The AI Opportunity Map: How to Find Where AI Actually Pays in Your Business (in One Week)
Most AI readiness assessments produce a maturity model and a roadmap. This one produces a ranked list of your actual workflows with build estimates, and a written list of what not to build. Here is the method, step by step, so you can run it yourself.
- An AI opportunity map scores every workflow on two axes: hours lost a month and weeks to build.
- It takes one week, is done by the engineer who will build the result, and ends with a ranked shortlist plus a written "not worth building" list.
- The inputs are conversations with the people doing the work and a look at the real data, not a questionnaire for managers.
- Build the top-right corner first. Write one sentence per candidate: hours lost today, hours expected after.
- Template included below so you can run a first pass yourself in an afternoon.
An AI opportunity map is a one-week assessment that scores every repetitive workflow in a business on two things: what it costs today in hours or errors, and what it would take to fix. The output is a ranked shortlist of two or three things to build, each with a build estimate and an expected saving, plus a written list of what is not worth building and why. It replaces the maturity models and roadmaps most AI readiness assessments produce with a decision.
This is the method Wope uses in week one for clients who arrive without a spec. We publish it here because it is not secret, and because a business that runs a first pass itself arrives at the real assessment with better questions.
Why most AI assessments do not help
A typical "AI readiness assessment" costs $1,000–$5,000, takes three to six weeks, and produces a document: a maturity score, a capability model, a phased roadmap. The problem is not that it is wrong. It is that nothing in it is a decision. The business still does not know what to build first, what it will cost, or what it will save, and the people who wrote the assessment are usually not the people who would build anything.
The map fixes that by narrowing the question. Not "how ready are we for AI?" but "which of our workflows loses the most hours and would be cheapest to fix?" That question has an answer, and the answer can be checked in thirty days.
The two axes
Axis 1: what it costs you now
Measured in hours a month, from the people who do the work. Not from their managers, who are usually off by a factor of two in one direction or the other. Ask: how many times a week do you do this, how long does it take, and how often does it go wrong. Multiply out. Add a loaded hourly rate if you want the number in dollars; for most small businesses $30–$60 an hour is realistic.
Errors count too. A workflow that takes four hours a month but produces a wrong quote once a quarter may cost more than one that takes forty.
Axis 2: what it would take to fix
Measured in weeks of senior engineering time. The rough bands we use:
- One week: inputs are structured, the output is a draft a human reviews, and the systems involved have decent APIs. Example: call recording into CRM fields.
- Three weeks: the answer has to be found in documents, so it needs retrieval and a test set. Example: repeat support questions.
- Six weeks or more: a new integration has to be built, the data does not exist in usable form, or the workflow is ambiguous enough that "done" is hard to define. Example: demand forecasting on two years of inconsistent sales data.
The honest version of this axis requires an engineer who has built these before. That is the main reason the map is done by the person who will build the result rather than by a consultant: the estimates are promises, not guesses.
The one-week method
- Monday: list. Every repetitive workflow anyone can name. Thirty-minute calls with four to six people who do the work, not the people who manage it. Aim for ten to twenty candidates.
- Tuesday to Wednesday: look at the real data. The ticket archive, the document store, the CRM, the call recordings. Deduplicate. Find the contradictions. This is where "three weeks" becomes "six weeks" or, happily, "one week".
- Thursday: score. Each candidate gets hours a month and weeks to build. Plot them. The top-right corner (most hours, fewest weeks) is the shortlist.
- Friday: write it down. Two or three items with one sentence each: "Support spends 62 hours a month on repeat questions; retrieval over tickets and docs should take 40 of them; three weeks." And the bottom-left list, with the reason for each: "Demand forecasting: real value, but sales history is too thin to beat the spreadsheet. Revisit in twelve months."
- The following week: a 45-minute call. You pick one. The first commit lands.
A worked example
A 40-person services company, no engineers on staff. Six workflows scored:
| # | Workflow | Hours lost / month | Weeks to build | Verdict |
|---|---|---|---|---|
| 1 | Repeat support questions | 62 | 3 | Build first |
| 2 | Quote and proposal drafting | 28 | 2 | Build second |
| 3 | Sales notes into the CRM | 19 | 1 | Build in week one of building; cheap win |
| 4 | Demand forecasting | High (but uncertain) | 9+ | Not yet; data too thin |
| 5 | Invoice data entry | 6 | 1 | Buy a tool instead |
| 6 | Staff onboarding docs | 5 | 3 | Not worth it |
Item 1 was built in three weeks and took support's repeat-question time from 62 hours to under 25. Item 3 shipped alongside it because it was a week of work. Item 2 started in month two. Item 4 is in the diary for next year. Items 5 and 6 were never built, and nobody misses them. You can see the same map rendered on the Wope home page.
Want the map done properly, by the person who will build the result?
Week one of a Wope subscription is the opportunity map. If it finds nothing worth building, we say so in writing and refund the month. Plans from $1,200 a month, cancel any time.
Start with the map →A template you can use this afternoon
Copy this into a spreadsheet. One row per workflow.
| Column | What to put in it |
|---|---|
| Workflow | One line, in the words of the person who does it |
| Who does it | Names, so you know who to ask and who will be the first user |
| Times per month | From them, not from their manager |
| Minutes each | Honest average, including the fiddly cases |
| Hours per month | The two above, multiplied |
| Error cost | What a mistake costs and how often one happens |
| Inputs | Where the information comes from: email, documents, a system, a call |
| Output | What "done" looks like: a draft, a record, a sent message, a decision |
| Weeks to build | 1 / 3 / 6+ using the bands above; mark it as a guess until an engineer checks |
| Never-alone actions | Anything the AI must always hand to a human (money, prices, commitments) |
Sort by hours per month, then by weeks to build. The first row that is one or three weeks is where to start. If the first row is six-plus, that is usually a sign the data needs work before the AI does, and that is worth knowing before anyone quotes you for a build.
Three mistakes to avoid
- Scoring from the org chart. Managers know what they think happens. The map needs what actually happens.
- Treating "weeks to build" as a formality. It is the axis that decides whether the first project pays back in month two or month nine. Get an engineer to check it.
- Skipping the bottom-left list. Writing down what you will not build, and why, is what stops the same idea coming back every quarter. It is also the part consultants leave out because it shrinks the follow-on work.
Frequently asked questions
What is an AI opportunity assessment?
A short, structured review of a business's workflows to find where AI would save the most hours or errors for the least build effort. A good one ends with a ranked shortlist, build estimates, and a list of what is not worth building.
How long should an AI readiness assessment take?
One week for a business under 300 people. Longer assessments accumulate slides, not insight. At Wope the opportunity map is week one of the subscription, not a separate engagement.
How do you prioritise AI use cases?
Score each workflow on two axes: hours lost a month (ask the people doing the work) and weeks to build (one for structured drafts, three for retrieval over documents, six or more for new integrations or missing data). Build the top-right corner first.
What does an AI opportunity assessment cost?
Consultants charge $1,000–$5,000 for a fixed-scope assessment. At Wope it is included in the first month on any plan (from $1,200), and the same engineer who scores the map builds the first item.
What if the assessment finds nothing worth building?
Then it has done its job. Wope puts that in writing and refunds the month. Finding out in a week beats finding out after a $40,000 build.
- Pareto Talent, "AI Consultant for Small Business (2026)" — on what a 30-day engagement should produce
- Dorian Media Group, "AI Consultant Cost in 2026" — fixed-scope assessment pricing
- Worked example figures are from a Wope client engagement, rounded.