How We Test Window Cleaning Robots

Every robot reviewed on Window Robot Guide goes through the same six-test battery, run on the same window configurations, with the same timing and measurement tools. The protocol does not change based on price tier, brand, or manufacturer relationship. A $130 budget robot and a $500 flagship robot are tested identically.

The testing period is a minimum of 14 days and 20 complete cleaning cycles. No verdict is written before that threshold is met.

Testing Environment

All testing is conducted in a standard residential home with mixed window types. Window configurations tested for every review:

• Standard double-hung vinyl-framed window (36″ × 48″) — primary benchmark surface
• Casement window
• Sliding glass door panel
• Interior mirror surface
• Exterior-facing window, light grime (pollen, dust film, 4–6 weeks uncleaned)
• Exterior-facing window, heavy grime (mineral deposits, bird residue, 8–12 weeks uncleaned)

Solution types tested on every robot:

• OEM branded cleaning solution supplied or recommended by the manufacturer
• Distilled water + isopropyl alcohol (70/30 mix)
• Tap water + diluted dish soap

The Six Standard Tests

Test 1: Light Soil Test

Window condition: interior glass cleaned within the past 4–6 weeks with accumulated dust film and fingerprints. The robot runs in its standard Quick Clean or equivalent mode using OEM solution first, then distilled+IPA. What is recorded: completion time in minutes and seconds, streak outcome under direct overhead light, streak outcome at 45° side angle, and Corner Residual at all four corners measured with a ruler.

Test 2: Heavy Grime Test

Window condition: exterior-facing glass, last cleaned 8–12 weeks prior, with visible pollen film and light mineral residue. The robot runs in Thorough Clean or Deep Clean mode using OEM solution. What is recorded: completion time, what the robot removed versus what remained, whether pre-treatment was required for acceptable results, and Corner Residual.

Test 3: Streak Assessment — Three Solution Types

The same interior window is cleaned to a neutral baseline between each run. Three consecutive runs use three different solution types: OEM solution, distilled+IPA, and tap water + dish soap. Each run produces a streak outcome assessment under direct light and at 45° side light. The result is recorded as None, Faint, or Visible for each condition. These six data points constitute the robot’s Streak Index — a metric that exists nowhere else in this category and is the primary basis for solution recommendations in every review.

Test 4: Corner and Edge Coverage

The standard 36″ × 48″ double-hung window is cleaned in a full Thorough Clean cycle. Corner Residual is measured at all four corners before any Edge Clean mode is activated. If the robot has an Edge Clean mode, it is then run as a finishing pass and Corner Residual is measured again. The time added by the Edge Clean pass is recorded. The difference between pre- and post-Edge Clean measurements is reported as a real data point in every review.

Test 5: Navigation Stress Test

Three non-standard window configurations are tested: a double-hung window with a center horizontal rail, a casement or arched window, and a narrow window under 12 inches wide where available. What is recorded: Dead Zones identified (reproducible areas the robot’s navigation consistently misses), false edge-stops, how many times manual repositioning was required, and whether the robot completes a full cleaning pass without intervention.

Test 6: Day 1 vs Day 14 Pad Degradation

The same interior window is cleaned with the same solution in the same mode on Day 1 and Day 14 of the testing period, after 20+ cycles of regular use. Completion time, streak outcome, and Corner Residual are recorded on both days. The difference documents how pad fiber degradation and absorption capacity reduction affect cleaning quality over time — information that is entirely absent from manufacturer testing claims.

The Metrics That Define This Site

Corner Residual

Corner Residual is the width in inches of uncleaned border remaining at window corners after a full cleaning cycle. It is measured with a ruler at all four corners. The average and worst-case measurements are both reported.

Category benchmarks: under 0.2 inches is excellent (premium robots with dedicated corner systems), 0.2–0.4 inches is good (mid-range with corner brushes), 0.4–0.7 inches is acceptable (entry-level), over 0.7 inches is poor.

Streak Index

The Streak Index is a robot’s aggregate streak rate across three solution types — OEM branded, distilled water + IPA 70/30, and tap water + dish soap — measured post-clean under both direct and 45° light.

It produces six data points per robot: None, Faint, or Visible for each solution type under each lighting condition. It is the primary basis for solution recommendations and the most direct indicator of which robots are chemistry-sensitive.

Pad Tax

The Pad Tax is the true annual cost of owning a window robot: OEM or third-party pad replacement cost, cleaning solution consumption, and the time cost of pad management at $15 per hour. For a 10-window home cleaned bi-weekly, pad management time runs approximately 8.7 hours per year.

Combined with pad and solution costs, the annual Pad Tax for most households runs $55–$90. This is a recurring cost that compounds against the robot’s one-time purchase price. It is calculated for every robot reviewed on this site.

Dead Zone

A Dead Zone is a surface area the robot’s navigation algorithm consistently skips or terminates early — not a one-time navigation anomaly, a reproducible mapping failure confirmed across multiple windows of the same configuration.

Dead Zones are documented by window type and confirmed through repeated observation.

Cross-Robot Comparison Standard

Every major test section in every review contains a direct comparison to at least one previously reviewed robot using the same measured metrics.

A review that says ‘performs well in corners’ without a number or a named comparison robot does not meet the publication standard for this site. All cross-robot data is maintained in a central benchmarks table, updated after each new review.

What Testing Does Not Cover

Window Robot Guide does not conduct high-rise drop testing, accelerated wear simulations, or multi-year lifespan projections. Safety testing at this site is limited to the power-interruption Recovery Time test and the suction adhesion tap-pressure test.

For high-rise use, the safety cord that ships with every corded robot is always recommended. The safety system section of every review states the test conditions explicitly — a safety claim without a named test condition does not appear on this site.