# A sample AI opportunity assessment Canonical page: https://opconsulting.ai/sample-assessment.html **This is a fictional sample.** The business in it, Cedar and Stone Landscapes, is invented to show the method. It is not a real client, and it reports no real results. This shows what an AI Opportunity Assessment covers and what it asks of you, before you buy. It answers one question: what do I get, and what does it ask of me. The method is real. The business is made up. The weight sits on the parts that do not change from client to client: the questions I ask, the rule for what to automate and what to leave alone, the data boundary, and the stop or go call at the end. - Fixed scope, fixed price - Quoted after the first call - Written answer about two weeks after the call ## Who this is for A local, owner-led service business that runs on multi-party coordination across jobs, crews, vendors, clients, and schedules, where a dropped handoff costs money, a client, or a callback the owner eats. The owner is the system holding it together, and there is no tech team to build the fix. Not for you if you run a solo shop with few moving parts, have an in-house AI or data team, or just want marketing or a website. If the fit is not there, I say so on the first call, and you will not have wasted it. ## The business and the workflow I examined Fictional example: Cedar and Stone Landscapes. Owner-led design and build. One owner, a small field crew, two regular subs for irrigation and masonry, and a handful of suppliers. The owner sells the job, schedules it, orders materials, dispatches the crew, keeps the client posted, and sends the invoice. It all runs through one phone, one inbox, and one spreadsheet. I looked at one workflow, end to end: accepted quote to finished job to paid invoice. Not the whole business. One chain, the one with the most handoffs. 1. Quote accepted, job gets a start week 2. Materials ordered for that week 3. Crew and subs booked around delivery 4. Client told when the crew is coming 5. Job finished, walked, invoiced Each step hands off to the next. Every handoff is a place the ball can drop. ## The six questions Six questions drive the assessment. The same six every time. 1. What repeated workflow is failing, and who owns it? 2. What is the observable before-state: time, delay, error, missed follow-up, service level, or cost? 3. Which people and systems touch the workflow? 4. What data could go into a tool, and what stays out? 5. What does a usable result look like, and who signs off on it? 6. What is the stop or go date, and what evidence supports doing more? For Cedar and Stone, the owner answered all six on the first call. That is a good-fit signal. When an owner cannot name the failing workflow or who owns it, that is the finding. Fix that first, before automating anything. ## Where the work breaks I describe breakage by how often it happens and what it costs when it does. No made-up numbers. A real assessment attaches your numbers. This is the shape. - Materials and crew fall out of sync. A supplier slips a delivery, and the crew shows up to a job with nothing to install. The owner loses a crew-day and scrambles to fill it. - The client hears about a delay late, or from the crew instead of the owner. The owner spends the evening on an apology call. The client trusts the next date less. - Follow-up quotes go out slow, or not at all. The owner is in the field, and warm leads sit in the inbox. Some go cold. What a real assessment measures here: how often each one happens over a set period, and what each one costs in a crew-day, a callback, or a lost job. Measured from the owner's own calendar, inbox, and invoices. Not a survey. That before-state is the baseline the work gets judged against later. ## What doing nothing costs Stated as how often and what it costs, because a made-up dollar figure is worse than none. The material-and-crew miss is the expensive one. It does not hit every job. When it hits, it burns a paid crew-day and a client's confidence at once. The late client update costs no cash directly. It is why a client does not call for the next project. The slow follow-up is missed revenue that never shows up as a loss, because the job was never booked. Doing nothing has a cost. The owner just absorbs it, on evenings and weekends, and has stopped counting it. ## What to automate first, and what to leave alone This is the core of the assessment. The rule: automate a handoff when it happens often, follows a stable rule, and costs something clear when it drops. Leave it alone when it is rare, needs judgment, or a wrong automated move costs more than doing it by hand. ### Automate first: material-readiness check and client status update Both happen every job. Both follow a stable rule. Has the material for Monday's job landed, yes or no. If a date moves, tell the client the same day. Both cost something clear when missed. A small system watches supplier confirmations against the schedule, flags a job at risk before the crew rolls, and drafts the client update for the owner to send. ### Leave alone, for now: change-order pricing on site A client asks for something extra mid-job, and the owner prices it on the spot. That call runs on judgment about the client, the crew's remaining time, and the relationship. Low volume, high judgment. A tool that auto-prices a change order could quote a number the owner never would have, in front of the client. That costs more than the few minutes it takes by hand. Leave it human. Revisit only if change orders get frequent enough to need a set rule. That second call is not a hedge. An assessment that finds nothing to leave alone is selling, not assessing. ## The non-AI alternative I weighed Before recommending any AI, I check whether a plainer fix does the job. For the material-readiness check, a shared calendar rule and a standing Friday review would catch some misses. It helps. It still leans on the owner remembering to run it in a busy week, which is the thing that fails now. So the fix is a light system that runs the check without the owner having to remember. For the client update, a saved template covers the wording. The part worth automating is the trigger, knowing a date moved, not the typing. I recommend AI or automation only for the part a plainer tool does not cover. Where a template or a calendar rule is enough, use that and spend nothing more. ## Data and provider boundary What goes into a tool, and what does not. - **In:** job schedules, supplier confirmations, client names with the job address and status. - **Out, unless you decide otherwise case by case:** client payment details, and anything a client shares in confidence. Before any client information goes to an AI provider, I explain the sharing, the purpose, the provider, and how long it is kept, and you approve it. Then it gets written down. Providers get named, not hidden. This matches the practice in the site's privacy notice (https://opconsulting.ai/privacy.html). It is a working boundary for the engagement, not a compliance certificate. ## Risk level and human review The material-readiness flag is low risk. It tells the owner something. The owner acts. Nothing leaves the business on its own. The client status update is consequential. It goes to a client. So it gets drafted, not sent. The owner reads it and sends it. Nothing reaches a client without a person choosing to send it. Anything consequential gets a human review first. That holds every engagement, not just this one. ## Recommended sequence The four stages, on this business. 1. **Assess.** This document. One workflow mapped, breakage found, a first move and a leave-alone picked. 2. **Prioritize.** Build the material-readiness check and the client-update draft first. Nothing else yet. 3. **Build.** Connect the schedule and supplier confirmations, set the at-risk flag, draft the update for the owner to send. Test it on real jobs before trusting it. 4. **Embed.** Write down how it works, show the owner and one crew lead, and re-measure the before-state to see if the misses actually dropped. ## Operating assumptions A few things have to be true. The owner takes part. The schedule and supplier confirmations are reachable in one place. One person owns the workflow. If the schedule lives only in the owner's head, the first job is to get it written down, and the assessment says so. ## Stop or go At the end you get a straight recommendation and a decision point. Not a contract you have to sign to learn anything. - **Go:** build the material-readiness check and the client update, measure the misses over a set period, and expand only if the misses drop and the owner gets evenings back. - **Stop:** the misses were rarer than they felt, or a plainer fix already handled it. Then the honest answer is to spend nothing more. What you provide: a kickoff call, one working session, and read access to the one workflow's records. Schedule, inbox, and invoices for that workflow. Nothing outside it. The first call: you bring one workflow. Together we decide whether it is even worth assessing. If it is not, I say so. That is a fine outcome. ## Want this for one of your workflows? Bring the one part of the work that gets stuck. We decide together whether it is worth assessing. Fixed scope and fixed price, quoted after the first call. - Book a paid discovery session: https://calendly.com/justin-opconsulting/discovery - Email: info@opconsulting.ai Cedar and Stone Landscapes is a fictional example. Not a real client, no real results.