Software for service businesses is usually sold as a feature checklist.
Operators do not buy checklists.
They buy fewer dropped balls.
I learned that by running Wade’s Plumbing & Septic, growing it to $2.4 million in annual revenue, and eventually losing the company. The failures were not mysterious. They showed up as missed calls, incomplete job information, estimates that did not carry cleanly into the field, costs that appeared after the invoice, customer promises trapped in text messages, and money that looked available before the obligations behind it were understood.
A service company is a chain:
lead → booking → dispatch → diagnosis → scope → approval → work → invoice → payment → follow-up → return
Software earns its place by protecting that chain.
A platform can have 200 features and still fail if the customer’s photo never reaches the technician, the estimate assumptions never reach the crew, the completed work never reaches the invoice, or the invoice never becomes collected cash.
The best field-service software is not the product with the most activity. It is the product that prevents important work from disappearing between people.
That is the standard I use when thinking about Thorbis and the category as a whole.
The phone is still the front door
Home-service software companies like to talk about digital transformation.
The customer often starts with a phone call.
Someone has no hot water, a sewage backup, a leaking ceiling, an HVAC system that stopped during extreme weather, or an electrical problem they do not understand. The customer is not studying the contractor’s software stack. They are trying to reach a competent human being quickly.
Current industry research supports the urgency.
Jobber’s 2026 survey of 1,050 U.S. home-service business owners reports that more than 55% of customers expect a response within an hour and 28% expect one immediately. A 2026 Pied Piper study evaluated 3,211 locations across major home-service brands and found that the most common follow-up experience was no follow-up at all, occurring 72% of the time.
Vendor-sponsored research should not be treated as a neutral census of every contractor. The pattern is still familiar to anyone who has operated the phones: companies spend money generating demand, then lose the customer at the handoff.
CallRail’s small-business benchmark placed the missed-call rate for home services at 14%. Use that as a directional benchmark, not a universal rate.
Consider a company receiving 300 inbound calls per month:
300 calls × 14% missed = 42 missed calls
Now make conservative illustrative assumptions:
- 40% of those calls would have become booked work
- The average first invoice would have been $450
The immediate revenue at risk is:
42 × 40% × $450 = $7,560 per month
Now assume only 25% of the customers who would have been won later produced one additional $450 job:
42 × 40% × 25% × $450 = $1,890 in future revenue
The modeled exposure becomes:
$7,560 + $1,890 = $9,450 per month
That does not include referrals, maintenance, larger future replacements, advertising cost already spent, or the customer who never calls again.
A missed call is not a phone statistic.
It is a broken acquisition funnel, an empty schedule slot, and possibly a lost lifetime relationship.
Software should help the company answer, route, identify, schedule, and follow up on every legitimate lead. Whether the first response comes from an employee, an answering service, or a carefully bounded AI voice system, the next person should inherit the full context.
The customer should not have to start over.
Context has to travel with the job
Most service businesses do not suffer from a complete absence of information.
They suffer from information that cannot travel.
The CSR knows the customer is a tenant and the owner must approve the work. The dispatcher knows access is limited after 3 p.m. The customer texted a photo to an office phone. The technician knows the equipment was repaired six months ago. The estimator included an assumption about excavation. The bookkeeper knows the deposit was short.
Each person can be correct while the company still makes the wrong decision.
A useful field-service record needs to connect:
- the customer and every relevant contact,
- the property,
- equipment or assets at the property,
- the original request,
- calls, texts, email, photos, and voicemails,
- appointments and technician assignments,
- inspection findings,
- estimates and approved scope,
- work performed,
- materials and labor,
- invoices, payments, refunds, and financing,
- warranties, callbacks, and future recommendations.
That does not mean placing everything on one giant screen.
It means the system knows the relationships so each person receives the right context at the right moment.
The technician should see what matters for today’s call. The dispatcher should see capacity and risk. The customer should see clear choices and status. The owner should be able to trace how the job moved from lead to cash.
Unified communication is not useful because one inbox looks cleaner.
It is useful because promises stop disappearing between channels.
The field application has to respect field labor
Technicians work under conditions most desktop product teams never experience.
They use the application with one hand. They have weak signal. They are wearing gloves or have dirty hands. They are standing in front of a customer. They may be trying to photograph a dark mechanical space, read a model number, calculate a repair, and explain the situation at the same time.
Every unnecessary tap has a labor cost.
Every repeated field invites a mistake.
Every slow transition turns into more time in the driveway.
This is why “mobile responsive” is not a meaningful finish line. The product needs a field-specific interaction model:
- clear primary actions,
- large and forgiving controls,
- minimal typing,
- fast customer and property search,
- photo capture tied directly to the job,
- local drafts,
- visible sync state,
- an offline strategy for essential work,
- and the ability to complete paperwork without returning to a desktop.
The correct performance metric is not only a Lighthouse score.
It is the number of seconds and decisions required to finish the job accurately.
Assume ten technicians each spend eight avoidable minutes per day waiting, searching, re-entering information, or correcting software friction:
10 technicians × 8 minutes × 250 days = 20,000 minutes
That is 333 hours per year.
At an illustrative loaded labor cost of $45 per hour:
333 × $45 = $14,985 per year
That is almost $15,000 spent on friction before counting the revenue those hours could have produced.
The software should make a technician more capable.
It should not turn a skilled person into a data-entry clerk.
The estimate has to become the work
Many field-service systems treat an estimate as a sales document.
Operationally, it is the beginning of a contract and production plan.
A strong estimate should carry forward:
- the exact approved scope,
- included and excluded work,
- labor assumptions,
- materials and equipment,
- permits and inspections,
- access and restoration responsibilities,
- payment terms,
- warranty,
- customer selections,
- known unknowns,
- and the conditions that require a change order.
When that information remains trapped in a PDF or a salesperson’s memory, the crew is forced to reconstruct the job.
That creates the worst kind of operational failure: the customer, estimator, office, and technician each believe something different was sold.
The software should turn the approved option into structured work. The crew should know what success means. The office should know what has changed. The invoice should reflect the approved and completed scope. The job-cost report should compare what was expected with what actually happened.
An estimate should not disappear after the customer signs it.
It should become the source of truth for delivery.
Owners need financial truth, not prettier revenue
Service-business dashboards often make revenue the hero.
Revenue matters. It does not answer whether the company is healthy.
An owner needs to know:
- gross profit by job type,
- estimated versus actual labor,
- material committed versus used,
- unbilled work,
- unpaid invoices,
- warranty and callback cost,
- cash expected and cash available,
- payroll and tax obligations,
- financing fees,
- marketing cost per collected job,
- and repeat-customer contribution.
The difference between booked, completed, invoiced, and collected revenue should be impossible to miss.
A job can be sold and never completed.
It can be completed and never invoiced.
It can be invoiced and never collected.
It can be collected and still lose money.
The system should show the movement, not collapse every stage into a celebratory number.
This is where job costing must connect to field behavior. If actual labor runs 40% over estimate, the owner should be able to see whether the cause was a weak scope, dispatch error, missing material, technician training issue, unexpected condition, or unrecorded change.
Reporting without operational context creates another dashboard.
Reporting with context creates a decision.
Customer experience should be an operating system feature
Homeowners do not separate “the work” from “the process” as cleanly as contractors often do.
The plumbing repair can be technically correct while the customer still dislikes the company because nobody answered, the arrival window changed without notice, the technician lacked the original context, pricing was unclear, or nobody followed up.
Housecall Pro’s 2025 survey of 1,040 U.S. homeowners found that 68% would hire the same professional again after an excellent service experience and 73% would refer the company. The same survey found that 68% expected photo or video proof of completed work.
Again, Housecall Pro sells software in this market, so its findings should be understood in that context. The business implication is still strong: customer experience is not a layer added by marketing after the job.
It is produced by operations.
Software should make the trust-building behaviors easy and consistent:
- fast acknowledgment,
- clear arrival communication,
- technician identity,
- documented findings,
- understandable options,
- explicit approval,
- photos of completed work,
- clean invoices,
- warranty visibility,
- and follow-up tied to the real job history.
A good system does not make every interaction robotic.
It makes sure the human interaction begins with the right information.
Automation should remove memory work, not judgment
Service businesses contain hundreds of repeated reminders:
- confirm tomorrow’s appointments,
- follow up on an unsold estimate,
- notify a customer that the technician is delayed,
- request a missing deposit,
- send a completion summary,
- ask for a review,
- remind the office about a permit,
- flag an invoice that passed its due date,
- and schedule recommended maintenance.
These are good candidates for automation because the trigger and next step can be defined.
The danger is automating situations where context changes the correct response.
A long-term customer with a family emergency should not receive the same collections sequence as an unknown commercial account ignoring a signed payment schedule. A high-risk electrical condition should not be summarized like an optional cosmetic recommendation. A technician’s uncertain note should not become a confident customer statement without review.
The Pied Piper study offers a useful warning about automation without dependable handoff. The study found wide variation in how home-service brands answered questions and provided next steps; at many locations, the process failed before a useful human conversation occurred.
AI can help answer, summarize, classify, draft, schedule, and surface risk.
It should also expose:
- the trigger,
- the data used,
- the proposed action,
- the owner,
- the approval requirement,
- and the result.
Automation should reduce the number of things employees have to remember.
It should not remove the company’s ability to understand what happened.
The software has to support return customers
New-customer acquisition gets most of the attention because it is easier to attach to marketing.
Returning customers create a different kind of strength.
They already know the company. The customer record exists. The property history exists. The office has less explaining to do. The technician can see prior work. Price comparison may be lower. The customer is more likely to accept a recommendation because trust was built before the current emergency.
In the plumbing operation I am part of today, 46% of revenue comes from returning customers. That is revenue share, not a claim that exactly 46% of individual customers return.
Software should make that value visible.
The company should be able to understand:
- repeat revenue by acquisition source,
- return rate by technician and job type,
- time between services,
- equipment or property history,
- referrals connected to the original customer,
- warranty versus new revenue,
- and the lifetime gross profit of a relationship.
A platform centered only on today’s average ticket will undervalue the customer who produces years of profitable work.
The customer timeline should not end when the invoice is paid.
The settings should not become the product
Field-service platforms become complicated for a legitimate reason: service businesses are complicated.
The wrong response is exposing every internal possibility as another setting.
Configuration is valuable when the company genuinely needs to express a different policy, trade, permission, workflow, or financial rule. Configuration is harmful when the owner has to assemble basic software behavior before using it.
A good default should reflect how a competent service company normally operates.
Advanced control should appear when the operation needs it.
The system should progressively reveal complexity instead of making every customer understand the full platform on day one.
This matters especially for smaller companies. A two-truck operation needs professional behavior without hiring an implementation team. A 75-person operation needs deeper controls without being trapped by the simple defaults.
Simplicity is not the absence of capability.
It is capability revealed at the right time.
The outcome scoreboard
I would judge field-service software by the business outcomes it changes.
Useful metrics include:
- Answer rate
- Time to first response
- Lead-to-booking rate
- Booking-to-dispatch time
- On-time arrival rate
- Estimate-to-approval rate
- Estimated-versus-actual labor
- Gross profit per field hour
- Callback and warranty cost
- Invoice-to-payment time
- Days sales outstanding
- Repeat-customer revenue
- Referral-generated revenue
- Task completion time in the field
- Number of jobs requiring manual correction
- Percentage of automated actions reviewed or reversed
Feature usage is supporting evidence.
The outcome is the reason the feature exists.
A dispatch board is valuable if it reduces gaps, conflicts, drive time, and customer uncertainty. A communication hub is valuable if it prevents missed messages and repeated explanations. AI is valuable if it reduces administrative load without damaging trust. Reporting is valuable if an owner makes a better decision earlier.
If the feature cannot be tied to an operational result, it may be a demo feature.
Build from the field up
The best product ideas I have had did not come from competitor matrices.
They came from standing in a wet basement while the tool in my hand fought me. They came from an office employee searching for a message. They came from a crew discovering that the scope did not match the site. They came from a large revenue month that did not create enough cash. They came from customers who simply wanted a clear explanation and a company that did what it said.
That is the bar for service-business software.
It should understand that one call becomes a job, one job becomes several obligations, and one customer experience can become years of revenue.
The industry does not need another feature checklist.
It needs software that catches the handoff before the ball hits the ground.