Site Selection

Operators in this industry have a saying that’s directionally honest: a great site with a mediocre operator beats a great operator on a mediocre site, almost every time. The data supports this. Cross-sectional analysis of express tunnels shows that location-driven factors — traffic, demographics, competitive density, and site characteristics — explain the large majority of variance in per-site performance. Everything else operators do matters, but operating excellence on a poor site is bailing water.

This is also where most of the hardest, most defensible analytical work happens in this industry. Wash Index was built specifically to systematize it.

The four factors that determine site performance

1. Traffic

The basic input is vehicle traffic past the site, typically expressed as Annual Average Daily Traffic (AADT). State and municipal DOT data publish AADT counts for most named roads. Useful benchmarks:

• Below ~20,000 AADT: generally too thin to support a modern express site unless other factors are exceptional.
• 25,000–40,000 AADT: workable range, but performance is highly dependent on demographics, capture, and visibility.
• 40,000+ AADT: the zone where most top-performing express sites operate.

AADT is a starting point, not an answer. Two sites at 35,000 AADT can perform very differently based on the composition of that traffic (commuter vs. through-traffic vs. local), the speed at which it passes the site, and the ease of pulling in.

2. Demographics

The demand for paid car washes correlates with household income, vehicle ownership rates, and population density within a defined trade area, typically a 3–5 mile radius or a 10-minute drive-time isochrone. Useful guideposts:

• Median household income in the trade area above ~$60,000 is a reasonable floor for membership-driven sites; premium-priced sites benefit from $80,000+.
• Vehicles per household above ~1.8 supports stronger demand than lower-vehicle-density areas.
• Population in trade area depends on competitive density, but 30,000–80,000 people in a 5-mile radius is a common range for viable sites.
• Daytime population matters too — sites in employment-dense areas capture lunch and commute washes.

The right framework is not absolute income thresholds but relative attractiveness within the local competitive set: an income level that supports a premium site in a low-cost metro may not support the same site in a high-cost metro where customer expectations are higher.

3. Competitive density and saturation

This is the factor that has changed most dramatically over the last five years and is now the single largest underwriting risk in many metros.

In an underserved market, demand for paid car washes is elastic — adding a site grows the market. In a saturated market, demand is largely fixed and a new site primarily cannibalizes existing ones, including potentially your own. Several markets have crossed that threshold:

• Phoenix, Houston, Dallas-Fort Worth, Charlotte, and parts of Florida have seen rapid express tunnel buildout. New-site economics in these metros are materially worse than they were five years ago.
• Saturation is hyper-local. Within the same metro, one trade area may be saturated while another two miles away has none. The relevant unit of analysis is the trade area, not the metro.

The right metric is express tunnels per 10,000 households in the trade area. Below roughly 0.5, the market is generally still expandable. Above 1.0, you are competing for existing washers. Above 1.5–2.0, you are almost certainly destroying value with new sites.

4. Site characteristics

Two sites with identical traffic, demographics, and competition can perform very differently based on the physical and operational characteristics of the parcel:

• Visibility from primary road. Sites visible from a half-mile out perform better than tucked-away sites by a meaningful margin.
• Ingress and egress. Easy in, easy out, ideally with right-in/right-out from the primary travel direction. Sites that require a U-turn or a left across traffic lose meaningful demand.
• Stacking capacity. Number of cars that can queue before the tunnel without spilling into the street. Undersized stacking caps peak throughput and trains regular customers to go elsewhere.
• Vacuum count and layout. Vacuum availability affects member retention; members who consistently can’t find a free vacuum churn faster.
• Side of street. “Going-home side” sites (typically the right side of the primary commuter direction in the evening) outperform “going-to-work side” sites, by enough that operators specifically target the former.
• Lot size and ability to add tunnels later. Optionality on future expansion is a real value driver, particularly for platforms underwriting long-hold strategies.

The Wash Index approach

The framework above is industry-standard. What separates rigorous site analysis from generic site analysis is the underlying data work — and specifically, the willingness to do at scale what most diligence processes do anecdotally for one or two sites.

Wash Index pulls together four primary data layers:

Vehicle traffic and demographics

• Vehicle traffic data (AADT) at the parcel level, not just the corridor average. Two sites on the same road can have meaningfully different actual exposure based on intersection geometry, signal placement, and the specific count station the data is drawn from.
• Census and household demographics at trade-area granularity — typically 3-mile and 5-mile radii and 10-minute drive-time isochrones. Includes household income, vehicle ownership, daytime population, and age distribution.

Competitive site mapping

We maintain a structured inventory of every car wash within the relevant trade area, classified by operator, format (express, in-bay, full-serve, self-serve, hybrid), parent platform if applicable, approximate age, and observable equipment generation. This is the base layer that every other piece of competitive analysis sits on top of.

LLM-based review analysis on 55 dimensions

This is the most distinctive piece of the methodology and the one that produces signal nothing else in the industry currently provides at scale.

Every Google review for every competing site is processed by a large language model that scores the review across 55 structured dimensions. We don’t just count stars and read sentiment. We extract specific, comparable signal across categories including:

• Wait time and throughput. How often customers mention long lines, stacking spillover, slow tunnel cycles, or congestion at peak times.
• Wash quality. Streaking, residue, missed spots, tire shine quality, drying performance, brush damage complaints.
• Membership experience. Mentions of value, RFID reader reliability, member lane effectiveness, ease of cancellation, billing issues, perceived churn drivers.
• Pricing perception. Price-fairness language, comparisons to competitors, comments on price increases, single-wash vs. membership value framing.
• Staff and service interactions. Greeter behavior, capture pressure (positive and negative), conflict resolution, perceived training quality.
• Facility and ancillary. Vacuum count and reliability, vacuum suction strength, mat washers, free vs. paid vacuums, lot cleanliness, lighting at night.
• Equipment condition and damage. Reports of vehicle damage, antenna or mirror issues, equipment downtime, broken arches or dryers.
• Site access. Difficulty entering or exiting, traffic light timing, ingress queue blocking the road.
• Trip context. Whether customers self-identify as members, single-wash, first-time, or returning; commute-driven vs. errand-driven visits.
• Solicitation and conversion proxies. Mentions of being asked to join membership, perceived sales pressure, signing-up experience.

The 55 dimensions are applied consistently across every reviewed site, which is the part that matters. Five-star averages collapse a lot of useful signal — a 4.6 site with chronic wait-time complaints is a different investment than a 4.6 site with universally positive reviews and one or two equipment outliers. Looking at thousands of reviews per site across 55 consistent dimensions surfaces these differences in a way that reading reviews manually cannot.

We use the resulting signal as a proxy for things that are otherwise hard to observe from outside: how busy a site actually is, where its operational weaknesses are, how its membership program is perceived, and how it compares to the rest of the competitive set on dimensions that affect customer churn and acquisition.

Pricing and membership intelligence

We systematically scrape the pricing and membership structures of competing car washes — directly from operator websites, from in-tunnel menu photos surfaced in reviews, and from third-party listings — and structure them for comparison. For each competing site we maintain:

• Single-wash pricing across all tiers. Base, mid, top, and any premium-named tiers.
• Membership plan structure. Number of tiers, monthly price points, what’s included at each tier, and how the plans are positioned against single wash.
• Plan-level perks. Free vacuums, member lanes, premium add-ons (ceramic, graphene, undercarriage, tire shine), guest passes, multi-vehicle discounts.
• Promotional pricing and trial offers. First-month discounts, referral structures, seasonal promotions.
• Pricing history. Tracked over time, so we can see which competitors are raising prices, when, and by how much.

This produces a real competitive pricing map for any trade area — what the local market looks like, where pricing power sits, and where the target site or platform is over- or under-priced relative to peers. For a buyer evaluating an acquisition, it answers the “is there pricing optionality here” question with data rather than with the seller’s narrative.

What the combination produces

Layered together, these data sources support site-level scoring that is meaningfully more defensible than either drive-by intuition or any single-source traffic study. Specifically, they answer questions that matter directly for underwriting:

• Is this trade area genuinely underserved, or is it on the wrong side of the saturation line?
• Among the competing sites in the trade area, which are operationally strong and which are operationally weak — and where does the target sit in that ordering?
• Is the local pricing environment healthy or compressed?
• Are competitors visibly struggling with wait times, equipment, or member experience in ways that create an opening for a better-run site?
• What does the membership maturity look like across the local market, and how much penetration headroom exists?

For PE underwriters evaluating a portfolio of sites — especially in bolt-on diligence where the time available per site is limited — this is the level of analysis that should be brought to investment committee, not anecdote from a site visit and a static demographic pull.

How site selection actually goes wrong

Most failed sites we’ve examined share one or more of the following:

• Cheap real estate that no one drives past. A 30% discount on land cost is not a 30% discount on returns when traffic is structurally insufficient.
• Site visibility lost to existing buildings, mature trees, or signage rules.
• Saturation that the buyer didn’t model. The trade area looked fine on day one and was already over the line by the time the third nearby competitor opened.
• Demographic mismatch. Premium format pricing in a market that supports only base-tier demand.
• Operationally compromised site characteristics — bad ingress, undersized stacking — that capped peak throughput and limited membership growth.

None of these are operator failures. They are underwriting failures, made before the deal closed.

The decision

Site selection is the most leveraged decision in this asset class. Spending an extra $200,000 on data, traffic studies, and trade-area analysis to validate a $5 million site is some of the highest-return work an investor can do, and the cost of getting it wrong compounds for the entire hold period. The next page covers another decision that compounds: the membership model that turns transactional traffic into recurring revenue.