Where to start: the one boring task to automate first
Not sure what to automate first? A one-week, one-sheet audit finds the one task worth it, and the first project we'd suggest for your kind of business.
Founder & Engineer
The winner is usually the most boring task in the office: frequent, done the same way, and visible to a customer when missed. Think the missed call, the reminder, the invoice nobody chased.
Why owners stall on AI
In Thryv’s 2026 outlook, released January 22, 67% of owners expected AI to have no impact on their business this year and only 38% felt they had the tools to grow. A small survey of Thryv’s own clients, but it matches what we hear around Orlando: owners are open, just unsure where to start.
OnDeck and Ocrolus found 56% of small businesses use AI, with 63% of that use in marketing, and Reimagine Main Street found 51% of owners were “Explorers”, 37% lacked the time to explore tools, and 34% saw no clear use case. A one-week audit gives you the time and the use case: task first, tool second.
The one-week sheet
Five columns, one ordinary week, kept by the person doing the work. Owners guess low on how often and high on how long.
| Task | How often | How long each time | Who does it | What breaks when it’s skipped |
|---|---|---|---|---|
| Call back a missed call | 8 to 12 a day | 4 minutes, plus the voicemail | Front desk | The caller books someone else |
| Send the appointment reminder | 8 a day | 1 minute | Front desk | Patient forgets, the chair sits empty |
| Chase an unpaid patient balance | 4 a day | 5 minutes, then a note to self | Owner, evenings | Cash arrives three weeks late |
| Re-type the visit into QuickBooks | 12 a day | 5 minutes | Office manager | Books are wrong until month end |
| Ask for a review | When someone remembers | 2 minutes | Whoever checked the patient out | Nothing visible, which is the problem |
| Write up an estimate for an unusual case | 1 or 2 a week | 40 minutes | Owner | Nothing, it needs the owner anyway |
“How long” includes the voicemail you play twice. “What breaks” settles ties: losing a customer beats losing an afternoon.
Score it, then count the hours
We use Databrain’s three filters. Give each task a 0, 1 or 2 on each and add them up.
Score each task
Frequent and predictableFilter 1
It happens often and you can see it coming. The missed call scores 2; the unusual estimate, 0.
Same steps each timeFilter 2
Fits on an index card for a new hire: 2. “It depends”: 0. Whether a step needs AI or just a rule sets the price.
Costly, visible mistakesFilter 3
A skip loses a customer, a payment or a day, like the missing reminder that leaves a chair empty.
Then multiply how often by how long. A business is a strong candidate when staff spend more than five hours a week on repetitive, predictable tasks. Above, re-typing clears it alone at an hour a day, and the call-backs, texts and invoices add five or six hours. The unusual estimates stay with the owner.
The calculator takes your top task’s times a week, minutes each and hourly cost, and counts a system taking four fifths of the time.
Try your own numbers
Does the task clear the five-hour bar?
Fill it in for the task that scored highest on your sheet. We count a system taking four fifths of the time; a person still handles the odd one.
- Hours a week now
- 2.7 h
- Hours a week a system takes
- 2.1 h
- Hours a year
- 107 h
- Worth a year
- $2,667
- Payback on a $1,500 build
- 6.5 months
Not yet. A template, a checklist or a decision is the right fix for now.
Fifty working weeks a year.
The free fixes that come first
Two calendars plus a booking assistant means double bookings at speed. Databrain’s advice is to fix broken processes first, and we do that together before any build. Most fixes cost only a decision:
- One calendar for the business, with personal views into it.
- One home for each customer record. If a phone number lives in three tools, pick the one that wins.
- One named owner for the inbox or the voicemail, even if it rotates weekly.
- A written rule for the top exception. “If a caller says emergency, we page the on-call dentist” is enough.
Do these during the audit week. The tasks still on the sheet after them are the real candidates.
The first project, by kind of business
The winner is usually the front door. Each price is fixed in writing, with 60 days of free fixes, in accounts in your name.
| Business | Task that usually wins | What we build | Starts at | Roughly how long | The number behind it |
|---|---|---|---|---|---|
| Trades and home services | The missed call | AI receptionist, text-back first | $1,500 (text), phone answering priced in the plan | Two to four weeks | Home services miss about 14% of calls, per CallRail’s 2025 benchmark. |
| Clinics and practices | Reminders | AI follow-up, reminders with confirm and reschedule | $1,500 | Two to three weeks | Healthcare misses about 32% of calls, the worst rate CallRail measured. |
| Brokerages | The web lead that waits | AI receptionist, text-back on every form and inquiry | $1,500 | Two to three weeks | In a vendor’s own 2025 audit, 41% of the top 74 brokerages never responded and 9% answered within five minutes. |
| Restaurants | The phone during the rush | AI receptionist with phone answering, hours and reservations first | $1,500, phone answering priced in the plan | Three to five weeks | The vendor Popmenu’s own figure: 49% of diners are less likely to visit if their call goes unanswered. |
Build times are ours, for one location. Texting customers needs carrier registration, which takes one to three weeks and has been mandatory since February 1, 2025, so we start it on day one. And since 59% of contractors using AI adopt features built into software they already use daily, we may set up the receptionist built into software you already run rather than build one, as in our piece on medical and dental front desks.
When “nothing yet” is the right answer
Sometimes the sheet says wait: the top task is under five hours a week, every instance differs, the exceptions live in your head, or you’re two people who answer the phone anyway. Two hours saved for an hour of babysitting is a subscription, not a system.
Meanwhile, make the fixes, keep the sheet again in your busy season and write the exceptions down. When the hours cross the line, the build will be smaller for it.
Start the sheet on Monday. By Friday you’ll know which task goes first, or that none does yet.
Benjamin J. Ververis
Questions owners ask
What should a small business automate first?
The task you do most often, the same way every time, where a miss costs you a customer. For most local businesses that’s the missed call, the appointment reminder or the unpaid invoice. A one-week sheet finds yours.
Who should fill in the sheet, me or my staff?
The person who does the task. Ask your front desk, office manager or bookkeeper to keep it for the week; their count will be closer than one from memory.
Is a week long enough?
For most offices, yes. Anything done more than five times a week will show up. If your business runs in seasons, keep the sheet during a normal week, not your slowest one.
Do I need to pick a tool before I do the audit?
No. The sheet is about the work, not the software. Once you know which task wins, the tool is usually the one you already pay for, or a small system built around it.
What does the first project cost?
Our smallest first projects start at $1,500, quoted as one fixed price in the written plan before any work begins. The call, the visit and the written plan are free.
Will the first project replace anyone on my staff?
No. The projects on this page take one repeated chore off a person who already has a full day. Your staff still answer the hard calls, approve the exceptions and talk to the customer.
Sources
- Thryv, Cautious optimism for 2026, but uncertainty is high among small businesses (January 22, 2026)
- OnDeck and Ocrolus, via Barchart, Small businesses enter 2026 with sustained confidence, expanding access to capital and growing use of AI (January 2026)
- Reimagine Main Street, Small business AI survey press release (June 10, 2025)
- Databrain, Workflow automation audit: how to find the work worth automating first
- Bushe, AI consulting for small business
- ServiceTitan, AI in the Skilled Trades report (December 16, 2025)
- CallRail, via SkipCalls, The 62 percent problem: contractor speed to lead (citing CallRail's 2025 benchmark) (CallRail benchmark January 2025)
- Roof AI, Lead response time: secret-shopper audit of U.S. brokerages (2025)
- Popmenu, 5 ways AI phone answering can make your restaurant more efficient
- CallHub, 10DLC registration in 2025: the complete guide (2025)
Know an owner this would help?
Talk it through
Have something like this on your plate?
The first call and the visit are free, and the written plan is yours either way.
Where this leads
Written plan