Platform Dependence Risk Survey

Key takeaways

• Platform risk is the danger that one algorithm, policy, or fee change guts your revenue overnight.
• The exposure metric is simple: what share of revenue and customer access runs through channels you do not control.
• The defense is an owned audience (email, phone, your own store) that no platform can take away.
• Diversification without ownership is still renting: two landlords are not a freehold.

The tenant problem, stated precisely

A business that reaches its customers through a platform (a marketplace, a social feed, an app store, a search engine, a lead-gen site) is a tenant on someone else's land. The platform owns the address: the search ranking, the feed placement, the buy button, the customer data. The tenant owns inventory and hope.

Platform risk is what follows from that structure. Fees can rise, and across the major marketplaces and app stores the long arc of take-rates has pointed one direction. Algorithms change without notice or appeal: a ranking update, a feed rebalance, a "helpful content" pass, and each one reallocates fortunes among tenants overnight, and the platform owes no one an explanation. Policies shift and enforcement automates: accounts suspended by classifier, appeals answered in weeks, businesses guilty until reviewed. And platforms compete with their tenants: the marketplace's private label arrives next to your best seller with better placement; the platform that was your channel becomes your rival with your sales data in hand.

None of this requires malice, which is what makes it structural rather than anecdotal. A platform serving millions optimizes for itself; every adjustment is a rounding error at their scale and possibly existential at yours. The asymmetry is the risk.

Measuring your actual exposure

Exposure hides in vibes until it is measured, and the measurement is a one-page exercise. Revenue concentration: what percentage of revenue arrives through each channel you do not control? Sum the platform-dependent share, and flag any single platform above roughly a third, the level where its policy team effectively co-manages your business. Traffic and discovery: where do new customers actually come from? A store whose sales look "direct" but whose discovery is 80% one search engine or one social feed is measuring the wrong hop; dependence lives at discovery, not checkout. Relationship ownership: for what share of customers do you hold a direct, usable contact, meaning an email or phone number you may lawfully use rather than one locked inside a platform's messaging system? Recourse: if your biggest channel suspended you tomorrow, what is the realistic timeline and process to restoration, and could the business pay its bills through it?

Score it quarterly, because drift is the pattern: platform shares creep upward in good times precisely because platforms are good at delivering demand. That is the seduction: rented demand arrives faster than owned demand, right up until the terms change.

One more honest cut: diversification across platforms reduces single-landlord risk but is still renting. Three marketplaces and two ad platforms is a portfolio of leases. The metric that changes the business's risk class is the owned share: revenue and reach that no third party can reprice or revoke.

What the wipeouts have in common

The pattern repeats across every platform generation, and practitioner accounts of it are numerous enough to treat as data. Businesses built entirely on early search-ranking arbitrage vanished in the algorithm updates that cleaned up those tactics. Publishers who reorganized around a social feed's referral traffic, and then around its pivot to video, were reorganized out of existence when the feed's priorities changed. App businesses have been repriced by store-policy changes and review-queue decisions. Marketplace sellers have watched their category's fees rise and their best-selling niche gain a first-party competitor in the same season. Ad-dependent brands have seen acquisition costs double as auctions crowded and targeting rules shifted.

The common thread is never that the platform was evil; it is that the business's core asset, customer access, was never theirs. The wipeout was always technically survivable for a business with reserves and other channels; it was fatal for the ones where the platform was the business.

The inverse cases share a thread too: the sellers and creators who absorbed the same shocks and continued had spent years converting borrowed reach into owned relationships: lists, communities, repeat direct customers, brands worth searching for by name. The shock cost them a bad quarter instead of the company. That difference was built years before it was needed, which is the entire lesson.

The conversion playbook: borrowed reach into owned reach

The strategy is not to abandon platforms, since they are where the demand is, but to run them with an extraction discipline: every platform interaction should have a chance of becoming a relationship you own. Capture: make joining your list genuinely worth it (a real discount, real content, warranty registration, order tracking) and ask everywhere you lawfully can: on-site, post-purchase, in packaging inserts where marketplace rules allow, in bio links, at checkout. Deepen: treat the list as a product, not a megaphone; the businesses whose emails get opened send useful things on a rhythm, and their "we're over here too" messages work when a platform shock arrives because trust already exists.

Redirect: nudge repeat purchases toward your owned store with loyalty pricing, subscriptions, or bundles the platform version cannot match. Margin funds the incentive, since the owned channel skips the take-rate. Fortify the platform presence itself: obsessive policy compliance, diversified SKUs and keywords, reserves sized so a suspension is a crisis rather than an ending.

Then manage by the metric: set a target for owned-channel share of revenue and contactable-customer share, and review both quarterly alongside the platform-risk scorecard. Movement of a few points a year compounds into a different company.

The end state is not platform-free; it is platform-optional, with platforms as acquisition engines feeding an owned core rather than the ground the business stands on. Tenancy is a fine way to start. It is a dangerous way to stay.

Put it to work

Score your exposure quarterly: platform share of revenue, discovery dependence, contactable-customer share, and recourse if your biggest channel vanished tomorrow. Then run the conversion discipline: capture contacts everywhere lawful, make the list genuinely useful, and shift repeat purchases to owned channels. Track owned-share like a KPI; a few points a year changes the company's risk class.

Sources & references

Linked entries open the named source directly. Entries without a link say exactly what kind of reference they are — and how to check them yourself.

• Each platform's own policy & fee pages — Fee schedules, API terms, and seller policies live on each platform's official site and change over time — always read the current version rather than summaries.
• Corlova synthesis of documented policy-change episodes — The pattern described (dependence → policy shift → margin squeeze) recurs across many publicly documented platform changes; the lesson is the structure of the risk, not a claim about any one company.