How to Make an App Like Redfin in 2026: Features, Tech Stack & Cost

At Make An App Like, we are a US-based app development agency, and over the past three years our team has shipped 26+ production marketplace platforms — including five real-estate clones live in our catalogue: our Zillow Clone, Redfin Clone, Realtor Clone, 99acres Clone, and MagicBricks Clone — plus seven automotive marketplaces (Carvana, CarGurus, Cars24, Vroom, AutoTrader, TrueCar, Edmunds), live commerce (Whatnot, Bambuser), audio streaming (Pocket FM, Kuku FM), ride-hailing (Uber), neobanking (Revolut), AI companion (Candy AI), and quick commerce (Zepto, Mrsool, DeliverIt). Real estate is one of our deepest verticals — we have built MLS feed adapters, Zestimate-style AVM engines, agent-CRM modules, mortgage calculators, and IDX-compliance layers across multiple production builds. In this guide, we walk through exactly how to make an app like Redfin in 2026 — the features that matter, the technology stack that scales to millions of listings, the development process, the MLS and IDX integration layer, the realistic cost across budget tiers, and the monetization layers — based on what we have shipped and what we have seen actually work in the field.

What is an app like Redfin?

Redfin is a US-based residential real-estate platform with a structural difference from every other property portal in the market — it is both a marketplace AND a licensed real-estate brokerage. Where Zillow earns money by selling buyer leads to outside agents through its Premier Agent program, Redfin employs its own real-estate agents on salary, pays them a base plus performance bonus, and refunds part of the buyer-agent commission back to the buyer at closing.

An app like Redfin combines five distinct product layers into a single experience. Understanding each one is essential before you scope an MVP.

Map-first property search. Buyers browse listings on an interactive map rather than a list-first search. Redfin's map UX — pin clusters, draw-to-search polygons, neighborhood overlays, school-zone overlays, walk-score and commute heatmaps — is the product surface most clients ask us to replicate first.

Redfin Estimate (their automated valuation model). Every listing carries a confidence-scored AVM price estimate similar to Zillow's Zestimate. Redfin trains their model on millions of comparable sales and publishes per-property accuracy bands so buyers can see how reliable each estimate is for that specific home and that specific market.

In-house salaried agents. Buyers can tour, negotiate, and close with a Redfin agent who is paid by Redfin rather than by commission per deal. This is the single feature that defines the Redfin model and is the hardest to operationalize because it requires real-estate licensing in every state or country you operate in.

Bundled services. Mortgage (through Bay Equity, a Redfin subsidiary), title insurance (Title Forward, also Redfin-owned), and rentals (Apartment Guide and Rent.com, acquired in 2021) all sit inside the same app — letting Redfin capture revenue across the entire transaction lifecycle.

Buyer commission refund. Redfin returns a portion of the buyer-agent commission directly to the buyer as cash at closing — typically 0.5 to 1.5 percent of the home price. This refund is the trust signal that defines the brand and the marketing hook that converts price-sensitive buyers away from incumbent brokerages.

Most new builds replicate the marketplace and map layer first, then add brokerage operations in a Phase 2 expansion. The brokerage layer is the heavier lift, because it requires real-estate licensing, agent recruitment, and state-by-state compliance overlays that the pure-marketplace path avoids.

Why build an app like Redfin in 2026?

The numbers are large. As per the National Association of Realtors housing statistics, US existing-home sales totaled roughly 4.1 million transactions in 2023, with median sale price around $389,800. At a 5 to 6 percent total commission rate, the US residential brokerage market generates roughly $85 to $100 billion in annual commission flow. Globally, residential real-estate technology platforms manage trillions in transaction value annually, and online property discovery now drives the majority of buyer journeys in every major market.

Three shifts in the last 24 months made 2026 a foundational year for new builders. First, the NAR commission settlement finalized in late 2024 fundamentally reshaped how buyer-agent commissions work in the US — agents now must negotiate compensation directly with their buyers rather than relying on MLS-published cooperative-compensation offers. This change opens space for new brokerage models that build commission transparency into the product from day one. Second, Rocket Companies acquired Redfin in 2025 for approximately $1.75 billion, creating consolidation in the buy-side bundled-services category and signaling that vertically-integrated real-estate-plus-mortgage businesses are the strategic frontier for the next decade. Third, AI-generated listing descriptions and AI-staged photos matured enough in 2025 that smaller brokerages can produce listing-grade content at scale without a full editorial team.

International demand is strong and underserved. Markets without a native Redfin equivalent — the UK (where Rightmove dominates listings but no brokerage hybrid exists), Australia (REA Group's Realestate.com.au is marketplace-only), Canada (Realtor.ca aggregates listings but offers no integrated brokerage), India (99acres, MagicBricks, NoBroker do not bundle agent services), and the GCC (Property Finder and Bayut serve listings but not transactions) — all have room for a Redfin-style hybrid that combines listings with managed agent operations and bundled mortgage.

Builders entering now can claim regional or vertical lanes the global incumbents do not yet serve well. A Redfin-style platform for the UK leasehold market, a regional Redfin-equivalent for tier-2 Indian cities, a commercial-real-estate Redfin for office and industrial leasing, or a build-to-rent (BTR) Redfin for the fast-growing institutional-landlord segment are all live opportunities our team has been asked to scope.

Who needs an app like Redfin?

Six buyer profiles drive most of our discovery calls for Redfin-style platforms in 2026.

• Discount-brokerage founders — launching a flat-fee or commission-refund brokerage in a US metro where established players have left supply gaps. The post-2024 NAR settlement environment specifically rewards founders who build commission transparency into the product from day one.
• International proptech founders — building a Redfin-style hybrid for the UK, Canada, Australia, India, GCC, or Southeast Asia, where the marketplace-plus-brokerage model has not yet been replicated by local incumbents.
• Existing brokerages going digital-first — traditional Keller Williams, Coldwell Banker, or Century 21 franchise affiliates wanting their own branded buyer and seller experience without losing their NAR membership or MLS access.
• Regional MLS consortia — replacing legacy IDX vendors with a modern white-label consumer-facing app powered by their member feeds.
• Commercial real-estate platforms — adapting the Redfin pattern to office, retail, industrial, and multifamily leasing where the marketplace-plus-broker model has not yet matured.
• Build-to-rent operators — institutional landlords managing thousands of single-family rentals who need a buyer-style discovery experience layered on top of their leasing operations.
• Real-estate franchise networks — bundling a unified consumer app across all offices under one brand, with shared MLS feeds, agent CRM, and lead routing.

Core features

The platform has four distinct user roles plus a cross-cutting MLS-and-AVM pipeline that powers every search, every estimate, and every listing page. The role-by-role feature breakdown below is the minimum viable scope for a launch-ready V1.

Buyer features (home shoppers and renters)

The buyer experience is what every new entrant gets judged on. Redfin's buyer app is famously polished, and copying the basics is table-stakes — the differentiation comes from the brokerage layer wrapped around it.

• Map-first property search — interactive map with pin clusters, draw-to-search polygons, neighborhood overlays, school-zone overlays, walk-score and commute heatmaps, and saved-map-view URLs that buyers can share with their agent.
• Smart filtering — price, beds, baths, lot size, square footage, year built, days on market, school district, HOA fees, listing type (for sale, sold, open house, new construction, foreclosure), and amenities (pool, garage, AC, basement).
• AVM (automated valuation model) — Redfin-Estimate-style price estimate with confidence interval (low / median / high), comparable-sales reasoning, and accuracy track record for the local market.
• 3D tours and virtual walkthroughs — Matterport or in-house 3D tours embedded directly in the listing page, plus video tours and photo galleries with high-resolution zoom.
• Mortgage calculator — payment breakdown by principal, interest, taxes, insurance, HOA, and PMI; affordability calculator pulling income, debt, and down-payment scenarios.
• Tour booking — request in-person or video tours directly with a salaried agent, with calendar availability and instant confirmation.
• Saved searches and instant alerts — push, email, and SMS notifications when a matching listing hits the market — typically within minutes of MLS publication.
• Hot Homes prediction — algorithmic flag for listings predicted to sell within 14 days, based on price, location, and historical comparable-sale velocity.
• Offer flow — guided offer-submission flow with offer-price recommendation, financing contingency selection, inspection contingency, and seller-counter-offer handling.
• Bundled mortgage and title — pre-qualification through the in-house lender, title quote, and closing-cost estimate surfaced inside the same app.

Seller features (home sellers and homeowners)

The seller side determines whether the platform has supply to sell. A modern seller experience has to combine instant property valuation with concierge upgrades that compete with the iBuying offer.

• Instant property valuation — homeowner enters address, gets a confidence-scored AVM estimate with comparable sales, price history, and tax-assessment context.
• Sell-with-an-agent flow — connect to a salaried in-house agent for a free consultation, listing-prep walkthrough, photographer scheduling, and CMA (comparative market analysis) review.
• Concierge listing service — optional paid upgrades for staging, paint, repairs, landscaping, deep cleaning, with payment deferred to closing.
• Listing builder — guided flow to upload photos, write descriptions (AI-assisted), set price, schedule open houses, and select premium-listing tiers.
• Showings dashboard — see who toured the home, agent feedback after each showing, offer count, and offer-strength comparison.
• Closing cost estimator — net-proceeds calculator showing commission, title fees, transfer tax, and seller credits.
• Offers dashboard — compare incoming offers side-by-side with financing strength, contingency count, escalation clauses, and closing timeline.
• Direct-buy or iBuy program flow — optional cash-offer flow if you operate an iBuying program, with offer expiration timer and acceptance flow.

Agent features (salaried in-house agents and showing partners)

The agent portal is where the brokerage operation actually runs. Salaried agents need a different toolset than commission-only agents because they care about pipeline velocity and customer satisfaction scores rather than per-deal commission optimization.

• Lead inbox and routing — auto-assigned leads by geography, agent capacity, language match, and customer rating; SLA timers for first-response.
• Tour calendar — multi-property tour scheduling, route optimization, calendar sync with Google or Outlook, and showing-partner handoff for overflow.
• CRM with deal pipeline — every active client tracked through tour, offer, contract, inspection, appraisal, closing, with stage-specific reminders.
• Performance dashboard — closes per quarter, customer satisfaction score, tour-to-offer ratio, offer-to-close ratio, average time-from-tour-to-close.
• Customer feedback collection — automated post-close survey, public reviews, and internal feedback used for performance reviews.
• Showing-partner network — overflow tours assigned to part-time agents for fixed fees, freeing salaried agents for high-intent buyers.
• Document signing — DocuSign or Dropbox Sign integration for offers, counter-offers, and closing documents.
• Comp tools — AVM-backed CMA generation, recent comparable-sales pull, price-recommendation engine, and listing-history visualization.

Admin features (brokerage operator)

The admin console is where the brokerage business runs. Real-estate compliance, agent payroll, MLS feed management, and commission accounting all live here.

• Agent management — onboarding, licensing verification, state-by-state compliance, performance reviews, salary plus bonus schedule, and territory assignment.
• MLS feed manager — schedule imports, map fields per MLS, handle deduplication across overlapping MLS jurisdictions, log feed-health metrics.
• Compliance ledger — immutable record of every offer, counter-offer, fair-housing disclosure, agency-disclosure form, and closing transaction for state-board audit.
• Commission accounting — track buyer-side commission earned, refund issued, seller-side commission earned, and split between agent and brokerage.
• Mortgage and title routing — route eligible buyers and sellers to the in-house lender and title vendor; track conversion rates by agent and metro.
• Concierge service queue — manage staging, paint, and repair vendor scheduling and payments.
• Reporting and analytics — GMV, transaction volume, refund volume, agent utilization, lead conversion, mortgage attach rate, customer NPS by metro.
• Multi-market support — state-by-state regulatory overlay, multi-currency where international, and multi-language UI.

MLS and AVM pipeline (cross-cutting)

The data pipeline is what makes the platform credible. Buyers compare AVM accuracy and listing freshness directly against Zillow and Realtor.com, and any visible lag or error kills trust fast.

• MLS RETS and RESO Web API ingest — pull listings from 600+ regional MLSes in the US plus their international equivalents (CREA DDF in Canada, Rightmove and Zoopla scrape feeds in the UK, REA Group XML feed in Australia).
• IDX compliance layer — each MLS has its own display rules, attribution requirements, opt-out registry, and refresh frequency; the compliance layer enforces all of them.
• AVM training pipeline — train and retrain the valuation model on comparable sales, listing history, neighborhood trends, and property features. Most builds use scikit-learn or XGBoost for the regression model, with Apache Airflow for scheduling weekly retrains.
• Photo CDN and processing — store millions of listing photos on S3 with on-the-fly resizing via Imgix or Cloudinary, plus AI-staging variants.
• Geo engine — PostgreSQL with PostGIS for parcel boundaries, school-zone polygons, commute-time isochrones, and walk-score caching.
• Map tile server — Mapbox or self-hosted vector tiles for sub-100-millisecond pan and zoom even at 1 million-listing scale.

Development process — 11 phases

Our team works through the build in eleven sequential phases. Skipping any of them creates risk that surfaces later as regulatory exposure, MLS de-listing, or expensive replatforming.

1. Define the market segment. US single-family residential, international leasehold, US multifamily, commercial real estate, build-to-rent, or a regional hybrid. This decision drives every other choice.
2. Choose the MLSes and IDX strategy. Each US MLS has its own license agreement, fee structure, IDX rules, and approved vendors. Build a strategy for which MLSes to pursue in V1 (the metros you launch in) and which to add in Phase 2.
3. Decide on brokerage operations or pure marketplace. A pure marketplace (Zillow-style) takes 6 to 9 months. A hybrid with in-house brokerage operations (Redfin-style) takes 9 to 14 months and requires real-estate licensing in every state.
4. Design the agent compensation model. Salaried, hybrid, or commission-only — this drives recruiting, retention, and the commission-refund math that defines the brand. Salaried is the hardest to operate but the strongest differentiator.
5. Design role-based architecture. Buyer, seller, agent, admin, plus the MLS-and-AVM pipeline. Define each role's UI, permission set, and data ownership.
6. Build core platform. Map-first search, AVM, listing pages, tour booking, offer flow, agent CRM, admin console. This is roughly 40 to 50 percent of the engineering budget.
7. Integrate KYC and licensing verification. Persona, Sumsub, or Onfido for buyer ID; state real-estate license API integration for agent verification.
8. Integrate payment and mortgage layer. Stripe Connect for transactional flows, mortgage referral integration or in-house lender (Encompass, Blend, Lower) for V2.
9. Build the AVM training pipeline. Ingest comparable sales, train the regression model, validate accuracy against held-out properties, and ship a confidence interval that buyers can trust.
10. Pilot with one metro. For example, Seattle for a US launch, Bangalore for an Indian launch, Dubai for a GCC launch. Pilot for 90 days, identify gaps, and harden the core.
11. Scale. Add MLSes, add metros, add mortgage and title bundles, add iBuying or concierge programs, add multi-language support.

The full process typically runs 6 to 9 months for a marketplace-only build and 9 to 14 months for a full hybrid with brokerage operations. A white-label fork of our existing Redfin Clone codebase compresses to 4 to 8 weeks for the marketplace layer.

Tech stack

The technology stack below is what our team uses on every real-estate marketplace build. Every component named here is in production on at least one of our 26+ shipped marketplaces. The stack is the same chassis behind our Zillow, Redfin, Realtor, 99acres, and MagicBricks clones, adapted per market.

This stack has been hardened across five production real-estate marketplace builds, so the operational edges (MLS feed retry behavior, photo-CDN cache invalidation, AVM retrain failure modes, KYC retry handling) are well understood. Alternative stacks (Ruby on Rails, Django) work fine for the core platform — we recommend Node and Python only because the AVM training pipeline is dominated by Python machine-learning libraries.

Cost — three tiers

Building an app like Redfin costs more than most marketplaces of similar UI complexity because the MLS integration layer, AVM training pipeline, and brokerage compliance overlay are substantial engineering surfaces on their own. The table below is the realistic 2026 range for a custom build from scratch.

A white-label fork of our existing Redfin Clone codebase compresses the timeline substantially. The marketplace and map layer ships in 4 to 8 weeks for $20,000 to $45,000, with the brokerage operations layer added as a Phase 2 expansion. This is the fastest path to a launch-ready V1 for a founder who wants to validate the model in a specific metro before committing capital to a full custom build.

Factors that drive cost

Seven variables move the project total up or down by tens of thousands of dollars. Getting these right on day one keeps the budget on track; getting them wrong creates drift.

• MLS count and IDX strategy — each additional MLS adds 3 to 6 weeks of engineering for the feed adapter, field mapping, compliance overlay, and licensing fee. A V1 with five MLSes costs noticeably more than a V1 with one.
• Marketplace-only vs full brokerage — the brokerage operation adds 4 to 6 months of engineering plus real-estate licensing fees in every state, agent recruiting, and payroll infrastructure. This is the single biggest cost lever.
• AVM depth — a basic AVM that uses one regression model on comparable sales is cheap. A confidence-banded, locality-aware, photo-feature-aware AVM trained on millions of records adds 2 to 3 months of machine-learning engineering plus ongoing retraining costs.
• Mortgage integration depth — a mortgage-referral layer (passing leads to Bay Equity-style partners) is light. A full in-house mortgage origination stack with Encompass or Blend integration plus state-level lender licensing is a major investment in its own right.
• Geographic scope — multi-language support (Spanish, French, Mandarin, Hindi, Arabic) and multi-currency add roughly $15,000 to $40,000 of engineering plus ongoing translation maintenance. International builds also need country-specific compliance overlays (RERA in UAE and India, MLS-equivalent feeds, lender licensing).
• Team location — hourly rates vary widely. As a 2026 benchmark, $15 to $40 in India, $80 to $200 in the United States, $70 to $150 in the United Kingdom. Most teams run a hybrid model — a senior US-based product lead paired with a 6 to 10 person engineering pod in India or Eastern Europe to balance time-zone coverage and burn rate.
• Compliance scope — a US-only build needs NAR compliance, state real-estate-commission rules, and fair-housing-disclosure flows. An international build adds RERA (India, UAE), DPDP (India), GDPR (EU and UK), and country-specific brokerage licensing — each of which carries its own engineering and audit cost.

How Redfin-style platforms make money

Seven revenue streams stack on top of one another as the platform matures. Most new builds launch with two or three and layer the rest as transaction volume grows.

• Buyer-agent commission (with partial refund) — the platform earns the buyer-agent commission on each closed transaction (typically 2.5 to 3 percent of the home price) and refunds a portion (0.5 to 1.5 percent) to the buyer. The retained portion funds the salaried-agent operation.
• Seller-agent commission — the platform earns the seller-agent commission (typically 2.5 to 3 percent) when listing the home for sale. Many platforms offer a flat-fee or discounted seller commission to win listings away from traditional brokerages.
• Mortgage origination revenue — when a buyer uses the in-house lender, the platform earns origination fees (typically 0.5 to 1 percent of the loan amount) plus servicing revenue. This is the highest-margin layer Redfin built post-2022 with Bay Equity.
• Title insurance and closing services — the platform earns commission on title insurance, escrow services, and home warranties bundled into closing. Title insurance alone can generate 0.5 to 0.8 percent of home price per transaction.
• Concierge service fees — staging, paint, landscaping, and repair services charged to the seller with payment deferred to closing. Typically generates 1 to 3 percent of home price for sellers who upgrade.
• Premium listings and featured placement — sellers pay for top-of-search placement, homepage featuring, and enhanced photo galleries. Typically $50 to $500 per listing in lower-cost markets, $200 to $5,000 in premium markets.
• Agent subscriptions or seat fees — for marketplace-only platforms that license access to outside agents, monthly subscriptions run $50 to $1,500 per agent depending on tier. Redfin itself does not use this model because it employs agents directly, but most regional Redfin-alternatives layer it on.
• Data and API licensing — the platform's AVM, comp data, and market analytics license out to ESG platforms, lenders, and corporate-relocation services. Typically $25,000 to $300,000 per licensee per year.

The hardest engineering problems and how we solve them

Five sharp edges differentiate a credible Redfin-style platform from a hobby project. These are the parts of the build that buyers, agents, and regulators scrutinize hardest.

MLS feed normalization

Every US MLS publishes its data in a slightly different format — different field names, different photo URL schemes, different status enums, different refresh frequencies. A platform that operates across 10 MLSes has to maintain 10 adapters and a normalization layer that flattens every record into a unified schema. We solve this with a registry-adapter pattern (one adapter per MLS) feeding a unified internal listing schema, plus a nightly contract-compatibility test suite that catches MLS-side schema changes before they corrupt the listing index.

IDX compliance per MLS

Each MLS has its own IDX (Internet Data Exchange) rules — required attribution text, opt-out registry handling, refresh-frequency minimums, image-watermark requirements, and listing-removal SLAs. Violating these rules gets a platform de-listed, which kills the supply pipeline overnight. We solve this with a per-MLS compliance overlay that enforces rules at render time, plus a compliance-audit log that tracks every listing's display state for state-board review.

AVM accuracy and trust

An AVM that is publicly visible and wrong costs more trust than no AVM at all. Zillow's Zestimate has been the subject of public lawsuits and FTC complaints. We solve this with locality-aware models (one regression model per metro rather than a single national model), confidence-interval transparency (always show the low and high band, not just the median), and a public accuracy track record per market that buyers can audit.

Commission transparency post-NAR settlement

The 2024 NAR commission settlement requires agents to negotiate buyer-side commissions directly with buyers rather than relying on MLS-published cooperative-compensation offers. Platforms that hide commission terms violate the settlement. We solve this with an explicit commission-disclosure flow inside the buyer onboarding (where the refund math is shown clearly), plus a per-transaction commission ledger that survives audit.

Salaried-agent payroll versus commission split

Redfin's biggest operational complexity is paying agents on salary rather than commission. The math has to balance fixed labor cost against transaction volume, and a quiet quarter can turn the operation cash-flow negative fast. We solve this with capacity-based lead routing (agents only get more leads when their pipeline has room), per-agent performance scorecards that drive bonus tier, and a real-time profit-and-loss view at the metro level so the operator can rebalance before margin compresses.

What to watch in the next 12 to 24 months

Four trends will reshape the Redfin-style platform landscape between mid-2026 and 2028. Builders who anticipate them in the V1 architecture avoid expensive replatforming later.

• NAR commission settlement implementation continues — the 2024 settlement is still rolling through the industry. Expect further FTC and state-AG action through 2026 and 2027 that will reshape how buyer-side commission is disclosed and paid. Platforms that build commission transparency into the product from day one will lead the next cycle.
• Rocket Companies + Redfin integration outcomes — the 2025 Rocket-Redfin merger is the largest real-estate-plus-mortgage consolidation event in a decade. Watch how the unified product lineup ships through 2026 — the playbook will define what bundled real-estate-plus-mortgage looks like for the rest of the industry.
• AI-generated listings reach scale — AI listing descriptions, AI-staged photos, and AI-generated 3D walkthroughs are all production-grade in 2026. Platforms that adopt them at the listing-creation step cut content cost by 60 to 80 percent and ship higher-quality listings faster than incumbent brokerages.
• Build-to-rent (BTR) takes share — institutional landlords managing tens of thousands of single-family rentals (Invitation Homes, American Homes 4 Rent, Tricon Residential) need a Redfin-style buyer experience layered on top of their leasing operations. Expect a dedicated BTR-Redfin to emerge by 2027.

Frequently Asked Questions

How long does it take to make an app like Redfin?

A marketplace-only build (Zillow-style, without in-house brokerage) takes 6 to 9 months. A full Redfin-style hybrid with salaried-agent operations, in-house mortgage, and bundled title takes 9 to 14 months. A white-label fork of our existing Redfin Clone codebase compresses to 4 to 8 weeks for the marketplace and map layer, with the brokerage operations layer added as a Phase 2 expansion. Most founders pilot with one metro for 90 days before scaling, and that pilot is where 70 percent of the operational lessons get learned.

How much does it cost to make an app like Redfin?

The realistic 2026 range is $45,000 to $90,000 for a basic single-MLS marketplace-only V1, $90,000 to $200,000 for a multi-MLS multi-metro platform with custom-trained AVM and tour booking, and $200,000 to $450,000+ for a full hybrid with salaried-agent operations, in-house mortgage, title, concierge, and multi-region compliance. White-label forks of our existing Redfin Clone compress to $20,000 to $45,000 for the marketplace layer.

Should we be marketplace-only or full brokerage?

Marketplace-only is the right starting point for 80 percent of new entrants. You skip the real-estate licensing, the salaried-agent payroll, and the state-level compliance overlay. Most successful Redfin-alternatives launch marketplace-only in V1, build a relationship with one mortgage partner and one title partner, then add in-house brokerage in Phase 2 once GMV justifies the licensing and operations investment. Going full-brokerage on day one is feasible but doubles the build cost and timeline.

How do MLS integrations work?

Each US MLS publishes its listings via either the legacy RETS protocol or the modern RESO Web API standard. The platform pulls listings on a schedule (every 5 to 15 minutes for premium MLSes), normalizes the data into a unified schema, applies the MLS-specific IDX compliance overlay (required attribution, opt-out registry, image watermarks), and indexes the result into Elasticsearch for search. Each MLS license costs $50 to $500 per month plus IDX setup fees. Most platforms ship V1 with 1 to 3 MLSes and grow to 10 or more by year two.

Should we build our own AVM or license one?

For a V1, licensing an existing AVM (CoreLogic, House Canary, Quantarium, Black Knight, ATTOM) is the fastest path. License fees run $10,000 to $200,000 per year depending on volume. For a long-term moat, building your own AVM trained on local market data is the better path because the data improves with every listing and closing on your platform. Most successful Redfin-alternatives license in V1 and build their own AVM in Phase 2 once they have 12+ months of proprietary closing data to train on.

How does Redfin make money if they refund part of the commission?

The buyer-agent commission in the US is typically 2.5 to 3 percent of the home price. Redfin refunds 0.5 to 1.5 percent and retains the difference (1 to 2 percent) to pay salaried-agent salaries, technology, and operating costs. On top of that, Redfin earns mortgage origination fees through Bay Equity (0.5 to 1 percent of the loan amount), title insurance commission through Title Forward, and concierge service fees from listing-side sellers. The bundled-services revenue is what makes the discounted-commission model viable at scale.

What changed in the 2024 NAR commission rules?

The 2024 NAR commission settlement (resulting from the Sitzer-Burnett antitrust case) made two structural changes. First, buyer-agent commissions can no longer be advertised on the MLS — buyers and their agents must negotiate commissions directly through a written buyer-broker agreement. Second, buyer-broker agreements are now mandatory before agents can show homes. These changes open space for new brokerage models that build commission transparency into the product from day one. Platforms that hide commission terms inside the legacy MLS flow violate the settlement; platforms that surface commission and refund math explicitly to the buyer win trust faster.

What is the moat in a Redfin-style platform business?

Engineering is not the moat — local market density and bundled-services attach rate are. Platforms that dominate transaction volume in a metro (top 3 brokerage market share) build a feedback loop where every additional close trains a better AVM, attracts more agents, and feeds more mortgage and title attach revenue. Platforms that operate in multiple metros at low share tend to lose to specialists. The right strategy is to dominate one metro completely (50+ closings per month) before expanding to a second.