+ of revenue despite being "side business." This transforms revenue model from pure subscription to subscription + transaction.
TAM Considerations: Individual vertical TAMs are smaller (~$10-50B for niche verticals) but often less competitive. "Big fish, small pond" strategy. Example: Toast's total addressable market is ~$50B (restaurant industry), but they dominate that segment.
Valuation Premium: Vertical SaaS with embedded payments trades at premium to horizontal SaaS (15-20x ARR vs 10-15x for horizontal) due to dual revenue streams and higher switching costs. Harder to benchmark against pure SaaS comps.
The Full Chain Explained:
Critical Insight: GMV is a growth and health indicator (showing transaction volume), but NOT equivalent to revenue. A $10B GMV marketplace is NOT comparable to a $10B revenue SaaS company. Always translate GMV to revenue first via take rate.
Using GMV Multiples Indiscriminately: a $10B GMV marketplace at 0.5x GMV sounds cheap ($5B valuation). But if take rate is 2%, revenue is only $200M. Implied EV/Revenue is 25x—actually expensive. Always convert GMV to revenue first.
Overstating Cohort Performance: early cohorts often include enthusiastic power users (selection bias). Projecting their LTV/CAC onto later cohorts overstates true lifetime value. Demand by-cohort data; don't average them.
Gross vs Net Confusion: some companies report gross bookings as "revenue." ASC 606 requires net revenue recognition (excluding pass-through amounts). Verify what the income statement shows; demand a revenue bridge from gross bookings to net revenue.
Ignoring Unit Economics: chasing revenue growth while LTV/CAC is below 1x is capital destruction masquerading as growth. A company with $500M revenue and 0.5x LTV/CAC is losing $1 on every $2 of revenue. Eventually, this hits the wall.
Direct Network Effects: Each new user increases value for all existing users. Social networks (Slack, WhatsApp, Discord), messaging apps. Measured by DAU/MAU ratio. Example: WhatsApp — each new user makes it more valuable for everyone. First-to-scale has massive advantage.
Cross-Side Network Effects: More users on one side attract more on the other. Marketplaces (Uber, Airbnb, DoorDash). More sellers → better selection → more buyers → more sellers (virtuous cycle). Measured by liquidity metrics (time-to-match, availability, selection). Cross-side effects are weaker than direct (buyers don't care how many other buyers exist) but often sufficient for defensibility.
Data/Tech-Driven Network Effects: More usage improves product via algorithms or shared data. Navigation apps (Waze — more drivers = better routing), recommendation engines (TikTok — more data = better recommendations), search (Google — more queries = better index). Each user's data makes the product better for everyone, but not directly — through the product's improvement.
Why Network Effects Matter for Retention: In non-contractual settings (marketplaces, consumer apps), network effects ARE the retention mechanism. Without them, users can leave freely. The strength of network effects directly correlates to churn rates. Platforms with strong direct network effects (Discord, Slack) have extremely low churn (1-2% monthly). Platforms without network effects (commodity marketplaces) have high churn (5-10% monthly).
Measuring Network Effects Strength: Cohort analysis comparing Day 1 retention to Day 365 retention. Strong network effects show steep drop initially (0-30 days) then flattening (30+ days as the user has invested in the network). Weak network effects show consistent decline with no flattening.
Contractual vs. Non-Contractual: SaaS (contract signed, customer must actively cancel) vs. marketplace/consumer (no contract, user can leave instantly). Contractual retention is easier to forecast; non-contractual requires constant engagement.
Churn Ladder Concept: Users pass through stages: Active → Inactive → Dormant → Deleted. Churn risk decreases as users pass engagement milestones. Week 1 churn is highest (70-80% of signups churn in week 1 for many consumer apps). Users who survive month 3 are much stickier (20-30% monthly churn vs 70%+ week 1 churn).
Key Metrics: 1-day, 7-day, 30-day, 90-day retention (% of Day 0 users still active at each milestone). DAU/MAU ratio (average engagement). Retention curves should show steep initial drop then flattening — not consistent decline.
Multi-Homing Risk: Users participate in multiple competing platforms simultaneously (Uber/Lyft drivers, TikTok/Instagram creators). Erodes network effect moat. Increases switching likelihood. CAC is wasted if customer multi-homes because they have reduced stickiness.
Switching Costs in Non-Contractual Models: User profiles (takes time to rebuild), social graphs (friends list is valuable), accumulated data (reputation, seller ratings), loyalty rewards (sunk spend). These are weaker than contractual switching costs but still meaningful.
The Fundamental Challenge: Attracting sellers without buyers is impossible (why list if no one will see it?). Attracting buyers without sellers is equally impossible (why browse empty inventory?). Solving this is the critical first mile of any marketplace.
Bootstrapping Strategies:
Liquidity Metrics: Time-to-fill an order (how long from buyer posting request to seller responding), percentage of listings that sell (conversion rate), match rate (% of demand matched with supply), inventory depth (how many options does buyer see).
Buyer Retention: Driven by selection, price, convenience. Measured by repeat purchase rate (3-month, 6-month, lifetime repeat). Expected repeat frequency varies: ride-sharing (3-5x/month), food delivery (1-2x/week), e-commerce (quarterly).
Seller Retention: Driven by sales volume (am I making money?) and platform costs (take rate). Measured by active listing rate (% of sellers still listing after 6 months), seller churn by GMV tier (small sellers churn more).
Supply-Demand Balance: Too much supply → seller churn (no sales, abandon). Too much demand → buyer churn (out of stock, long wait times, poor experience). Healthy marketplace has slight supply constraint (healthy price floor) with high match rates.
Take Rate Structures:
Take Rate Ranges by Vertical: Product e-commerce 5-15%, service marketplaces 15-30%, financial services 1-3% (high volume, thin margin), luxury goods 10-20%.
GMV Is NOT Revenue (Critical): Revenue = GMV × take rate. A $100M GMV marketplace at 5% take rate generates $5M revenue, not $100M. This distinction is load-bearing for valuation — many investors mistake GMV for revenue.
CAC Dynamics — Two Cohorts: Separate CAC for buyers and sellers. Supply-side CAC often lower (sellers are eager for new distribution channels). Demand-side CAC can be high (need to acquire many buyers to create demand). Total CAC = blend of both cohorts.
Disintermediation Risk: Users bypass platform after initial connection. High risk in services (freelancer and client connect then transact off-platform). Platforms combat via escrow (hold payment until delivery), insurance, dispute resolution, mandatory use of platform payment system.
Horizontal vs. Vertical: Horizontal (Craigslist, eBay, Amazon Marketplace) serve many categories, broad TAM, shallow moat. Easy to enter but hard to dominate any segment. Vertical (StockX for sneakers, Airbnb for hospitality, Veepee for flash sales) serve one category, narrow TAM, deep moat, higher take rates. Vertical marketplaces trade at premium multiples due to defensibility.
Trust Mechanisms — Critical Infrastructure: Ratings/reviews (visible seller quality), verification (identity, credentials), escrow (hold payment), insurance (refund guarantees), fraud detection (detect bad actors). Trust is a direct driver of retention. Without it, liquidity collapses. Trust-building is expensive but mandatory.
Diligence Checklist:
Mature Marketplace Signals: Rising organic acquisition (users finding platform without ads, true network effect), improving cohort economics (newer cohorts performing better than earlier ones due to platform improvements), take rate stability (established brand can defend pricing), margin expansion (fixed costs spread across growing volume).
DAU/MAU as Core Metric: Higher ratio = more habitual use. Benchmarks: messaging >60%, social 30-50%, content/media 20-30%, e-commerce <15%. Stickiness is the #1 predictor of ad revenue (more engagement = more ad impressions).
Onboarding Conversion Funnel: Signups → install → first action → repeat use → daily user. Typical conversion: 100% signups → 50% install → 30% first action → 10% repeat → 3% daily users. Many consumer platforms lose 70-80% of new signups within the first week. Improving onboarding is highest ROI investment.
Engagement Loops (The Core Engine):
Retention Curves: Shape is critical. Healthy curves show steep initial drop (0-7 days, half of signups churn), then flattening after day 30 (retained core is much stickier). The "retained core" size (% of signups still active at day 365) multiplied by ARPU = sustainable revenue base.
Cohort Decay Metrics: Measure each signup cohort's retention over time. Track if each new cohort is stickier (retention curve improving) or leakier (retention curve declining). Declining retention = product-market fit is eroding.
Advertising (CPM-Based): Revenue = Impressions × CPM / 1000. Key metrics: ad ARPU (monthly ad revenue per user), fill rate (% of available inventory that's sold), ad load (% of feed that's ads). Higher ad load = more revenue but worse user experience and potential churn.
In-App Purchases (IAP): Virtual goods (cosmetics, skins), power-ups, battle passes, currency. Conversion rate typically 2-5% (free → paying). Revenue heavily skewed: top 10% of payers generate 50%+ of IAP revenue ("whales"). ARPPU is the key metric — wide gap between median paying user (~$20/year) and average paying user (~$100/year indicates whale concentration).
Subscriptions (Premium Tier): YouTube Premium, Spotify Premium, LinkedIn Premium. Recurring and predictable but requires compelling differentiation from free tier. Conversion to premium: 1-3% typical. Premium ARPU: $60-120/year typical. More stable than ad revenue (less volatile with ad market conditions).
Transactions (Take Rate): Platform takes percentage of commerce transacted within app. Instagram Shopping (10-15% from checkout), WeChat Pay (1-2%), mobile games (30% of IAP from Apple/Google). Can add up — even low take rates are valuable at scale.
Blended LTV Calculation: A single user may generate ad revenue (free tier) + have probability of converting to subscriber + probability of making IAP. Total LTV = (months active × monthly ad ARPU) + (P(subscriber) × subscriber LTV) + (P(IAP converter) × IAP LTV). This is why engagement metrics (DAU/MAU) are so critical — they determine ad inventory, which drives ad revenue.
Monetization Curve Over Time: Many platforms launch free only, then layer monetization. Mature platforms: 80-90% free users, 8-12% premium subscribers, 2-5% whale IAP users. The mix shifts based on user base composition and platform focus.
Definition: Single app aggregating many services — messaging, payments, ride-hailing, food delivery, e-commerce, financial services. Examples: WeChat (China, 1+ trillion message volume/day), Grab (Southeast Asia, ride + food + payments), Gojek (Indonesia, 5+ services).
Economics of Cross-Sell: Each user's ARPU increases as they use more services. Ride-hailing user base → cross-sell payments → cross-sell food delivery → cross-sell fintech. Each additional service increases switching cost (data lock-in across services) and daily engagement (more reasons to open app). WeChat average user is active 4+ hours/day vs. single-service apps at 30-60 minutes/day.
Operational Complexity: Each service has its own unit economics, competitive dynamics, regulatory environment. Segment reporting is crucial to understand which services are profitable. Example: ride-hailing may be unprofitable (thin margins, regulatory costs), but payments + fintech are highly profitable and subsidize growth of ride-hailing.
Regional Differences: Superapps flourished in Asia (less friction from platform rules, mobile-first economies, regulatory support). Western attempts (Uber: rides + delivery + freight, Meta: social + messaging + commerce) struggled due to iOS/Android app store restrictions and consumer habits favoring specialized apps. The superapp model is fundamentally incompatible with iOS app store economics (30% take rate, app focus).
Diligence Focus: Track revenue by service, contribution margin by segment (not all services are equally profitable), user overlap across services (are users actually cross-using or just siloed?), and whether cross-sell is actually happening or just theoretical benefit claimed by management.
Valuation Complexity: Superapps are valued as sum-of-parts: profitable payments segment (high multiple) + growth ride-hailing segment (low or negative multiple) + emerging fintech segment (speculative). Don't apply single multiple to total revenue — break down by segment first.
Revenue Sharing Models with Specific Splits:
Edge Monetization Concept: Revenue generated at the "edges" (between users/creators and consumers) rather than by the platform centrally. Platform takes a slice. Digital goods have near-zero marginal cost, so revenue-split decisions are about ecosystem incentives, not cost coverage. Platforms want to maximize creator earnings to attract supply, but also need to be profitable.
Creator Incentive Alignment: Better creator splits attract higher-quality creators (Patreon's 5% fee attracts serious creators over YouTube's 45% split). But platform needs profit to fund features. Tension: creative economy platforms are usually cash-negative because they prioritize creator economics over platform profitability.
Platform Stickiness Through Creator Dependence: Creator earnings dependence on platform (e.g., Twitch streamers earning majority of income from subs + ads) creates moat. Streamers' audience and earnings are tied to the platform. Switching costs: lose all subscribers if you leave.
Creator Concentration — Pareto Distribution: Top 1% of creators generate disproportionate revenue (often 30-50% of total). Top 10% generate 80%+ of total. Losing a top creator can crater the platform. YouTube's top 100 creators drive significant traffic; losing PewDiePie would be catastrophic.
Rake Compression — Competitive Pressure: Platforms forced to offer better terms to creators. Apple's 30% cut faces legal challenges (Epic Games lawsuit) and developer backlash. Patreon keeps rates low (5-12%) to attract creators. OnlyFans' 20% take is higher but still competitive in creator space. As competition intensifies, platforms race-to-the-bottom on creator split.
Multi-Platform Presence — Diversification: Smart creators spread across YouTube + TikTok + Twitch + newsletter (Substack) + Patreon. Reduces platform lock-in. If YouTube changes payout structure, creators have alternatives. Platforms know this and it limits pricing power.
Virtual Economies and Edge Transactions: Skins, tips, donations, badges. These edge transactions align creator and platform incentives (both benefit from high spend). Creates engagement loop: creators incentivized to engage audience (more tips), users incentivized to support creators (cultural status from tips), platform takes a cut. More stable revenue stream than ads (less dependent on ad market).
Diligence Checklist:
Definition: Platforms built on real-world connections (Facebook, LinkedIn, Snapchat). Strength: strong initial lock-in (your network is there). Weakness: ceiling on growth once addressable population is saturated and connected.
Saturation & Demographic Shifts: Once everyone is on the platform, growth stalls (Facebook — 3 billion users, 90% of addressable market in developed countries). Younger demographics actively avoid platforms their parents use (teens leaving Facebook, older demographics driving Discord usage decline). Network that was value becomes burden.
Content Fatigue: Social-graph content depends on friends posting novel updates. Over time, posting declines (people sharing less personal information due to privacy concerns, "performance" burden, competition for attention from other platforms). Feed becomes stale, engagement declines, users churn.
Monetization Ceiling: Ads in a personal social feed feel intrusive. There's a limit to ad load before experience degrades and users leave. You can't monetize a dead network. Facebook hit monetization ceiling when engagement plateaued.
Privacy & Regulation Pressure: GDPR, Cambridge Analytica scandal, iOS App Tracking Transparency (kills third-party cookies/identifiers). Targeted advertising (the main revenue source) faces structural headwinds. Can't target as precisely → lower CPM → lower ad revenue.
Competition from Interest-Graph Platforms: TikTok proved algorithmic content discovery can compete with social-graph lock-in. Algorithm-selected videos are often more engaging than friends' real-world updates. Legacy platforms had to add Reels/Shorts to compete, cannibalizing their own ad inventory and engagement.
Social-graph platforms face structural decline as they mature. Demographic churn, content fatigue, privacy regulations, and algorithmic competition create a long-term headwind. Growth comes only from geographic/demographic expansion (emerging markets) or acquisition of younger platforms (Instagram acquisition by Facebook, TikTok's growth despite competitive threats).
Definition: Platforms that created entirely new behaviors/categories (Uber created ride-hailing, Airbnb created home-sharing, DoorDash created food delivery). Initial growth is explosive as they educate market.
Hyperlocal Network Effects: Network effect is city-by-city, not global. Dominance in London doesn't help in Paris. Expansion = series of mini-launches, each requiring critical mass in that city. Scaling is operationally complex (managing 100+ city P&Ls). Cost to acquire market leadership in a new city is high (subsidies, marketing).
Multi-Homing — Core Problem: Drivers/couriers work for multiple platforms simultaneously (driver list is not exclusive). Riders switch based on price/ETA (low switching cost). This destroys pricing power. Platforms can't raise prices because users immediately switch. Lyft's existence creates price ceiling for Uber (can't go too much above Lyft pricing or riders churn).
Thin Margins & Unit Economics: High variable costs per transaction (driver pay = 75-80% of fare). Take rate ~20-25% minus incentives often = near-zero or negative unit margin. Can only scale profitably at enormous scale with brand maturity (Uber eats losses in nascent cities to eventually become profitable in mature cities).
Regulatory Risk — Structural & Ongoing: Gig worker classification (are drivers employees or contractors?), price controls, licensing requirements, safety regulations. These are not one-time risks — they're permanent regulatory threats. Classification as "employees" would fundamentally break unit economics (add 20-30% to driver costs).
Need for Constant Stimulation: Usage is occasion-based (call a ride occasionally), not habitual. Without daily use, DAU/MAU is low. Marketing spend must be perpetual to maintain demand. High customer acquisition cost. Every customer acquired costs heavily in subsidies/discounts.
Valuation Reality: Focus on unit economics at steady state in mature markets. Is a single ride transaction profitable net of all costs? Example: $15 ride, 20% take = $3 gross revenue, minus $2.50 driver subsidy, $0.30 payment processing, $0.05 support cost = −$0.85 per ride. At this unit economics, scaling just digs the hole deeper. Profitability comes only from scale (spreading fixed costs) and moat (reduced subsidy dependency).
Examples: Apple (iPhone + iOS + services), Peloton (bike + content subscription), gaming consoles (hardware + game sales/subscriptions + digital services), IoT devices (thermostats + cloud services).
Lower Blended Margins: Hardware gross margin 30-50% vs software 70-85%. Blended margins are dragged down. Example: Apple: iPhone hardware ~40% margin, services ~70% margin, blended ~50%. Over-reliance on hardware sales = margin compression.
Capital Intensity & Supply Chain Risk: Manufacturing, supply chain, inventory risk. Very different from pure software. Peloton's supply chain disaster during COVID disrupted hardware production while demand was high (and they couldn't capitalize). Hardware cycles are measured in years (iPhone refresh every 12-24 months); software can iterate continuously.
Product Cycles & Obsolescence: Hardware becomes obsolete quickly (Nokia missing smartphones). R&D must span both hardware and software generations simultaneously. Example: Apple invests in iPhone hardware, iOS software, iCloud backend, and watch ecosystem — all concurrently. Risk of hardware misstep is existential.
Ecosystem Lock-In Strategy (Apple Model): iPhone → AirPods → Apple Watch → MacBook → iCloud → Apple TV. Each additional product increases switching cost. Ecosystem is extremely powerful but expensive to build. Requires excellence in hardware AND software, simultaneously, across categories.
Attach Rates & Subscription Layers: What % of hardware buyers also subscribe to services? Peloton's crisis: hardware sales slowed AND subscriber churn rose simultaneously (people stopped using the bike, canceled subscription). This revealed the subscription layer was dependent on hardware usage, not lock-in. At maturity, Peloton needed ~60% of bike owners as active subscribers to achieve profitability; they achieved only 20%.
Valuation Methodology: Hardware companies valued at 1-3x revenue. Add a subscription layer and the software portion may get a higher multiple (5-10x). Key focus: mix of hardware vs services revenue and trend over time. If hardware is shrinking and services growing = improving business. If hardware stalling and services not growing fast enough = compression.
Inflated Take Rates: Stated take rate includes one-time fees, pass-through revenues (seller pays marketing costs and platform calls it revenue), or promotional credits. Always calculate: effective take rate = actual platform revenue / actual completed GMV (excluding canceled/refunded orders).
Artificial GMV Boosting: GMV includes canceled/refunded orders (should only count completed transactions), intercompany transactions (platform buying from itself), or double-counting (counting full payment amount when only a percentage goes to platform). Red flag: GMV growing much faster than revenue. If GMV up 50% but revenue up 10%, take rate is compressing due to either (a) incentive programs or (b) high refund rates.
Gross vs Net Revenue Confusion: Some companies report gross bookings as revenue. Ticketing platform counting full ticket price as revenue when 70% goes to event organizer. Only revenue that stays with the platform (and isn't passed through) should be counted. GAAP requires net revenue reporting, but companies sometimes bury this in footnotes.
Diligence Steps:
If GMV is growing much faster than revenue, the take rate is compressing. This is usually hidden or downplayed in earnings calls. Compressing take rate is a red flag because it indicates deteriorating unit economics (platform is giving away more to sellers to compete or incentivize volume).
Ignoring True Costs: Only counting direct ad spend as CAC while excluding referral incentives ($50 signup bonus), first-purchase discounts (40% off), sales team costs, onboarding support, customer success allocated. Real CAC is often 2-3x the stated figure. Example: stated CAC $100 = direct ad spend only. Real CAC $300 = includes $80 signup bonus, $60 discount, $40 support costs, $20 in payment processing.
Unrealistic LTV Assumptions: Projecting 5+ year customer lifetimes with zero churn. Or assuming customers stay at premium pricing forever (no competitive pressure leading to discounts). If the payback calculation requires zero churn beyond the observed period, that's a red flag — unobserved future churn is speculation.
Blending Cohorts to Mask Issues: Averaging early cohorts (loyal power users with 30% month-1 retention) with newer cohorts (weaker retention, 5% month-1 retention) to show healthy-looking LTV/CAC. Always demand cohort-level data by signup period. A blended 20% month-1 retention could hide one cohort at 40% and another at 5%.
One-Time Viral Bumps Misattributed: A period of low CAC from organic/viral growth (a viral moment, press coverage, celebrity mention) gets extrapolated as the permanent cost structure. Once the viral moment passes, CAC normalizes upward. Management acts surprised when CAC rises 3x post-viral bump.
Channel-Blind Reporting: Blending CAC across organic, referral, paid search, paid social, partnerships. Organic CAC is near-zero; paid is $200. Blended average looks good ($80) but hides that paid channels are money-losing at current pricing. Demand CAC by channel.
Verification: Compare stated LTV/CAC to observed payback. If LTV/CAC is 3.0x, payback should be 3-4 months. If payback is 12 months, LTV/CAC is misstated. Do the math independently; don't trust management's unit economics.
Selective Cohort Reporting: Company highlights best-performing cohorts (early adopter segments, high-LTV geographies) but omits weaker ones (newer geographies, younger segments with lower ARPU). Demand full cohort detail across all acquisition channels, geographies, and customer segments before accepting unit economics claims.
Over-Adjusted EBITDA: Adding back stock-based compensation, "one-time" restructuring charges (that recur every year), litigation settlements, impairments, and reclassifying operating expenses as non-recurring. A sound business should not need to obscure its core metrics. Example red flag: company reports -$50M GAAP operating loss but +$20M "adjusted EBITDA" by adding back $70M in adjustments. What are these adjustments hiding?
Gross Margin Gimmicks: Classifying customer support or cloud hosting costs as R&D or S&M instead of cost of revenue to inflate gross margin. True gross margin should reflect the actual cost of delivering the product. If support costs 15% of revenue, they belong in COGS, not buried in OpEx.
Revenue Recognition Games: Recognizing revenue early (upfront for multi-year deals instead of ratably over time), including non-recurring revenue in recurring revenue metrics, or counting potential future transactions as booked revenue. ASC 606 requires conservative recognition, but companies push boundaries.
Customer Concentration Disclosure: Burying concentration risk in footnotes. If top 3 customers are 25% of revenue, that's material risk. Demand customer concentration metrics (top 10 as % of revenue) and customer churn rates by size tier.
A sound business should not need to obscure its core metrics. Reasonable take rates, genuine transaction volume, transparent revenue accounting, and realistic customer acquisition math — these form the foundation of a credible growth story. If management is heavily adjusted metrics, digging deeper often uncovers deteriorating fundamentals.
Marketplace Types: P2P (eBay, Airbnb — individual-to-individual), B2C (Amazon Marketplace — business-to-consumer), B2B (Alibaba, Faire — business-to-business). Services vs Goods. Horizontal (multiple categories) vs Vertical (single category).
LTV Formula Derivations:
Payback Period Inversion: If LTV = $5000 and CAC = $1000, LTV/CAC = 5.0x. Payback = CAC / (monthly contribution) = $1000 / ($200/month margin) = 5 months. Payback inversely relates to LTV/CAC: higher LTV/CAC = faster payback.
Strategic Exit Patterns:
You've now traversed the full TMT landscape: from subscription SaaS fundamentals and cloud infrastructure, through marketplace network dynamics and embedded finance, to consumer platforms and creator economies. You understand the metrics that matter, the models that work, the weaknesses embedded in each, and the red flags that separate genuine opportunity from value destruction.
The TMT space moves fast. Regulations shift, user preferences evolve, competitive dynamics flip. But the fundamentals remain constant: network effects, unit economics, retention mechanics, and the eternal tension between growth and profitability. These principles apply whether evaluating Stripe, Shopify, or the next generation of platforms.
Remember: Understand the model deeply. Stress-test the metrics rigorously. Question every narrative. Verify the math independently. The best TMT investments are built on defensible economics and honest reporting — not narrative momentum or founder charisma.