Skip to main content

Marketplace Design Review

Does a first-time visitor know within 5 seconds what kind of listings are here, how to start comparing, why to trust what they find, and where to go for depth?

When to Use

Use this checklist when building or auditing a site whose primary job is presenting business listings, venture ideas, tools, offers, or content packages for discovery, comparison, acquisition, or commitment.

The test: Cover the nav and logo. Can a visitor in 5 seconds identify (a) what kind of listings are here, (b) how to start comparing, (c) why to trust the listings, and (d) where the deeper know-how or proof lives? If no to any — fail.

Marketplace vs Gateway vs Chooser

Not every marketplace is trying to sell a product immediately. Some are gateways into ideas, know-how, proof, or implementation. Name the surface job before reviewing the page.

Surface typeUser jobPage must exposeFailure mode
MarketplaceBuy, invest, subscribe, acquirePrice, trust, availability, comparison attributes, checkout pathLooks interesting but cannot convert
ChooserCompare options and pick the next actionSort/filter, same mini-IA per item, proof state, disqualifiers, CTA verbBecomes a gallery of identical cards
GatewayRoute to the right depth surfaceDestination types, idea vs know-how vs proof, route labels, escape pathMixes all content jobs into one confusing surface
Portfolio indexDecide what deserves attentionStage, model, score, target, kill signal, next actionRewards browsing instead of commitment
Data catalogFind a specific record or trusted datasetSearch, filters, freshness, source, schema, download/API pathForces manual scanning and unverifiable reuse

Research Basis

This checklist applies three external UX findings to Dream surfaces:

  • NN/g list-entry guidance: a list entry should behave like a small webpage, with attribute priority reflected in the mini-information architecture and visual design.
  • NN/g faceted-search guidance: filters should be appropriate, predictable, jargon-free, and prioritized.
  • Baymard product-list research: product lists, sorting, and filters determine whether users can reduce a large set to a manageable, relevant set.

Interpretation: a venture, offer, or idea card must expose the same decision attributes in the same order. Otherwise the reader or agent cannot compare honestly.


1. Hero Section

CheckThresholdMeasure
Outcome-led headlineHeadline says what the user gains, not what the product isRead aloud — does it say "learn/find/acquire" or "we are a marketplace for..."
Headline length≤10 wordsCount words
One primary CTASingle above-fold action, unique colorSquint test — one element stands out without focusing
Subheading bridges pain to benefitPain → solution → outcome structureDoes it name a real friction before offering the answer?

Diagnosis:

SymptomFix
Headline starts with "We" or "Our"Rewrite: lead with what the visitor walks away with
Two CTAs compete above foldPick one. Demote the second below the fold.
Subheading explains featuresReplace with the problem it solves

2. Listing Card Anatomy

CheckThresholdMeasure
Specific namesReal company, founder, or idea name — no "Major Retailer"Scan 10 cards — any generic title fails
Demand signal on card≥1 quantified signal per card (votes, downloads, revenue, search volume)Count cards without a number
Problem-first hierarchyCard text communicates what it solves, not just what it isCover test: strip visuals — does text alone communicate value?
Card scannable in 3 secondsKey info visible without hover or expandTime yourself reading 5 cards cold
Same mini-IAEvery comparable item uses the same attribute orderCompare five cards — attribute order must not drift
Discriminator above foldThe highest-value decision attribute is visible before body copyCover lower half of card — can you still compare?

Diagnosis:

SymptomFix
Cards look identicalAdd demand signal as the differentiating visual element
Generic listing titlesRequire actual name + one-line problem statement in listing submission
Users hover to understandPromote the hover content to the card face
Cards compare poorlyDefine a mini-IA contract before styling: tension, descriptor, proof, metric, action

3. Trust Signals

CheckThresholdMeasure
Named testimonialsCEO/investor/founder quotes with name and title — no anonymousCount unnamed quotes
Media logosTop-tier publications only, ≤6 logosRemove any logo your target audience would not recognise
Usage metricsConcrete numbers (600K listeners, 200+ listings, $4M raised)No "thousands of users" — find the exact number
Methodology transparency (data tools)Explain how data is sourced and validatedFor AI/data-driven tools — show the method, not just the output

Diagnosis:

SymptomFix
Logo wall without quotesLogos signal category; quotes signal conviction. Add a quote or remove the logos.
Vague social proof ("loved by teams")Replace with a specific number and a specific named person
Data source unexplainedAdd one sentence: "Data sourced from X, validated by Y"

4. Discovery and Navigation

CheckThresholdMeasure
Nav items ≤6Minimal header navigationCount nav links
Multiple discovery pathsAt least: recent, trending/ranked, and search3 entry paths minimum
Return mechanicDaily email or notification opt-in visibleIs there a low-friction way to come back?
Mobile-first CTAEmail signup preferred over app installApp install adds friction on mobile — email captures intent first
Sort/filter controlsRequired when 6+ comparable items existCount comparable cards; controls must appear before the grid
Result countVisible when filtering or sorting existsChange filter; count updates clearly
Reset pathOne action clears filtersFilter to zero results; reset must be visible
Empty stateExplains what the filter meansForce no-match state and read message

Diagnosis:

SymptomFix
List-only discoveryAdd a ranked/trending sort — not all visitors know what they want
Nav with 10+ itemsCollapse into 6 or fewer; route the rest to footer
No return mechanicAdd email capture — even a "new listings" digest is enough
Filters mirror internal jargonRename to user decision attributes, or add explanatory labels

5. Trust, Proof, and Kill Signals

For idea, venture, and agent-facing marketplaces, trust is not only social proof. It is evidence state.

CheckThresholdMeasure
Evidence stateEvery claim says proven, validating, projected, unvalidated, or consumedCount cards without an evidence label
Commercial basisRevenue, price, or ROI includes time basis and avoids false precision"$36k/yr projected" beats "$692.31/wk"
Kill signalEach unproven opportunity names what would stop itScan cards for disqualifier, deadline, or constraint
Proof routeStrong claims link to proof, case study, receipt, or methodologyClick proof cue; route must explain why to believe
Destination typeIdea, know-how, proof, demand, and implementation are visibly distinctFirst-time visitor can say which route they are opening

Diagnosis:

SymptomFix
"Conviction" presented as proofLabel as projected or validating; link to what would prove it
Revenue looks precise but untestedRound the number and add projected / actual / target
All CTAs say "View"Match verb to stage: inspect, pressure-test, see proof, compare, commit
Gateway links all look interchangeableLabel each destination by job: idea, know-how, proof, demand, build

6. Pricing (SaaS and Tool Marketplaces)

CheckThresholdMeasure
Freemium activationFree tier with daily or limited access — removes "I'll try later"Does the page let someone start without a card?
Pricing visibleNot hidden behind an email gateHomepage or one click away
Launch urgencyTime-limited or seat-limited pricing if early stageReal constraint only — no fake countdown
Tier differentiation3 tiers with feature differences readable at a glanceCover test: can you explain each tier in 5 words?

Diagnosis:

SymptomFix
No free tierAdd a limited daily access tier — even 3 free views/day qualifies
Pricing hiddenMove above the fold or add persistent "Pricing" nav link
Tiers have 20 feature rowsReduce to 5 rows; move detail to a comparison page

7. Anti-Patterns

  • Logo carousels without quotes — logos signal category, quotes signal conviction. If you have logos, add a quote from that organisation.
  • Generic listing titles — "Interesting opportunity in logistics" is not a listing. Require actual company name and one-line problem statement.
  • Pricing hidden on content sites, shown on tool sites — match visibility to monetization model. Membership sites can gate pricing; tool sites cannot.
  • List-only discovery — always offer ranked or trending as an alternative sort. New visitors do not know what to search for.
  • "Coming soon" on a premium tier without a waitlist — kills urgency and captures no intent. Add waitlist capture if the tier is not ready.
  • Hover-dependent card content — mobile users cannot hover. Promote key information to the card face.
  • Decorative taxonomy colour — colour must encode model, stage, risk, or action; arbitrary gradients are visual noise.
  • Gateway without depth route — if a card raises a method question, the page must route to know-how instead of trapping the reader in more cards.
  • No disqualifier — an opportunity without a kill signal is a pitch, not a decision surface.

Component Decision Checklist

Record these decisions in the implementation PR, page comment, or design receipt before closing.

ComponentDecision to record
HeroWhat decision does this page help the visitor make?
Gateway doorsWhich destination types are intentionally separated?
Control barWhich attributes are sortable/filterable, and why those attributes?
Listing cardWhat is the mini-IA order, and which attribute is the discriminator?
Trust/proof cueWhat evidence state appears before the click?
Kill signalWhat stops this option from consuming more attention?
Empty stateWhat does a no-match result teach?
Mobile layoutWhat remains visible without hover or wide-screen comparison?
Agent-legibilityWhich data-*, schema, or route labels make the page machine-readable?

Verification Loop

Run this after implementation:

  1. 5-second gateway test: identify listing type, comparison start, trust basis, and depth route.
  2. Mini-IA scan: compare five cards; the same attributes appear in the same order.
  3. Filter test: apply one model/stage/type filter, confirm count, reset, and empty state.
  4. Mobile test: card content remains understandable at 375px without hover.
  5. Proof test: every strong claim links to evidence or labels itself as projected/unvalidated.
  6. Agent test: inspect data-*, JSON-LD, route labels, or semantic headings for machine-readable decision attributes.

Context