How a Solo Tax Prep Practice Took on 60 More Clients With Zero New Hires

She'd been doing this for 14 years. 280 returns a year. Solo. No staff. She wanted to grow but every prior attempt had cost her her weekends and her sanity.

By the time we finished she was preparing 340 returns. Working the same hours. Charging slightly more per return because she'd dropped the bottom tier of clients.

The system isn't fancy. It's three pieces wired together. A solo practice can run this. A 5-person practice can run this. You don't need a 12-person dev team to set this up.

Piece 1: Intake-form qualification

Her old intake was a 2-page PDF emailed back. About 1 in 5 turned into a return. The other 4 in 5 either ghosted, asked for a quote and disappeared, or filled it out and turned out to be the wrong fit (cash-only businesses she didn't take, multi-state filers above her comfort, etc.).

That triage was eating her February.

The new intake is a Tally form. 11 questions. The form output gets piped through Claude with a prompt that scores fit (good fit / borderline / not a fit), estimates effort (simple / moderate / complex), and drafts a personal reply.

Three buckets: - Good fit + simple effort: auto-reply with engagement letter and pricing - Good fit + complex effort, or borderline: route to her for human review, with the AI's draft response prefilled - Not a fit: polite rejection with a referral to a colleague who handles those returns

She spends 20 minutes a week on intake. Used to be 2-3 hours.

The auto-rejection feature was the controversial one. She worried it would feel cold. We A/B tested rejection emails with a small group. The auto-rejections actually scored slightly higher on "felt respected by the response" because they came back within 30 minutes instead of the prior 4-day delay.

Piece 2: Document intake

Most solo practices' worst week in busy season is "I'm waiting on documents from 40 clients and have no idea which 40." She had this.

We built a checklist generator. When a return is opened in her practice management software (Canopy), the system generates an expected-documents list based on prior year filings. The list lives in a shared Notion page that she and the client both have access to. Each line auto-updates from green-checked when a document is uploaded to her secure portal.

The shared list is the part that changed everything. Clients can see what they owe before she has to remind them. Most clients hate getting reminder emails. They love seeing a checklist that they're 80% done with. The visual progress drives self-completion.

When 3 days pass with no movement, a Claude-drafted reminder goes out (personalized per client based on their history). She reviews and sends.

Document-chase time per client dropped from 45 min over a 4-week return to about 8 min.

Piece 3: Drake Tax data entry

This is the dumb mechanical one that just saves time.

Most W-2s and 1099s are clean PDFs. We OCR them, parse the standard fields, and dump them into Drake via a CSV import. The OCR is right about 96% of the time on clean PDFs. The 4% it gets wrong it flags as low confidence so she reviews.

Scanned-from-phone-camera documents are still done by hand because the OCR breaks too often. That's about 20% of her intake. Future improvement.

For the 80% that are clean PDFs, entry time per document dropped from about 8 minutes to about 90 seconds (and the 90 seconds is mostly verification).

What it cost

Tally Pro: $29/mo. Notion Plus: $10/mo. Claude API: about $40-60/mo. OCR service: $25/mo. Drake CSV import templates: free, built by her with my help.

Total: under $130/mo. Less than one new client's annual fee. She picked up 60 new clients.

My time: 6 weeks of build. About 9 phone calls of training. I stopped charging her by week 4 because we'd already crossed the breakeven for that year. She paid me a year-end bonus instead of the back half of my engagement fee. Best deal I've made.

Why this works for a solo

Solo practices have an advantage that large firms don't. You make the call. You set the workflow. You own the standardization.

A large firm has 8 people with 8 different ways of running intake. The change management to align them is brutal. A solo has one way. Their way. Changing it is one conversation with one person.

If you're solo and you've been told "you can't afford AI," you can. The stack above is cheaper than most solo's monthly software bill already.

What's still on the list

Phone calls. She still does all her own intake calls. We've talked about a voice agent answering during her busy hours. She doesn't want one. Her edge is being available. I respect it.

Year-round bookkeeping for the 60 small businesses she preps for. We're piloting a Claude-powered transaction categorization for one of them this fall. If it works she'll add bookkeeping as a service line for the others.

What I'd tell another solo CPA

Start with intake. The intake qualification alone gives you back a week of February.

Add document chase second. Same client value (you respond faster), with operational savings you can measure week by week.

Do the OCR third. It's the smallest individual win per document but it stacks.

You don't need a developer. You don't need a 6-month implementation. You need a weekend, three subscriptions, and a partner who's willing to redo their intake form. The technology is the easy part.