Marketplace Provider Concentration: Test Lost Capacity

A marketplace can list twenty specialists while relying on one person for most of the work customers actually buy. Recruiting another twenty profiles does not necessarily repair that dependency: they may lack the right skills, available times or customer approval.

Quick answer: measure each provider’s share of completed transaction value, then remove the busiest provider from a fixed set of upcoming bookings. Check whether qualified alternatives can accept each job within its original promise, without assigning the same spare capacity twice. Price the recovery and unresolved commitments before deciding to buy more demand.

Verdict: conditional. Concentrated supply can be a workable starting point when the founder understands the exposure and limits commitments accordingly. It is weak when growth assumes that every listed provider is interchangeable or that customers will accept an undisclosed substitution.

Define the business before measuring dependence

For a constructed founder scenario, imagine a marketplace connecting small-business presenters with independent specialists for a remote slide-formatting service. Customers supply approved slide content; a provider applies agreed formatting and returns a reviewed file. The pilot excludes copywriting, research, regulated advice and access to unrelated company files.

The buyer pays for a bounded result by a deadline. The provider accepts the scope, compensation, working window and revision allowance. The marketplace earns a commission for arranging and supporting the transaction. Repeat demand, prices and costs below are unproven planning assumptions—not findings about this market.

Measure past concentration and future exposure separately

Andreessen Horowitz’s marketplace metrics framework describes concentration through the share of GMV attributable to top sellers or buyers and highlights exposure when a large participant leaves. That is a useful starting signal, not a universal safe percentage.

Define GMV consistently before calculating it. In this example, use the service value of completed, accepted jobs in one stated period, excluding taxes, tips and pass-through amounts. State how refunds and reversals are treated; do not mix gross bookings in one row with net completed value in another.

Top-provider share = that provider’s completed service value ÷ all providers’ completed service value in the same period. If one specialist delivered $4,000 of a constructed $10,000 total, the share is 40%. Also show job count: one expensive project and many small projects create different replacement work.

Then build a second view for commitments not yet delivered. Which accepted bookings depend on that specialist, on what dates, and for what required skills? A provider’s historical 40% share does not mean 40% of next week’s requests will fail. Likewise, low historical share can hide sole coverage for a new service or deadline window.

Use service, skill, language, timing and customer preference to define meaningful segments. A blended total can conceal a provider who is the only eligible person for a particular request type. Several provider names may also share one subcontractor or the same unavailable working hours.

Replay the bookings with one provider unavailable

Start with a simulation using test records or a copied planning ledger. Do not remove someone from actual work or change a customer’s confirmed booking merely to test a spreadsheet. Fix the affected window and preserve the original deadline, scope and customer choice.

Sharetribe’s liquidity explanation distinguishes participants achieving a transaction from merely joining a marketplace. Apply that outcome focus here: an alternate profile is not a recovered booking.

Give every proposed replacement a real capacity constraint

1. List the unavailable provider’s accepted jobs in the affected window, with duration, preparation, revision allowance and deadline.
2. Identify alternatives qualified for that exact scope. Confirm availability after their existing commitments and protected buffers.
3. Allocate a specific usable slot to each candidate recovery. Reduce remaining capacity immediately; never reuse one spare slot across several rows.
4. Check whether the customer bought an interchangeable service or a named specialist. Record whether substitution requires consent and whether that consent has actually been obtained.
5. Classify each row as simulated feasible, pending acceptance, accepted recovery, or uncovered. A plan remains a plan until the necessary participants accept.

The startup dispatch capacity template helps estimate usable time. This replay asks a different question: after removing a particular provider, which original promises still fit the remaining qualified capacity?

Show why three profiles do not equal three backups

Illustrative replay: the busiest specialist has five upcoming jobs, each requiring one protected two-hour block. Provider B has two suitable spare blocks; provider C has one. Provider D is listed but unavailable during the required window.

At most three of five jobs have a proposed slot. If J3’s buyer declines the alternative, only two remain potentially recoverable under these assumptions. Rescheduling J4 into next week changes the promise; record an accepted revised date separately rather than counting it as original-deadline recovery.

Price the recovery without confusing GMV with earnings

A recovered $200 booking does not add $200 to the platform’s contribution. Use your actual fee basis and provider agreement. The marketplace take-rate worksheet supplies the baseline pricing model; add only the incremental recovery costs here.

Constructed economics: each completed job has a $200 service value and a 20% commission. Provider proceeds are $160; the platform retains $40. Normal variable operating costs, including a consistent acquisition allocation, total $25 per completed job. Normal platform contribution is therefore $15 before fixed overhead, taxes and other excluded costs. The provider’s $160 is proceeds, not personal profit.

Assume extra recovery coordination costs $8 per recovered job, making its platform contribution $7. If all three proposed replacements are accepted and delivered, their modeled contribution is 3 × $7 = $21. Add a hypothetical $12 handling cost for each of the two uncovered jobs: the modeled affected-booking remainder becomes $21 − $24 = −$3.

This deliberately simplified case assumes uncovered jobs generate no retained customer charge and no provider compensation obligation. It excludes incremental nonrecoverable processing costs or acquisition costs on those failed jobs; if they exist, deduct them too. Any promised provider compensation, refund expense, paid backup reservation or additional revision work also changes the result. Return of a customer’s principal is a cash movement, not automatically another expense to deduct from commission never earned; have the actual treatment reviewed.

The original five successful jobs would have contributed $75 under the normal assumptions. The adverse scenario is therefore $78 worse before other obligations—not merely a reduction in GMV. It is a sensitivity test, not a probability-weighted forecast or an accounting statement.

Do not buy permanent backup capacity from this arithmetic alone. Compare its full commitment cost, likely use, provider willingness and the contribution it actually protects. A provider who requires paid reserved time has made a different offer from one who may freely decline an occasional request.

Copy the booking-recovery card

Keep one row per original commitment, linked to its replacement proposal. Use a reviewed ledger for calculations; no automated optimizer is required.

• Original promise: booking ID, customer, service version, named-provider requirement, inputs, price, delivery window and accepted terms.
• Exposure: unavailable provider, affected interval, reason known or unknown, original capacity block and required qualifications.
• Alternate evidence: qualified provider, exact spare block, existing workload, source/date of availability and compensation acceptance.
• Customer decision: whether consent is required, proposal sent, acceptance evidence, refusal or agreed revised deadline.
• Economics: fee basis, normal costs, incremental coordination, backup commitment, outstanding obligations and reviewed outcome.
• Closeout: delivery evidence, customer review, missed original deadline, exception owner and unresolved compensation or refund work.

Read the card as a chain of gates: eligibility before proposal, participant acceptance before commitment, and completion evidence before recovery is marked delivered. Missing inputs return to the customer; a declined provider proposal returns to operations; a disputed substitution or funding gap goes to the named commercial owner.

Test continuity before expanding demand

Recruit alternatives with relevant work samples and a small, properly compensated trial before relying on them. Ask whether they would accept the actual scope, price and response window—not whether they would like to join a network someday. Confirm that serving your request does not conflict with existing commitments.

• Hypothesis: suitable customers will accept a qualified alternative, and enough willing capacity exists to preserve the bounded service promise at affordable cost.
• Test: first replay a fixed planning window. Then offer a small, disclosed pilot to customers who knowingly accept the proposed provider; do not disrupt existing bookings to manufacture an outage.
• Signal: accepted alternatives, confirmed usable slots, delivered outcomes, original-deadline performance, actual operator time, provider proceeds and unresolved obligations.
• Threshold: prewrite the maximum uncovered commitments and cash exposure your business can responsibly carry. One illustrative pilot gate is three voluntary replacement bookings delivered by their agreed deadlines, nonnegative combined contribution after measured incremental costs, and no unresolved material quality or compensation issue.
• Decision: continue another capped cohort if every gate passes. Rework a specific skill, window or communication gap. Stop adding commitments if willing qualified supply is unavailable, customers reject substitution, or the business cannot fund the promise.

These are proposed learning gates, not industry benchmarks or statistical proof. A cooperative three-booking pilot does not establish resilience to a widespread outage. Test shared dependencies separately, and record founder intervention rather than presenting personal rescue as a scalable process.

If customers specifically buy one expert, the honest model may be a specialist-led service with limited capacity—not an interchangeable marketplace. Narrow the promise, disclose the dependency and decide whether the resulting business is still attractive. Ordinary launch performance belongs in the marketplace liquidity scorecard; keep this adverse-case replay separate.

Build This Business in Workhint

The Workhint Workflow Engine describes configurable request steps, assignments, role-routed approvals and required details, files or notes before work progresses. The Workhint founders solution describes distinct participant roles and onboarding, with requests moving through review, assignment and approval. Those capabilities fit the acceptance gates in this operating model.

Use the following as a copy-ready configuration brief to review—not a claim that Workhint calculates concentration, optimizes schedules or proves backup availability automatically:

Build a remote slide-formatting service marketplace for business customers, qualified formatting specialists, an operations reviewer and a commercial owner. Capture the approved input deck, permitted access, scope, deadline, chosen-provider requirement, price and accepted terms. Onboard specialists with reviewed samples, skills, availability evidence and compensation acceptance. Customers submit inputs and review delivery; providers review proposed work and submit completion evidence; operations reviews eligibility and replacement proposals.

Use Submitted, Needs Inputs, Scope Review, Provider Acceptance, Ready, In Progress, Completion Review and Closed states. Add a linked Recovery Review case when a provider becomes unavailable. Require a manually checked alternate slot, provider acceptance and any required customer consent before a replacement becomes Ready. Route unavailable capacity, missing inputs, substitution refusal, cancellations, rescheduling and disputed delivery to named owners. Commercial review handles incremental commitments and unfunded promises; it must not silently change original terms.

Keep the reviewed concentration and capacity calculations in the pilot ledger. Record original and revised deadlines, operator effort, delivery evidence and the review decision. Configure customer/provider views and administrative permissions; test every transition with fictional records. Keep collection, refunds and provider payment in the existing reviewed financial process, linking execution evidence where appropriate. Do not assume escrow, automatic settlement, automated recovery scoring or guaranteed service continuity.

FAQ

What provider concentration percentage is safe?

There is no universal safe percentage for this decision. Consider which commitments depend on the provider, whether customers accept alternatives, available qualified capacity, replacement effort and your ability to carry outstanding obligations. Show counts and amounts alongside shares.

Does onboarding more providers reduce the risk?

Only if they add useful, willing capacity for the exposed work. Profiles without relevant skills, compatible windows and accepted commercial terms do not recover a booking. Several providers can also share the same underlying dependency.

Should I move work away from my best provider?

Not automatically. Arbitrary redistribution can damage quality or disregard customer choice and provider agreements. Test qualified alternatives with willing participants, preserve existing commitments and compare the full cost of resilience with a narrower growth promise.

Scale the promise you can still keep

A concentration metric tells you where to look. The booking replay tells you what might actually fail. Expand only when the ordinary service economics work, a realistic adverse case is affordable, and accepted recovery no longer depends on the founder improvising every transition. Otherwise recruit specific missing capacity, narrow the offer or slow new commitments.