Replacement windows demand lives in pre-1990 owner-occupied single-family stock — not in a 15-mile radius. ZIP fits a windows-specific geo-MMM against Census housing-age, median HHI, and financing-tier overlays, then ranks which channels and ZIPs are driving sat appointments this week.
ZIP pulls windows-specific lead source, appointment set, sat, and close data from MarketSharp, improveit 360, HubSpot, or Salesforce — never blurred with other product lines.
Census ACS pre-1990 SF share, median HHI, and GreenSky/Synchrony approval-band proxy are joined to every ZIP so demand and closability are scored together.
Weekly list: shift $X from broad Meta into Google LSAs in Fairless Hills (pre-1980 stock, 74% sit rate), cut Display retargeting in oversaturated NE Philly.
Roofing is event-driven (hail, storm). Windows are lifestyle + housing-age driven. Modeling them together washes out the pre-1990 housing signal that separates a great windows ZIP from a bad one.
Windows tickets clear near-prime credit thresholds cleanly. ZIP flags ZIPs below the threshold and recommends cash-sale creative or Wisetack alternatives instead of GreenSky.
Rural low-density and dense-urban rental clusters both no-show windows appointments at 40%+. ZIP flags them upfront so your ad budget stops feeding the graveyards.
Full-frame vs. insert vs. multi-window jobs price 3× differently. ZIP correlates ticket size to ZIP archetype so channel bids match the expected job value.
"We were spending like windows and roofing were the same product. ZIP fit a windows-only model against pre-1980 housing and financing tier. We cut three rural ZIPs and doubled down on suburban pre-1990 stock. Sat appointments up 37% in a quarter."
Roughly $5,000/month across at least two channels and two service ZIPs, focused on windows. Below that, the causal signal is too noisy for honest ZIP-level attribution. Most in-home windows operators clear this easily.
ZIP tracks ticket size as a ZIP-level attribute so the model expects higher-value work in specific archetypes. It doesn't split into three separate models — one windows model with ticket-size weighting is more statistically robust than three thin models.
Yes. ZIP scores every ZIP in your metro on pre-1990 SF stock, median HHI, financing-tier fit, and competitive density — even before you've spent a dollar there — and recommends the highest-expected-lift expansion candidates.
ZIP measures your incremental lift against your own historical baseline in each ZIP, not against absolute market share. It flags ZIPs where a dominant national windows brand is suppressing your close rate so you can decide to concede or invest.
Tell us your business, city, and product line. We'll return a ranked ZIP-level budget shift you can execute this week.