Cleaning business quality control: the complete system
Five parts, in the order you should build them, and why most quality programs fall apart at the fourth one.
Ask a hundred building owners what they think of cleaning companies and you get the same answer. Cleaners are great for 90 days, and then the service starts to go downhill. They are not being unfair. They are describing something real. It is rarely laziness. The owner who sold the account and personally walked it for the first month eventually stops walking it, and nothing takes their place.
What replaces the owner is usually an assumption. I cannot get to that building this week, but Frank is a great cleaner, so it is probably fine. That is managing by assumption, and it works right up until the day it does not, which is generally the day the customer calls. By then you are managing a complaint, and the argument about quality is already lost.
A quality control program is what you build so the standard survives the owner not being there. It has five parts. Most companies have two of them and wonder why the program does not hold.
1. The standard: written per site, not per company
A company-wide cleaning standard is a nice document that changes nothing. The standard has to exist at the site level, because the scope of work is at the site level. What gets checked in a medical suite is not what gets checked in a warehouse, and the property manager at each one cares about different things.
Build the site standard directly from the scope of work you sold. Every line in the contract becomes one or more checks. Anything the customer complained about in the past becomes a check, permanently. Anything the crew keeps missing becomes a check. If a line cannot be answered with a straight yes or no by two different people standing in the same spot, rewrite it until it can. Start from the commercial cleaning inspection checklist and cut it down to the site rather than writing one from nothing.
2. The cadence: decided in advance, not by who is shouting
Inspection frequency should be set when the account starts and written down, so it is not competing with whatever is on fire that week. A cadence that works for most portfolios:
- Baseline, before the first clean. Inspect the building the day before you take over. It will score badly, and that is the point.
- Weekly for the first month. New crew, new site, highest risk of losing the customer you just won.
- Monthly once the scores hold. Stable site, stable crew, no complaints in the last 60 days.
- Back to weekly on any trigger. A complaint, a crew change, a scope change, or a new site contact. All four reliably precede a quality drop.
The baseline inspection is the one most companies skip and the one that pays for itself fastest. You take over a building and it comes in at 67. After the first night it is in the low 70s. A week later it is in the 80s. A month later it is in the 90s. That progression is the argument you make at the first review meeting, and you cannot make it if you never scored the building on day zero. The property manager only remembers the after. The baseline is the only place the before still exists, written down with a date on it.
3. The scoring: Pass, Fail, N/A
Score every item Pass, Fail or N/A. There is nothing in between. The moment you introduce a 1 to 5 scale you have created a place for the inspector to hide, and every ambiguous item becomes a 3. A 3 tells the cleaner nothing and tells the customer nothing. The site score just sits there, unmoving.
Pass and Fail forces a decision. Either the mirror streaks at an angle or it does not. Either the bin was wiped or a liner was dropped into a dirty bin. Use N/A only when the item genuinely does not apply at that site (a fixture that does not exist, an area that is locked) and require a note explaining why, or N/A quietly becomes the new 3.
Set the target at 90 percent or higher at all times, and watch the trend line harder than the number. A site drifting down four points a month is a cancellation forming in slow motion.
4. The escalation loop: where most programs die
This is the part everyone gets wrong. Companies buy an inspection form, run the inspections, produce a list of failed items, email that list to a supervisor, and never confirm that a single one of them was fixed. The inspection happened. The quality did not change. All you built was a record of your own problems.
Every failed item needs four things: a photo taken at the moment it was found, one named owner, a due date, and a verification step performed by someone other than the person who fixed it. Miss the photo and the conversation three days later becomes an argument about whether it was really that bad. Miss the verification and you are back to managing by assumption, one level down.
The verification step is the difference between a quality program and a filing system, and it is also the first thing to get skipped when everyone is busy. Get the evidence in front of the person who can act on it (the cleaner themselves, most of the time) with a photo of the problem and instructions to fix it, then have a supervisor confirm the fix before it closes.
You also want to find these things before the customer does. If you are sitting waiting for the phone to ring, you are behind on every account permanently. The goal is that the customer hears about a deficiency from you, already fixed, or never hears about it at all.
5. The proof: what you show at renewal
Customers think about their cleaning company maybe four times a year: when they have a complaint, when they pay the bill, when the contract comes up, and when a competitor knocks on the door with a lower price. Three of those four are moments where you either have evidence or you have an opinion.
Without a record, the renewal conversation is: we think we are doing a good job and we hope you are happy. With a record, it is: we were in your building 26 times this year, your scores held in the 90s, and the average deficiency was closed within a day. One of those conversations supports a price increase. The other one invites a competitor.
That record is also what keeps you from being a commodity. If the customer cannot tell the difference between your work and the next company’s, the only variable left is price, and someone will always be a hundred dollars cheaper.
The first supervisor is the real test
You can run all five parts yourself up to a point. Past that point you hire a supervisor, and the program either survives or it does not.
A supervisor is an employee. They do not look at the building the way an owner does, and you should not expect them to. If the instruction is ambiguous, they walk into a building, do not know what to check, and find nothing, partly because finding something means an awkward conversation with a cleaner they work beside every week. Frank is a great guy, so no report gets written.
The fix is a specific list, a required photo on every failed item, and an inspection they can finish by checking boxes instead of forming an opinion about the room. Do that and you can take a good employee and teach them to see a building the way you do. That is also what lets one supervisor cover more sites instead of you hiring another one, which in my experience is the single largest avoidable cost in a growing cleaning company.
Where to start next week
- Pick your three highest-risk accounts. Usually the newest, the largest, and the one that complained most recently.
- Write a site-specific check list for each, built from the scope of work and every past complaint.
- Inspect all three this week and score them Pass, Fail or N/A. Photograph every fail.
- Assign each failed item an owner and a due date, and verify the fixes yourself before you close them.
- Inspect the same three sites again next week and compare. The second score is the one that tells you whether the loop works.
Do that for a month on three sites and you will know more about your operation than a year of driving around looking at buildings. People do not run businesses. Systems run businesses, and people run systems.
Common questions
What is quality control in a commercial cleaning business?
A written standard for each site, a fixed inspection cadence, a scoring method that produces the same answer no matter who inspects, an escalation loop that closes every failed item, and a record you can show the customer. If any one of the five is missing, the program will not hold up once you pass the point where the owner can personally visit every building.
How often should you inspect each site?
Weekly for the first month of a new account, monthly once the scores hold steady, and back to weekly after any complaint, crew change or scope change. High-visibility accounts and anything with a demanding property manager stay on a two-week cadence permanently.
What is a good inspection score for a cleaning site?
Set the target at 90 percent or higher at all times, and treat anything under 80 as a site that needs a return visit rather than a note. The absolute number matters less than the trend: a site that moves from 74 to 91 over six weeks is a healthy account, and a site that drifts from 95 to 88 over three months is the beginning of a cancellation.
Should cleaners inspect their own work?
They should self-check against the same list, but someone else has to do the scored inspection. A self-check only tells you whether the cleaner understands the standard, not whether it was met, since nobody fails themselves.
How do you get a supervisor to inspect the way the owner would?
Remove the ambiguity. A supervisor who walks into a building with a general instruction to check that it looks good will find nothing, because finding something means an uncomfortable conversation with a cleaner they see every day. Give them a specific list, require a photo on every failed item, and turn the inspection into something they can finish, not something they have to form an opinion about.