AI Foundations for Small Business · Lesson 9 of 11

Sizing the Opportunity

Time: ~25 min reading + ~20 min worksheet Goal: Turn "this feels useful" into "here's what it's worth." When you can put a number on an AI opportunity, the owner can say yes.

---

SMB owners are pattern matchers. They've seen a dozen consultants pitch "AI solutions" with no numbers. Most of those consultants got politely ignored.

The ones who get hired show up with a realistic number and a realistic plan. Not a perfect number — a *defensible* number. You're not faking precision; you're doing enough math to make the decision easier.

This module teaches you to build that number.

Most SMB AI opportunities are time-savings plays. The basic model is:

Annual value = (hours saved per week) × (number of people) × (hourly loaded cost) × 50 weeks

"Loaded cost" means salary + benefits + overhead. As a rule of thumb:

• Admin/support roles: $25–$40/hour loaded
• Skilled professionals: $50–$100/hour loaded
• Specialized professionals (attorneys, engineers, clinicians): $100–$300+/hour loaded

Example: A tool that saves 4 CSRs 3 hours/week at $35/hour loaded = 4 × 3 × $35 × 50 = $21,000/year.

That's the headline number.

Never use the raw number alone. Adjust for:

Adoption rate

Not everyone will use the tool. Realistic adoption in year 1 is 50–80%. Multiply accordingly.

Time recovery vs. time re-deployment

Not all saved hours become new value. Some saved hours just shift to other work, some disappear into slack.

• Best case: saved hours redeployed to revenue-generating work (rare)
• Mid case: saved hours reduce overtime / avoid hiring (common)
• Worst case: saved hours simply... disappear into schedule slack (often)

Be honest about which one applies. For savings without a clear re-deployment path, discount by 30–50%.

Ramp

Value isn't day-one. Typical ramp: 30% of expected value in quarter 1, 60% in quarter 2, 90% by quarter 3, steady-state at 100% by year 2. Weight accordingly for year 1 ROI.

Time is the easiest to measure. But many AI wins show up as:

Revenue gains

• Higher close rates — Faster, more personalized proposals win more deals. 10% lift on $2M pipeline = $200K.
• More outbound activity — Rep does 30 outbounds/day instead of 15. Pipeline grows.
• Reduced customer churn — Better proactive service, better communication. Pick a retention lift × customer LTV.
• New revenue streams — The AI enables a service the company couldn't offer before.

Cost avoidance

• Avoided hire — Instead of a new FTE at $70K loaded, the AI handles the workload. Pure avoidance.
• Reduced errors — Fewer reworks, fewer write-offs, fewer refunds.
• Lower vendor costs — Sometimes AI replaces a mediocre SaaS subscription.

Quality/strategic gains (harder to quantify, don't skip)

• Faster response times — Customer experience improves. Estimate the effect on NPS and retention if possible.
• Better decisions — Owners have better reports, find issues earlier.
• Employee retention — People like their jobs more when AI takes the tedious parts. Hard to measure, real effect.

For a proposal, pick 1–3 quantifiable benefits and mention the strategic ones as supporting value.

For a real opportunity, build the math like this:

Step 1: Total annual value = time savings + revenue gains + cost avoidance Step 2: Total annual cost = subscriptions + API costs + consulting + internal time Step 3: Net annual value = value − cost Step 4: Payback period = initial investment ÷ net annual value (in months) Step 5: 3-year NPV or cumulative return

Example for a 40-person company:

Project: Custom GPTs for proposal writing + automation for lead triage.

Value:

• Sales reps save 4 hrs/week each, 6 reps, $75/hr loaded → $90K/year gross
• Adoption 70% → $63K
• Year 1 ramp average 65% → $41K in year 1, ~$63K in year 2+
• Close rate +5% on $3M pipeline → $150K new revenue (year 2)

Costs:

• Claude Team for 40 users: $12K/year
• Consulting (initial build): $22K
• Zapier plan: $1K/year
• Ongoing maintenance: $4K/year

Year 1: Value $41K − Cost $39K = +$2K (roughly break-even, expected) Year 2: Value ~$210K − Cost $17K = +$193K Year 3: similar to year 2

Payback: about 12–15 months, conservatively.

That's a clean pitch. Numbers are defensible, ramp is realistic, and you're not promising miracles.

Sometimes a high-value opportunity isn't the right one to tackle first. Readiness matters.

Score each opportunity on:

1. Data readiness

• Clean, accessible data → high
• Scattered, messy → low

2. Technology readiness

• Modern SaaS stack with APIs → high
• On-prem legacy systems → low

3. Workflow readiness

• Documented, repeatable process → high
• Tribal knowledge, exceptions everywhere → low

4. People readiness

• Curious, adaptable team → high
• Change-resistant, stressed → low

5. Leadership readiness

• Owner genuinely in → high
• "We should probably do something about AI" → low

A 5-point checklist. Score each 1–5. Total below 12 = "not yet." 12–18 = "yes with careful scoping." 19–25 = "go fast."

Score every candidate opportunity on two axes: Value and Readiness.

``` High Value │ Revisit │ GO — pitch these │ (need readiness) │ first │ work │ │ │ │──────────────────│────────────────── │ │ Low Value │ Skip │ Nice to have │ │ (later) │ │ └──────────────────┴────────────────── Low Readiness High Readiness ```

Top-right: clear winners — pitch these first. Top-left: get them ready, then pitch. Bottom-right: backlog. Bottom-left: let them go.

A good opportunity assessment almost always surfaces 3–5 top-right candidates. Lead with the one that has the fastest clear ROI — that builds the credibility to propose the rest.

Watch for these when sizing:

• Overestimating time saved. The "5 hours/week" claim is often really 1.5 hours/week once you're honest.
• Ignoring adoption cost. Training, change management, habit formation — these are real.
• Missing integration work. "We'll just connect it to their CRM" hides 40 hours of dev work.
• Forgetting API cost at scale. 10,000 prompts a day adds up.
• Forgetting maintenance. AI projects need tuning. Build in 10–20% of build cost annually.
• Promising perfect output. Review time is real time.

Before you send an ROI pitch, ask yourself: would I invest my own money in this project at these numbers? If the answer is no, go rework the numbers. If it's yes, you're ready to pitch.

---

• Time × People × Rate is the starting model, then discount for reality.
• Include revenue gains, cost avoidance, and quality gains — but quantify where you can.
• Ramp and adoption are the two biggest silent killers of ROI claims.
• Readiness matters as much as value — the right project done at the wrong time fails.
• The opportunity matrix (Value × Readiness) is your map. Lead with top-right.
• Honesty test — would you bet your own money on these numbers? If not, redo them.

---

Next up: Module 10 — How to Run a Discovery Conversation. The meeting that sells the engagement.

---

Time: ~20 min Goal: Put numbers on a real opportunity. Build the ROI muscle.

---

A 50-person insurance agency is considering a Custom GPT + Claude Team rollout to help CSRs write policy change confirmations, renewal reminders, and client emails.

The facts:

• 8 CSRs, each earning $52K salary (loaded cost ~$40/hour)
• They each spend about 6 hours/week on repetitive client emails
• You estimate AI would save them 50% of that time
• You expect 75% adoption in year 1, 90% in year 2
• Claude Team for all 50 users = $15K/year
• Your consulting package (3 Custom GPTs + templates + training) = $14K one-time
• Ongoing tune-ups = $3K/year

Build the math:

Gross annual time savings (hours): ______________

Gross annual dollar value (no adjustments): ______________

Adjusted for adoption, year 1: ______________

Adjusted for adoption, year 2+: ______________

Add ramp adjustment (assume 65% effective for year 1): ______________

Year 1 net: Value __ − Cost __ = __

Year 2 net: Value __ − Cost __ = __

Payback period: ______________

Same agency. Score their readiness (1=low, 5=high):

• Data readiness (client data mostly in agency management system, decent shape): _
• Technology readiness (modern AMS, Outlook, basic Zapier, no legacy blockers): _
• Workflow readiness (CSR processes are mostly documented but with exceptions): _
• People readiness (older workforce, mixed tech comfort, but curious): _
• Leadership readiness (owner enthusiastic, has a budget, wants to start): _

Total: __ / 25

What do you recommend? ______________

Look at your numbers from Exercise 1. Be brutally honest:

• Is 50% time savings realistic for "repetitive client emails"? What's a more conservative number?

______________

• Could adoption actually be lower than 75%? What would knock it down?

______________

• What's the risk that these "saved hours" just become slack instead of redeployed value?

______________

• Would you bet your own money on the net-positive ROI you calculated? Why or why not?

______________

---

Exercise 1 — worked example:

Gross annual hours: 8 CSRs × 6 hrs × 50 weeks × 50% savings = 1,200 hours saved

Gross annual value: 1,200 × $40 = $48,000

Adjusted for adoption (75%): $48K × 0.75 = $36,000 (year 1 ceiling)

Adjusted for adoption (90%, year 2+): $48K × 0.90 = $43,200

With year-1 ramp (65% of adopted value): $36K × 0.65 = $23,400

Year 1: $23,400 − ($15K + $14K + $3K) = −$8,600 (small loss, expected in year 1)

Year 2: $43,200 − ($15K + $3K) = +$25,200

Payback: $32K investment (year 1 cost) / ~$25K annual net year 2 = ~15 months

That's a reasonable pitch. Not a home run — a solid, defensible, boring-sounding good investment. Boring is good. Owners trust boring.

Exercise 2 — readiness:

A realistic score for this client might be something like 4 + 4 + 3 + 3 + 5 = 19/25 — firmly in "go" territory. Leadership enthusiasm is the strongest signal; the mixed tech comfort is the biggest risk to address with training.

Exercise 3:

• 50% is optimistic for "repetitive emails." A more honest number is often 25–40%. Redo the model with 35%: gross value drops to ~$34K, year-1 net goes more negative. Still a reasonable investment but the story is tighter.

• 75% adoption is optimistic. Older workforce + mixed tech comfort = bank on 50–60% in year 1. Plan training budget accordingly.

• High risk in a CSR role. If saved hours don't get re-deployed to more proactive client outreach, cross-sell, or retention work, the savings evaporate. Build the redeployment plan into the engagement.

• If you re-run with 35% savings and 55% adoption, the math gets tighter but still works over 3 years. That's the version you should pitch — ROI that survives honest assumptions. The 50%/75% version is too rosy.

---

Check yourself: Can you build the full ROI model for a new opportunity in about 30 minutes, with assumptions that survive scrutiny? That's the bar for Module 10, where you'll go gather the inputs to build these models live in a discovery meeting.

Previous: Build vs. Buy vs. Configure · Next: How to Run a Discovery Conversation