Enterprise DNA

Omni by Enterprise DNA

Enterprise DNA Resources

Step-by-step how-tos. Practical AI operating-system thinking for owners, operators, and teams doing real work.

220k+

Data professionals

Omni

AI agents and apps

Audit

Map the manual work

Guide Intermediate Omni Ops

Stop Losing Money on Change Orders

Stop technicians doing unpaid extra work with real-time change order prompts, photo proof, and customer approval before work begins.

Sam McKay |
Stop Losing Money on Change Orders

The extra work problem isn’t usually a pricing problem

A plumbing, HVAC, electrical, or roofing business rarely loses money on change orders because the owner doesn’t know what labour and materials cost. The loss happens much earlier, on site, when a technician finds something unexpected and starts fixing it.

The customer says, “While you’re here, can you look at this too?”

The technician opens a wall and finds damaged pipework. An HVAC tech sees that a failed part has caused a second issue. An electrician discovers that the existing panel can’t safely support the new circuit. A roofer finds rotten decking under the shingles.

The work needs doing. The crew wants to help. The schedule is tight. So someone says, “Take a few photos and we’ll sort it out later.”

Later is where the margin disappears.

The photos stay on a technician’s phone. The job notes say “extra work completed.” The office hears about it after the invoice has gone out, or after the crew has already left. The customer doesn’t remember agreeing to another $850, $2,400, or $6,000 in work. The owner is then forced into an awkward decision: bill it and risk a dispute, discount it, or write it off completely.

For trades businesses doing $1 million to $25 million in annual revenue, change order leakage often sits inside a wider $50K to $200K annual band of lost or delayed revenue. Not every dollar is recoverable. Some work genuinely should be goodwill. The issue is that most firms cannot clearly separate deliberate goodwill from undocumented chargeable work.

That is an operations problem. It can be fixed without asking your technicians to become admin staff.

What a missed change order looks like on a real job

Consider a service plumber called out for a leaking bathroom supply line. The booked work is straightforward. During the visit, they identify corrosion on nearby fittings and a shutoff valve that is close to failure.

There are three possible outcomes.

The first is the ideal version. The technician explains the issue, takes photos, sends a clear scope and price, gets approval, then completes the added work. The customer sees exactly what changed and why. The office invoices with evidence attached.

The second is common. The technician calls the office, waits for someone to pick up, then proceeds because the customer verbally says it is fine. The admin team later tries to reconstruct the scope from texts, call notes, and a half-complete job record.

The third is expensive. The technician does the work to prevent a further problem, tells the customer the office will follow up, and closes the job. Nobody creates the change order. The added fittings, labour, vehicle time, and warranty exposure become a cost against the original job.

This does not mean technicians are careless. In most businesses, they are responding to the incentives in front of them. They are measured on fixing the problem, getting to the next call, and keeping the customer happy. Asking them to stop, find the right form, calculate pricing, call an owner, wait for approval, and upload evidence can feel impossible during a busy day.

A good process has to meet the technician where the work happens, on a phone, in a text thread, or inside the dispatch app they already use.

Why “we’ll document it later” doesn’t work

Change order processes tend to break in four places.

First, the technician does not recognize that a job has moved beyond the original scope. This is especially common with smaller additions. Replacing a damaged connector, repairing concealed wiring, or adding material that costs a few hundred dollars can seem too minor to escalate. Repeated across hundreds of jobs, those “small” additions become a significant number.

Second, the evidence is weak. A blurry before photo, no photo after completion, and a note such as “customer approved” gives the office little to work with. It also does not give the customer confidence when the revised invoice arrives.

Third, approval happens informally. A customer might agree verbally to investigate a fault, but not agree to the full repair price. The gap between permission to diagnose and approval to proceed is where disputes form.

Fourth, the office does not see the issue until after the crew has completed the work. At that point, leverage has gone. The customer has the benefit, the technician is elsewhere, and nobody can clearly explain what was agreed.

Paper forms and mandatory office calls can improve this, but they often fail under field pressure. A more reliable approach is to automate the prompt, not just the paperwork.

How an AI change order agent works on site

An AI agent for change order capture should not independently price work, promise outcomes, or pressure customers into approving repairs. Its role is practical. It identifies the moment where a change is likely, gathers the right information, prepares a customer-ready approval request, and keeps the office informed.

Here is what the end-to-end workflow can look like.

A technician is at a booked HVAC maintenance visit. They find a cracked condensate line and water damage that was not in the original scope. Instead of calling the office, they use a short prompt on their phone or send a message to the operations workflow.

The agent asks only for what is needed:

  • What did you find?
  • Is the added work safety-critical, urgent, or optional?
  • Can you add two to four photos?
  • What is the recommended repair?
  • How long will it take?
  • Is there a standard price book item, or does the office need to price it?

The technician can answer with voice notes, short text, or photos. The agent turns that raw field input into a structured record. It links the record to the customer, job number, technician, and original scope.

If the repair fits an approved price book item, the system can prepare a plain-language request for the customer. For example:

During your scheduled maintenance, we found a cracked condensate line causing water to leak near the unit. We recommend replacing the damaged section today to prevent further water damage. The additional work is $475. Reply APPROVE to proceed, or call us with any questions.

The customer receives the message with annotated photos where appropriate. The agent records their reply and alerts the technician. No work starts until the required approval is captured, except where a genuine safety protocol calls for immediate action.

If the work is outside standard pricing, the agent routes the request to an estimator, service manager, or owner. It can include the technician’s notes, photos, likely labour time, and a suggested urgency level. The approver does not need to chase information across five different apps.

Once approved, the agent updates the job record, creates the change order, and sends the technician a simple confirmation. After the work, it prompts for completion photos and a short summary of what was done. That record then follows the job through invoicing.

This is where Omni Ops becomes useful. The aim is not a chatbot for its own sake. It is a workflow that removes the delay between discovery, customer communication, approval, and billing.

The right rules matter more than the AI label

The best automation is built around your operating rules. Before introducing an agent, decide what must happen in each category of extra work.

Set a clear threshold. You might require digital approval for every amount above $150, or for any addition that changes labour time, materials, safety exposure, or warranty terms. The exact threshold varies by trade and job type. The point is to stop leaving it to individual judgement every time.

Define emergency exceptions. If a gas safety issue, exposed live wiring, or an active flood needs immediate containment, the crew must act. The system should document the reason, notify the customer, and flag the exception to the office. It should not create a dangerous delay in the name of process.

Use trade-specific language. A homeowner may understand “replace failed shutoff valve” more readily than internal item codes. A commercial property manager may need a purchase order field, site contact name, and formal quote attachment. Your workflow should reflect both.

Give technicians a fast path. If it takes more than a minute or two to submit the evidence and trigger the approval, adoption will fall. The workflow should remember job context and avoid asking for details already in the dispatch system.

Escalate silence. A customer who has not responded in 15 minutes should not leave a technician standing in a driveway indefinitely. The agent can send a reminder, offer a call option, and notify the office. A manager can then decide whether to reschedule, complete only the approved scope, or authorize a limited goodwill fix.

These are the types of decisions we map in the AI audit for trades businesses. The technology comes after the rules, handoffs, and commercial boundaries are clear.

Change orders connect to the rest of your revenue system

It is tempting to treat leakage as a single field-service issue. It rarely is.

If the office is overwhelmed taking calls and moving crews, job documentation gets pushed down the list. An owner who spends 20 or more hours each week dispatching, checking schedules, and chasing updates will struggle to review every extra-work request before it becomes an invoice problem.

The 24/7 Dispatch Voice Agent can take pressure off that front end. It answers incoming calls, identifies an emergency versus a schedulable job, books an appropriate slot in the dispatch tool, and texts confirmation to the customer. That does not replace your dispatcher. It prevents the phone from dictating every minute of their day.

A better dispatch process also improves change orders because the office has cleaner job data from the start. The right customer contact, service address, original job description, appointment window, and technician assignment are already attached to the record.

The same pattern applies after the job. The Estimate Follow-Up Agent tracks estimates and follows up at sensible intervals, including day 2, day 5, and day 14. In many trades firms, follow-up on stale estimates can convert a meaningful 15% to 25% of opportunities that would otherwise sit untouched. That is revenue you have already paid to generate.

The Review and Reactivation Agent handles another common gap. It asks satisfied customers for a review the day after a completed job, then reactivates prior customers at the relevant service interval. A well-documented change order supports both workflows. It means your team knows exactly what was repaired, what may need future inspection, and what the customer agreed to.

You can see how these workflows fit together in Omni, or browse our practical AI guides for other operational bottlenecks.

What to measure in the first 90 days

Do not judge a change order workflow by how many messages it sends. Measure the commercial outcomes.

Start with the percentage of jobs where technicians identify extra work. If it suddenly rises after introducing a prompt, that may mean the work was always there but was not being recorded.

Then track approval rate. Break it down by trade, job type, dollar band, technician, and how quickly the request reached the customer. A low approval rate may point to poor scope descriptions, slow response times, unrealistic pricing, or customers receiving the message after they have become frustrated.

Track approved value versus invoiced value. This catches the second form of leakage, where approval exists but the final invoice does not include the change order.

Review cycle time from discovery to approval. On urgent service work, a long delay can destroy the practical value of the system. For larger replacement or commercial work, a longer approval process may be normal. Your targets should reflect that difference.

Finally, sample the customer experience. Read the messages. Look at the photos. Ask whether customers understood what was found, what was recommended, and what they were being asked to approve. Clean documentation is not just a back-office control. It is part of trust.

If you want an adjacent process to tighten while you work on field approvals, use the After-Hours Call Recovery Plan for Trades as a practical checklist. You can also get the direct worksheet download for mapping who responds, how quickly, and what happens to jobs that arrive when the office is closed.

Find the money before you buy more software

You may already have a dispatch platform, price book, CRM, and accounting system. The answer is not automatically another disconnected app.

Start with a review of the last 30 to 60 days of completed jobs. Pull jobs where actual labour, materials, or technician notes suggest work beyond the quoted scope. Compare those records with approved change orders and invoices.

Look for patterns:

  • Which technicians regularly identify extra work but rarely submit a change order?
  • Which job types generate the most undocumented additions?
  • How often are customer approvals held in texts, personal phones, or verbal notes?
  • How much time passes between a technician finding the issue and the customer receiving a price?
  • Which approved items fail to reach the invoice?

That gives you a baseline. It also tells you whether the primary constraint is technician adoption, office pricing speed, customer communication, or invoice handoff.

A 60-minute Omni Audit produces three useful outputs: a map of the revenue leaks in your current workflow, a shortlist of agent opportunities ranked by value and effort, and a practical next-step plan for implementation. There is no slide deck performance. We work through the jobs, systems, and handoffs that are costing you money.

If undocumented extra work is showing up across your service calls, Book a 60-min Omni Audit. We will identify where real-time photo, text, and approval prompts can protect your margin without slowing your crews down.

Stop asking crews to remember the process

Technicians should not need to remember every administrative step while managing a customer, a technical diagnosis, safety requirements, and the next job on the board.

Build the process so the right prompt appears at the right moment. Make the customer approval clear. Link the evidence to the invoice. Give the office an escalation path when pricing or urgency needs human judgement.

That is how you turn extra work from an uncomfortable recovery conversation into a documented commercial decision before the work begins.

For a broader view of where automation can reduce lost calls, missed follow-up, weak job records, and unpaid change orders, see Omni for trades businesses. When you are ready to map it against your own jobs and systems, Book my Omni Audit.