A sample sketch and quote: Heat pumps without retyping: requests, visits and inspections in one system
This is what you get from us at the start. First the sketch: one business day after five sentences, no meeting. Then a free week of research: your site and app, your competitors, their prices and colours. At the end of that week, an analysis with your office's data and a written quote.
ExampleThe company, its competitors, their numbers and this quote are illustrative and point to no existing business. The structure, the method we price by and the market prices beside the quote, each with its source, are real.
- For
- An installer in Greater Poland: heat pumps and ventilation, 14 technicians, two people in the office
- Prepared by
- The founders of BizBees Studio
- Part 1
- Sketch · one business day after five sentences · free
- Part 2
- Research and review · week 1, free: site, app, competitors, prices and colours
- Part 3
- Analysis and quote · end of week 1, with the office's data
Part 1
The sketch
1 business day · free · no meeting, no contract
Five sentences from the client and one question from us
That is enough to start. We ask the sixth question: how will you know it worked? The answer decides what we count next.
From the client
We install heat pumps and ventilation in Greater Poland, with 14 technicians and two people in the office.
Requests come by phone, e-mail and text; the office types them into a spreadsheet and phones the technicians with assignments.
In 2022 we ordered an Android app, but the technicians don't use it: they fill in a paper report, and in the evening the office retypes it into the manufacturers' portals so the client keeps the warranty.
We book the yearly inspections from memory, and some of them slip.
We want clients to report faults on our site and book inspections themselves, and the office to stop retyping.
How will you know it worked?
The office takes no calls about status, and every inspection is booked before it falls due.
What we read in it
- The bottleneck is the office. It links the client, the technician and the manufacturer, and everything passes through it by hand.
- Money leaks through inspections. Every one that slips is a lost margin and a risk of a warranty dispute.
- The goal can be measured: the number of status calls and the share of inspections booked before they fall due. We check both after three months.
A one-page drawing of the solution
Today and after the change, step by step. You correct this drawing yourself before we count anything more precisely.
Request
1Today: The client phones, e-mails or texts. The office types the request into a spreadsheet.
1After the change: The client reports the fault on a page, with a photo. The request joins the queue with the device's history.
Assignment
Today: The office phones a technician and reads out the address.
After the change: The office drags the job onto a technician in the calendar. The address and history wait in the app.
Status
2Today: The client phones: when are you coming?
2After the change: The client gets a text with a link: assigned, on the way, on site.
Visit
Today: The technician fills in a paper report. The app from 2022 sits unused.
After the change: The technician writes the report in the new app, offline too: photos, readings and the client's signature on screen.
Warranty
3Today: In the evening the office retypes the report into the manufacturer's portal.
3After the change: The report comes out in the portal's format. The office checks it and sends it the same day.
Inspection
4Today: Inspections are booked from memory. Some slip.
4After the change: 30 days before it falls due, the client gets a reminder and picks a time.
Mail
Today: The office reads e-mails and texts and retypes them into the spreadsheet.
After the change: An AI agent turns an e-mail into a draft request. The office approves or corrects it.
- where time or money leaks today
- how we close it
- prepared by AI, approved by a person
Three things to do first
In the order in which they pay back fastest.
Reports in an app instead of paper
The office loses most time to retyping. We start with the technicians' app and the panel, because without them no later step has any data.
closes 3
Inspections that remind themselves
The device base from the spreadsheet, a reminder 30 days before the due date and a time the client picks. This is the money that slips today.
closes 4
Requests and status without a phone call
The client reports a fault on a page and gets a text with a link to the visit's status. The office stops retyping requests and answering questions whose answer already exists.
closes 1 and 2
Price range
€4,000–11,000
net, depending on the scope
- Start€4,000–6,000
- Recommended€5,000–8,000
- Full€7,000–11,000
The scope we recommend: Recommended, about 4 weeks to launch. You get the exact quote a week later, with the office's data.
To check in week 1
- How many requests and inspections do you have a month, and how do they spread over the year?
- Which manufacturers do you service, and what are their warranty portals like?
- Where is the device base today, and how much is missing from it?
- Who has access to the code and database of the 2022 app?
- Which phones do the technicians use?
Part 2
Research and review
week 1 · free · the client's site and app, three competitors, their prices and colours
The site today: slow on a phone, with no way to report a fault
We measured the home page with Lighthouse on the phone profile (the median of three runs) and walked its 11 pages as a client looking for a service would.
Lighthouse on a phone
0–100: 90 and above is good, below 50 is poor
Performance
41
poor
Accessibility
72
needs work
Best practices
78
needs work
SEO
69
needs work
Loading and stability, same runs
Largest element (LCP)
6.8 s · poor
good up to 2.5 s · a 2.1 MB project photo in the header
Layout shift (CLS)
0.24 · needs work
good up to 0.1 · a promotion banner slides in above the content
Blocking time (TBT)
1,240 ms · poor
good up to 200 ms · a chat widget and a slider on the home page
What we found
| What is there | What it costs | Weight |
|---|---|---|
| There is no fault-report form: only a phone number and an e-mail address. | Every request passes through the office, by phone or e-mail. | high |
| The phone number in the header is an image, so a phone cannot call it with a tap. | On a phone it cannot be tapped or copied; clients type it in by hand. | high |
| The home page's images weigh 4.2 MB together and have no phone-sized versions. | On a phone the page takes almost 7 seconds to load. | medium |
| There is no page about inspections: what they cover, what they cost and why the warranty depends on them. | Clients do not know an inspection is lapsing until the office calls. | medium |
| The home page has no H1 heading, and 9 of the 11 pages share one title. | A search engine cannot tell which page answers “heat pump inspection”. | medium |
| The contact form does not say who processes the data, or why. | The notice GDPR Article 13 requires is missing. | medium |
What works
- Client reviews and a map of the service area on the home page.
- On a computer the page loads well: performance 88.
- The project photos are real and good. They only need to be smaller.
Fault reports and visit status get their own light page in the recommended scope. Fixes to the current site (images, the phone number, headings, the GDPR notice) take about a day; they are not in the quote, because your web developer can make them from this list.
The app from 2022: why the technicians do not go back to it
We got the installation file and access to its panel, and talked to three technicians. A developer wrote the code in 2022, and nobody is in touch with them any more.
2
of 14 technicians opened it in the last month
03/2023
the last version; no security updates since
3,100
visits in its database: history worth moving over
12 of 14 at the start, 2 today. The technicians say reports get lost without a signal, so they prefer paper.
Data as a table
| Quarter | Technicians |
|---|---|
| Q3 2022 | 12 / 14 |
| Q4 2022 | 11 / 14 |
| Q1 2023 | 9 / 14 |
| Q2 2023 | 7 / 14 |
| Q3 2023 | 6 / 14 |
| Q4 2023 | 5 / 14 |
| Q1 2024 | 5 / 14 |
| Q2 2024 | 4 / 14 |
| Q3 2024 | 4 / 14 |
| Q4 2024 | 3 / 14 |
| Q1 2025 | 3 / 14 |
| Q2 2025 | 3 / 14 |
| Q3 2025 | 2 / 14 |
| Q4 2025 | 2 / 14 |
| Q1 2026 | 2 / 14 |
| Q2 2026 | 2 / 14 |
| Q3 2026 | 2 / 14 |
What we found
| What is there | What it costs | Weight |
|---|---|---|
| Without a signal the report is not saved. In a boiler room or a basement the technician loses the whole visit. | The technicians went back to paper. | high |
| Photos are scaled down to 640 pixels wide. | The serial number on the rating plate cannot be read, and the manufacturer's portal requires it. | high |
| All technicians sign in to one shared account. | Nobody can tell who wrote which report, and the password stays the same when someone leaves. | high |
| The database on the server has no backups. | One server failure erases the history of every visit. | high |
| The technician cannot see a device's history. | On site they phone the office to ask what was done last time. | medium |
| There is no export to the manufacturers' portals or to a spreadsheet. | The office retypes every report anyway. | medium |
| It is installed from an APK file, outside Google Play; the last version is from March 2023. | It gets no security fixes, and new phones warn against installing it. | medium |
Decision: we build a new app
Offline mode, full-size photos and personal accounts amount to a new app on old code that nobody knows. We build a new one and move the visit history over from the old database; the import fits in the backend module.
Three competitors: none of them shows clients a visit's status
We checked the three firms the client named from the outside, the way their clients see them: the site, the way to report a fault, the price list and the Google reviews.
- Client
- 14 technicians, Greater Poland
- Competitor A
- an installer network in three regions, about 60 technicians
- Competitor B
- a local firm near Poznań, about 8 technicians
- Competitor C
- a heat-pump manufacturer's authorised service
What a client can do on the site
| On the site | Client | Competitor A | Competitor B | Competitor C |
|---|---|---|---|---|
| Report a fault through a form | no | yes | no | yes |
| Check a visit's status | no | no | no | no |
| Book an inspection themselves | no | partly | no | no |
| Get an inspection reminder | no | yes | no | partly |
| See the inspection's price | no | yes | yes | yes |
| Google rating | 4.123 reviews | 4.6412 reviews | 4.861 reviews | 3.9140 reviews |
| Site performance on a phone (Lighthouse) | 41 | 78 | 55 | 63 |
- Competitor A, booking: a form asking for a date, which the office confirms by phone.
- Competitor C, reminder: an e-mail from the manufacturer, with no link to book a visit.
We sorted the reviews into themes by hand; each has one main theme.
Data as a table
| Theme | Reviews |
|---|---|
| Cannot get through on the phone | 31 |
| A long wait for a date | 27 |
| No word on when they will arrive | 19 |
| An inspection lapsed or was forgotten | 9 |
| A price different from the call | 8 |
What follows
- Nobody nearby offers a visit's status or self-booked inspections. The client can take that place first.
- The three most common complaints, the phone, the wait and the status, are exactly what the drawing in part 1 closes.
- Competitor A wins on the number of reviews: 412 against 23. A text asking for a review after each visit fits in the client-page module.
The client's prices against the competition
Gross prices from the three competitors' published price lists (October 2026) and from the client's own. The authorised service is the dearest in everything it offers.
- client
- competitors A, B and C
- our recommendation
Heat-pump inspection
Ventilation unit inspection
Call-out and fault diagnosis
Annual service plan
Data as a table
| Service | Client | Competitor A | Competitor B | Competitor C | Recommendation |
|---|---|---|---|---|---|
| Heat-pump inspection | PLN 450 | PLN 590 | PLN 420 | PLN 690 | 520–PLN 550 |
| Ventilation unit inspection | PLN 300 | PLN 390 | PLN 280 | not offered | – |
| Call-out and fault diagnosis | PLN 250 | PLN 199 | PLN 300 | PLN 350 | – |
| Annual service plan | not offered | PLN 790 | not offered | PLN 990 | 690–PLN 790 |
What we propose
- A heat-pump inspection at 520–PLN 550 once reminders and online booking run. Today the client is PLN 140 below Competitor A and PLN 240 below the authorised service, and after launch it will give more than they do: a reminder and a time picked without a phone call.
- An annual plan at 690–PLN 790: the inspection, priority call-outs and a discount on parts. Only A and C have one; decide after six months, on the system's data.
- We do not count these price changes in the payback in part 3. The payback rests only on the office's hours and the inspections that lapse today.
Colours: three of the four firms are green
Green says ecology and heat pumps, so almost everyone chose it. We checked what sets the client apart, and whether its buttons can be read.
The colours on their sites
Client
- #4CAF50buttons
- #37474Ftext and ground
Competitor A
- #2E7D32buttons
- #263238text and ground
Competitor B
- #5DAA3Fbuttons
- #424242text and ground
Competitor C
- #1565C0buttons
- #1A2B3Ctext and ground
Proposal
- #4CAF50buttons
- #2F6B3Atext and ground
- #B4532Aactions
Client
Competitor A
Competitor B
Competitor C
Proposal · actions
The client, A and B sit within 20° of hue: from a distance they are one firm.
Data as a table
| Firm | Colour | Hue |
|---|---|---|
| Client | #4CAF50 | 122° |
| Competitor A | #2E7D32 | 123° |
| Competitor B | #5DAA3F | 103° |
| Competitor C | #1565C0 | 212° |
| Proposal · actions | #B4532A | 18° |
Can the button be read: white text on the button's colour
| Firm | Colour | Contrast | Text, 4.5:1 | Large text, 3:1 |
|---|---|---|---|---|
| Client | Aa #4CAF50 | 2.77:1 | fails | fails |
| Competitor A | Aa #2E7D32 | 5.12:1 | passes | passes |
| Competitor B | Aa #5DAA3F | 2.88:1 | fails | fails |
| Competitor C | Aa #1565C0 | 5.74:1 | passes | passes |
| Proposal · actions | Aa #B4532A | 4.98:1 | passes | passes |
| Proposal · text and ground | Aa #2F6B3A | 6.39:1 | passes | passes |
What we propose
- The logo and its green stay: clients know them.
- Text, links and headings in a darker green, #2F6B3A: 6.39:1 on white.
- Buttons and actions in a warm #B4532A: white text on it reaches 4.98:1, and nobody nearby uses it. It also fits what the client sells: heat.
- The new colours go to the request page, the app and the texts. The client can repaint the current site later.
Part 3
Analysis and quote
end of week 1 · after a call, if you want one, and a week of the office's data
Where time and money leak today
For a week the office logged where its time goes. We counted the inspections from the spreadsheet for the last four full quarters.
about €4,500
is what the manual process costs a month today
The office's hours on retyping, calls and booking: €2,800. The margin of inspections that slipped: €1,700.
56 of 80 hours
of the office's week go on retyping, calls, taking requests and booking
22%
of inspections in the last four quarters slipped: 352 of 1,610
about 420 calls
a month asking when the technician will arrive
- today
- after the change
Retyping reports into manufacturers' portals
−11 h
Client calls about a visit's status
−9 h
Taking requests and typing them into the spreadsheet
−4 h
Booking warranty inspections
−7 h
Assigning and moving visits
−3 h
In total 56 → 22 hours: 34 hours a week go back to the office.
Today: a week of the office's log. After the change: our forecast, checked after three months.
Data as a table
| Activity | today | after the change | difference |
|---|---|---|---|
| Retyping reports into manufacturers' portals | 15 h | 4 h | −11 h |
| Client calls about a visit's status | 13 h | 4 h | −9 h |
| Taking requests and typing them into the spreadsheet | 11 h | 7 h | −4 h |
| Booking warranty inspections | 9 h | 2 h | −7 h |
| Assigning and moving visits | 8 h | 5 h | −3 h |
- slipped
- done
Most slipped in Q2 2026: 104 of 447. In spring the technicians are installing, and inspections wait.
The client's inspection spreadsheet, four full quarters.
Data as a table
| Quarter | due | done | slipped |
|---|---|---|---|
| Q4 2025 | 376 | 302 | 74 |
| Q1 2026 | 373 | 292 | 81 |
| Q2 2026 | 447 | 343 | 104 |
| Q3 2026 | 414 | 321 | 93 |
The plan: 4 weeks, a demo every week
The recommended scope from the start of work to launch. At the end of every week you get a working version to click through and a written update.
| Process drawing, screen mock-ups and a written plan | wk 1 | 1 day |
|---|---|---|
| Backend, accounts, roles and the device base, imported from the spreadsheet and the old app | wk 1 | 2 days |
| Office panel: request queue, map and the technicians' calendar | wk 1–2 | 3 days |
| Technicians' app for iOS and Android that works offline: jobs, a report with photos, the client's signature | wk 2–3 | 3 days |
| A request and status page for clients, linked in a text, with a review request after the visit | wk 2 | 1 day |
| Inspections: reminders and self-booking | wk 3 | 1 day |
| Report export in three manufacturers' portal formats | wk 3 | 1 day |
| A pilot with two technicians, store release and office training | wk 4 | 1 day |
Risks, and what this plan leaves out
Every risk has a name, a plan and an effect on the price. The top of the range is all of them at once.
| Risk | What we do | Effect on the price |
|---|---|---|
| The manufacturers' portals have no API and no fixed import format | We export the report in each portal's format, and the office sends it. We add automatic sending where a manufacturer offers it. | +1 day |
| The device spreadsheet has gaps | An import with an error report. The office fills the gaps in the panel before reminders start. | +1 day |
| Technicians ask for changes to the app after the pilot | A pilot with two technicians in the last week, and fixes before the whole team gets the app. | +1 day |
| Clients do not answer the reminder text | After 7 days without an answer the office gets a call list. We measure the reply rate from the first month. | none |
| The AI agent misreads some Polish e-mails (full scope only) | The agent only prepares a draft, and the office approves it. Before launch we test it on 200 real e-mails. | +1 day |
What it leaves out
- Billing clients: it stays in the software you use today.
- A parts shop and online payments.
- Integration with the phone system.
- Moving paper reports from before launch.
- Fixes to the current site: you get them as a list in part 2.
The quote: three scopes, one rate
On a smaller budget we shrink the scope and keep the same rate. We price every scope the same way: days of both founders times the day rate. The range comes from the risks in the previous section.
For reference: the manual process costs about €4,500 a month today.
Start
When you first want to get rid of paper.
price: €4,800
- days of work
- 10
- range
- €4,320–5,760
- to launch
- 3 weeks
- gives back
- €570 / a month
- payback
- in month 10
Recommended
We recommendCloses all four points on the drawing.
price: €6,240
- days of work
- 13
- range
- €5,760–7,680
- to launch
- 4 weeks
- gives back
- €2,900 / a month
- payback
- in month 3
Full
Recommended plus the AI agent, parts stock and a report for the owner.
price: €8,160
- days of work
- 17
- range
- €7,680–10,080
- to launch
- 5 weeks
- gives back
- €3,900 / a month
- payback
- in month 3
| Item | Days | Start | Recommended | Full |
|---|---|---|---|---|
| Process drawing, screen mock-ups and a written plan | 1 | included | included | included |
| Backend, accounts, roles and the device base, imported from the spreadsheet and the old app | 2 | included | included | included |
| Office panel: request queue, map and the technicians' calendar | 3 | included | included | included |
| Technicians' app for iOS and Android that works offline: jobs, a report with photos, the client's signature | 3 | included | included | included |
| A request and status page for clients, linked in a text, with a review request after the visit | 1 | not included | included | included |
| Inspections: reminders and self-booking | 1 | not included | included | included |
| Report export in three manufacturers' portal formats | 1 | not included | included | included |
| A pilot with two technicians, store release and office training | 1 | included | included | included |
| AI agent: e-mails and texts as draft requests to approve | 2 | not included | not included | included |
| Parts stock in the service vans | 1 | not included | not included | included |
| A weekly report for the owner | 1 | not included | not included | included |
| days of work | 10 | 13 | 17 | |
| price | €4,800 | €6,240 | €8,160 | |
| range | €4,320–5,760 | €5,760–7,680 | €7,680–10,080 |
- A day of both founders: €480 net.
- Amounts are net, plus 23% VAT.
- Running costs of the recommended scope: about €105 a month for the server, database, texts and store accounts. You pay the providers directly, on your accounts.
- Our amounts are converted from złoty at the NBP average rate of 28 October 2026 (table 210/A/NBP/2026, EUR 1 = PLN 4.409) and rounded up. Figures from publications and from the example firm's own market stay in their own currency.
Why it costs less than the market
We have no salespeople, project managers or office to keep, and we write the code with AI agents on our own component library. That means fewer days of work at the same day rate, and we read and check every line ourselves.
- the publications' ranges
- our three scopes
CenaUslug.pl · simple app (MVP)
Netguru · simple app, one platform
Chop Dawg · small app, one platform
Groovy Web · simple app at an “AI-first” agency
BizBees Studio · Start
BizBees Studio · Recommended
BizBees Studio · Full
Dollar amounts converted at the NBP rate of 28 October 2026 (table 210/A/NBP/2026): USD 1 = PLN 3.962. A “simple app” in these lists means 3–5 screens on one platform, usually without a backend. The recommended scope covers more: an office panel, an app for iOS and Android, a client page and the portal export.
Data as a table
| Source | Range |
|---|---|
| CenaUslug.pl · simple app (MVP) · May 2026 | PLN 40,000–100,000 |
| Netguru · simple app, one platform · 1 June 2026 | $15,000–50,000 |
| Chop Dawg · small app, one platform · 3 April 2026 | $20,000–35,000 |
| Groovy Web · simple app at an “AI-first” agency · 2026 | $8,000–25,000 |
- Recommended
- Start
- Full
Data as a table
| Quarter | Start | Recommended | Full |
|---|---|---|---|
| launch | −€4.8k | −€6.1k | −€8.2k |
| quarter 1 | −€3.4k | +€230 | +€450 |
| quarter 2 | −€1.8k | +€8.8k | +€12k |
| quarter 3 | −€230 | +€17k | +€23k |
| quarter 4 | +€1.6k | +€26k | +€35k |
| quarter 5 | +€3.2k | +€34k | +€46k |
| quarter 6 | +€5k | +€43k | +€58k |
Why we recommend the middle scope
Recommended and Full both pay back in the same, third month. Full gives more after 18 months (+€58k against +€43k), but its surplus rests on two assumptions we cannot check today: how many e-mails the AI agent reads correctly, and how many second visits come from a missing part. So we propose Recommended now, and Full's modules after three months, on real data and at the same rate.
Terms, and what we need from you
Terms
A fixed price for a defined scope. We quote any change of scope in writing before we make it.
The code, repository, documentation and rights are yours once the final invoice is paid.
We sign a non-disclosure agreement before exchanging sensitive information.
Payment in three parts: 30% at the start, 40% after the technicians' app demo, 30% after launch.
draft term · to confirm
Maintenance and development after launch within the scope we agree on.
What we need from you
- One person who decides. In this example, the office manager.
- The device spreadsheet and a list of the manufacturers' portals, with test access where possible.
- Two technicians for the pilot in the last week.
- An hour a week for the demo and your comments.
Your sketch looks the same, only with your five sentences
One business day, no meeting, no contract. Write how your process works today and how you will know it worked.