Every canvass manager has walked the same Saturday morning. Forty volunteers, clipboards, two-hour shift. The targets are middle-class households with solid turnout history. The route is efficient. The plan is flawless.
The volunteers come back at noon with a 31% contact rate.
The problem isn’t the plan. The problem is that most of the targets weren’t home. Not because they don’t exist, not because the addresses are wrong — because the target voters were, predictably, at work.
This is the single most common mistake in volunteer field operations, and it’s the easiest to fix once you have the right data.
The home-address-only problem
Every voter file carries a home address. That address is accurate, geocoded, and legally verified. Canvass tools built on top of voter files take that address as the target for a knock. Routing optimizes for distance. The volunteer shows up, knocks, and hopes.
What’s missing is a simple question: Is this voter likely to be at this address at this hour?
The answer, for most working adults, is no — and the hours they’re not there are exactly the hours that volunteer canvasses are most likely to run. Saturday morning is the one exception, which is why Saturday morning gets overbooked and the rest of the week is wasted.
The fix isn’t to only canvass Saturdays. The fix is to know, for each voter, what hours they’re likely home.
What the business-license data unlocks
Business licensing is a dataset we aggregate from state and municipal registries. It tells us: what business exists, where it physically is, who owns it, what hours it’s published to be open. When a voter’s name matches a business owner in the same region — specifically, a sole proprietor, LLC member, or principal officer whose business address differs from their home — we have something powerful: a second geographic anchor for that voter.
Concretely, we now know:
- Where the voter sleeps (home address from the voter file).
- Where the voter works, specifically (business address from licensing data).
- The hours they’re likely at each (publication hours from licensing data, plus defaults per industry for businesses that don’t publish).
This generalizes. Even for voters who aren’t business owners, we can often infer work-hour patterns from employer data in FEC filings, census commute patterns for the precinct, or industry-level defaults.
The result is a field-ops plan that knows:
- Don’t knock this door between 9 and 5 on a weekday. The owner of Main Street Auto Body is at the shop. Their house is empty. Shift this knock to Tuesday evening 6–8pm.
- Do knock this door Wednesday 11am. A retiree on this block is home most weekdays. Their spouse is home every weekday.
- Prioritize this block Saturday 10am. Three households here have overlapping weekday work patterns; Saturday morning is their only reliable window.
Yard signs get their own model
Yard signs are a different question. The goal isn’t to reach the household on the sign — the household already put the sign there. The goal is to reach every driver, pedestrian, and neighbor who sees the sign on the way somewhere.
That means yard-sign placement is a visibility problem, not a residency problem. The right location for a yard sign is a high-visibility lot in the hands of a willing supporter — and the “high-visibility” part is exactly what our geocoded catalog can score.
For a given address we compute:
- Daily traffic volume from census commute patterns and road-class weighting
- Directional visibility (is the lot on the right side of the road for morning inbound? evening outbound?)
- Sign line-of-sight from setback, frontage, and nearby obstructions
- Supporter density within line-of-sight (is this a block that flips sign-to-sign into an identity moment?)
Every supporter who raises their hand for a sign gets a score for how much traffic their sign will actually reach. A campaign with 400 volunteers willing to take signs should put the 400 best signs up — not the 400 most convenient.
Business-hour routing, specifically
Here’s the routing rule we apply by default in Civitas walk plans:
- Weekday mornings (9am–12pm): Route to households with at least one member retired, unemployed, home-office, or in a service industry with common morning-off patterns (e.g., hospitality, night-shift healthcare).
- Weekday afternoons (12pm–5pm): Route to households with flexible-schedule members and supporters on leave. Skip dual-earner households with published 9-to-5 schedules.
- Weekday evenings (5:30pm–8:30pm): Prime window for business-owner households. Route heavily here if your targets are entrepreneur-heavy.
- Saturday mornings (9:30am–12:30pm): The universal window. Save this for the highest-EV targets — the ones where a conversation is most likely to persuade.
- Sunday afternoons (1pm–5pm): Strong secondary window. Culturally mixed — some districts skip Sundays, others don’t.
The routing is configurable. If a candidate or volunteer organizer wants to override the model for religious, cultural, or strategic reasons, they can. The default just exists because most campaigns default to “knock when we have volunteers,” not “knock when voters are home.”
What changes
Published 31% contact rates are the baseline. Early internal tests with business-hour-aware routing have moved contact rates to the 48%–54% range on comparable universes. Half again as many conversations per volunteer hour.
The second-order effect matters more: volunteers don’t quit. A volunteer who knocks 50 doors and has 27 conversations feels like they made a difference. A volunteer who knocks 50 doors and has 14 conversations is more likely to decide canvassing doesn’t work and stay home next week. Field teams die at the retention layer, not the volume layer. Good routing keeps the team alive.
The tooling
Inside Civitas, this shows up as a single toggle when you build a walk plan: “Respect work hours.” On by default. The router uses your target universe, the hour of your shift, and the business-licensing and employment signals we have for each household, and builds the route around contact probability, not just distance.
The output is a printable packet: street map, ordered door list, scripts, tally sheets. It prints cleanly on a canvass clipboard. The tech disappears into a paper workflow a 16-year-old volunteer can use.
The data was public. The licensing files were public. The voter file was public. What was missing was the join. That’s what we build.