Key takeaways
- Home size sets the base price; condition is what AI estimates from photos, and condition is where quotes go wrong.
- Write your pricing rules down before any tool sees them: base rates, condition levels, add-ons, frequency discounts and minimums.
- A person reviews every estimate where the photos show heavy buildup, clutter or pet hair, and every commercial job.
- Follow up open quotes with a held slot and the recurring rate, not with "just checking in".
AI estimates for a cleaning company work by reading a short form and a few photos, judging the condition of the home against your pricing rules, and drafting a quote for a person to approve. Home size sets the base. Condition is the part that sinks a first clean when it is guessed wrong, and it is what photos and AI help with most.
How photo quoting works
Photo quoting works by pairing facts the customer knows with a judgment they cannot make. The customer can tell you bedrooms, bathrooms, rough square footage, pets and how often they want service. They cannot tell you whether their kitchen is a standard clean or a two-hour degrease. Photos can.
A typical flow for a residential request:
- The customer fills in home size, pets and frequency, then uploads photos of the kitchen, the main bathroom and one living area.
- The AI rates condition against levels you define (for example light, moderate or heavy) and flags anything outside your rules, such as construction dust, mold or a hoarding situation.
- It applies your rate for that size and condition, adds any add-ons the customer picked, and drafts two prices: the first deep clean and the recurring visit after it.
- Your office reviews, adjusts if needed, and sends the quote from your CRM.
Move-out cleans work the same way with one difference: the home is usually empty, so the photos that matter are inside the oven, the fridge and the cabinets. Commercial work does not fit this flow. Offices, clinics and gyms need a walkthrough and a scope of work, and AI's job there is drafting the proposal from your walkthrough notes.
How photo quoting works
- The customer sends photos From a text or a form
- AI drafts the quote From your price book and rules
- A person reviews it Nothing is sent unchecked
- The customer receives it With a way to book
- Follow up open quotes Until a yes or a no
What you need first
- Write down your pricing rules. List your base rates by bedrooms and bathrooms or square footage, your condition levels with a sentence describing each, your add-ons, your frequency discounts and your minimum visit.
- Collect past quotes with outcomes. Pull recent quotes from Jobber, ZenMaid or your cleaning CRM, along with how long the first clean actually took. That is how you check whether the rules match reality.
- Take reference photos. Photograph homes your teams have cleaned, with permission, at each condition level. These calibrate what "moderate" means in your company.
- Decide what is out of scope. Biohazard, hoarding, post-construction and anything your teams do not do should trigger a "we will call you" reply, not a price.
- Set the review rules. Decide which estimates a person must approve before they go out. The table below is a starting point.
- Write the privacy line. Tell customers how their photos are used and that they are deleted after quoting. Then do that.
Accuracy and human review
AI estimates are usually close on size and less reliable on condition, because photos are taken in good light and from flattering angles. A kitchen photo cannot show the grease on top of the cabinets or the state of the oven door. That is why the review rules matter more than the tool.
| Estimate type | AI drafts it | A person approves when |
|---|---|---|
| Standard recurring clean, light condition | Yes | Spot checks only |
| First deep clean, moderate condition | Yes | Always, before sending |
| Heavy condition, pet hair or clutter | Yes, flagged | Always, often with a short video call |
| Move-out clean | Yes | When appliances or cabinets were not photographed |
| Out-of-scope work | No | A person calls back |
| Commercial janitorial | Proposal draft only | Always, after a walkthrough |
Track the gap between the estimated time and the real time for the first few weeks, and feed it back into the rules. If first deep cleans keep running long, your "moderate" level is too generous, and your teams are absorbing the difference.
Customer photos are for quoting. If you want to use a home in marketing, ask separately and in writing. If a client's words or photos appear in an ad, the FTC's endorsement guides apply (16 CFR Part 255).
Following up open quotes
An open quote is a household that wanted a price and has not said yes. Following up well means giving them a reason to decide, not a reminder that you exist.
- Hold a slot, with an end time. "We are holding Tuesday morning for you until Friday." It turns a vague maybe into a real choice.
- Show the recurring rate again. Many households stall at the first deep-clean price. Seeing that biweekly visits cost less afterward often moves them.
- Ask what changed. "Did the price or the timing not work?" The answer tells you whether to adjust or let go.
- Tie move-out quotes to the lease date. Those clients have a deadline. Follow up a week before it.
- Close it cleanly. After a week, mark it lost with a reason, and move it to a seasonal list.
Quotes that become first cleans only pay when the first clean becomes a plan. The repeat customers guide picks up from there, and the online booking guide covers letting customers accept and schedule without a call.
How VETS Growth Marketing helps
We turn your pricing into written rules, build the photo-and-form estimate flow, set the review steps your office uses, and connect it to the CRM you already run. Accounts and data stay in your name, the build runs to agreed dates, and we report how many quotes went out, how fast, and how many became recurring clients. Book a growth consultation and bring a few recent quotes so we can test the rules against them.
Sources
- 16 CFR Part 255: Guides concerning endorsements and testimonials in advertising, Electronic Code of Federal Regulations