The True Cost of Getting Home — Commute-Cost Data for Zillow

Zillow's commute filter answers "how long." It still can't answer the question that decides where families actually buy: "how much." In 30 of 30 NY/NJ tri-state towns we tested, choosing rail over driving saves an average of $1,924 a month — Summit, NJ alone is $2,302 — and today no US portal shows a buyer that number. Zillow can be first, in one quarter, with a line on every listing: "True commute cost: $15/day door-to-door."

The gap every portal has

Zillow has made commute a strategic re-investment (June 2024 commute-time filter, Sept 2024 AI natural-language search, Jan 2026 Front Porch re-promotion) and made affordability the other flagship theme (BuyAbility, the June 2026 Homebuyer Hub). The two never touch. The commute feature is time-only — no fares, tolls, parking, or fuel anywhere — single-destination, single-commuter, with siloed modes. BuyAbility prices the mortgage and ignores transportation, a household's #2 expense. With $5 gas and NYC's $9 congestion charge live, "what will this commute cost this household" is the consumer question the entire portal industry leaves open:

Redfin bought Walk Score outright in 2014 because a number on every listing becomes the number buyers search by. Post-Rocket ($1.75B, closed July 2025), it's a competitor's stagnant, dollar-less asset. The dollar-denominated successor is unclaimed.

What we built

One engine (live at commutingcost.com), four integration surfaces — each shippable inside a quarter:

1. Nightly per-listing feed — the day-one integration. A batch enrichment file (Parquet/CSV keyed by address, lat-lon, or H3 cell): door-to-door commute time and true monthly dollar cost to metro job anchors across 9 modes — walk, bike, bike-share, scooter, drive, transit, Uber, Park-&-Ride, Kiss-&-Ride, Uber+Rail — each itemized as fare + toll + parking + fuel (+ NYC's $9 charge), plus best-value mode chain. It's a file join and a number rendered on the page, the exact Walk Score pattern: no API dependency, no latency budget, no per-call metering. One GPU reprices a whole metro in ~21 seconds, so nightly refresh is routine, not heroic.
2. Live API — POST /v1/isochrone returns per-hex {times, costs, mode-chains}, AM-peak/off-peak, inbound/outbound, ~0.7 s warm and milliseconds cached. It powers the search filter and gives your Sept-2024 AI search an answer it can't currently compute: "homes under $400/month total commute from my office." The same fields complete BuyAbility into true monthly cost = mortgage + commute.
3. White-label hex-map embed — the interactive cost choropleth on listing and neighborhood pages.
4. Two-commuter search (the premium wedge) — a shipped /v1/multiuser endpoint (live code, not a spec) intersects 2–4 people's time and cost surfaces with per-person modes, minute caps, a combined monthly budget, and school/daycare anchors, and shades the homes that work for the whole household. No portal offers anything like it; gate it behind registration or Showcase as an engagement and lead-quality driver.

Why it can't be trivially copied

• The moat is the curated price database, not the algorithm. 723 cleaned GTFS fare feeds, tolls in 36 states/provinces, 1,515 geocoded CBD garages, 194 bike-share systems' pricing, per-ZIP fuel, per-metro calibrated traffic factors. RAPTOR is public; this corpus isn't.
• Google can't print this number. Its Routes rate card is $5–$15 per 1K with no native isochrone API and no cost dimension — brute-forcing even time-only enrichment at portal scale is seven figures a year — and Google's ToS restricts the storage and display a cost-on-listings product requires. That's a structural barrier, not just a budget one.
• Incumbent data models can't express it. TravelTime and Walk Score are single-traveler, time-or-score systems; household intersection and dollar receipts aren't a feature toggle for them.
• Fresh, not frozen. CNT's H+T index is 2022-vintage modeled data; Walk Score is static. When NYC's $9 congestion charge went live, we repriced the metro in one pass.

Proof (verified, honestly sourced)

Candor: we have no portal integrations or licensing revenue yet — which is exactly why first-mover category exclusivity is still available.

Pricing (proposed anchors — comps here are custom-negotiated)

Benchmark: replicating time-only enrichment on Google's rate card runs seven figures a year — and still can't print a dollar on a listing. Redfin found this category valuable enough to buy Walk Score outright.

Who we are

Krishna Malyala, founder, TLC Engine — a licensed broker (NMLS #1875937) who has watched buyers discover the commute-cost surprise after closing. The engine, map, and data pipeline are live at commutingcost.com.

The ask

A 30-minute demo on your own inventory: pick any of our 54 metros, send 25 listing addresses, and within a week we return the full true-monthly-cost enrichment — every column, real dollars — plus a live walkthrough of the two-commuter search at commutingcost.com/map. From there, a 60-day paid pilot in three metros (badge + cost filter behind a feature flag; measure dwell, saves, and lead quality against time-only controls), with the pilot fee crediting 100% against a national license — before the Rocket/Redfin affordability push makes this the number every portal has to have.

Also fits: Redfin/Rocket (the dollar successor to their own Walk Score, and a commute input to mortgage affordability), Realtor.com and CoStar's Apartments.com/Homes.com (the cost layer that snaps onto their existing TravelTime isochrones — additive, not rip-and-replace), and relocation firms (SIRVA BGRS, Cartus, Graebel), whose destination-selection dashboards have no commute-cost intelligence at all.