The Zero-Integration Dream: Running Your Practice on Beancount Alone (No QuickBooks, No Xero, No Subscriptions)

General
1

I’ve been running Martinez Bookkeeping Services for 10 years now, and I’m at a crossroads that I think a lot of you might relate to. My SaaS costs are up 8% this year alone—QuickBooks Online raised rates again, my receipt scanning app went up $15/month, my client portal subscription doubled. I’m now paying close to $400/month across five different tools just to run my bookkeeping practice.

So I’ve been thinking: What if I just… stopped? All of it?

What if I ran my entire practice on Beancount alone—no QuickBooks, no Xero, no monthly subscriptions eating into my margins? Pure plain text accounting for every client, with Fava for client portals, Python scripts for anything custom, and Git for backups. Near-zero software costs, complete data ownership, infinite customization.

The Dream Scenario

Here’s what I’m envisioning:

Client ledgers: All clients get their own .beancount files, versioned in Git (private repos per client). Complete transaction history in plain text.

Client access: Self-hosted Fava instances (maybe $30/month for a VPS hosting all clients). Each client gets their own view—they can check balances, run queries, see their P&L anytime.

Bank feeds: Custom CSV/OFX importers in Python. Most banks export cleanly enough. I’d run imports weekly or monthly, depending on client needs.

Invoicing: Either plain text invoice generation (Beancount has this built-in) or keep my $10/month invoicing tool—that’s the only subscription I might keep.

Reports: Custom queries for everything—P&L, balance sheets, cash flow statements, tax schedules. I’d build a library of standard reports my clients need.

Backups: Git push to private GitHub repo. Every change tracked, every version recoverable. Way better than hoping my cloud accounting vendor doesn’t lose data.

Total monthly cost: Maybe $40/month (VPS hosting + GitHub private repos), down from $400/month. That’s $4,320/year back in my pocket.

The Reality Check

But then I think about the challenges:

Client education burden: How do I explain plain text accounting to a restaurant owner who barely understands QuickBooks? Do I even try, or do I just handle everything behind the scenes?

No phone support: If something breaks, I’m on my own. No QuickBooks support line to call. Just community forums and my own debugging skills.

Missing features: Payroll integration, payment processing, automated bank feeds—these require third-party tools. How do you handle payroll when you’re Beancount-only? Do you partner with a payroll provider and import their data?

Time investment: Building importers, writing reports, troubleshooting Python scripts—that’s time I’m not billing clients. Is the upfront investment worth it?

Professional perception: Does going Beancount-only make me look like a scrappy solo operator versus a “real” bookkeeping firm? Or does it signal I’m technically sophisticated and privacy-focused?

What I’m Really Asking

For those of you running professional practices (or thinking about it):

  1. Is zero-integration realistic in 2026? Or am I being idealistic?

  2. What client types work best? Tech-savvy founders? Privacy-conscious professionals? Cost-conscious small businesses?

  3. What features would you miss most? If you went Beancount-only, what commercial software features would actually hurt to lose?

  4. What’s the minimal tool set? Are there 1-2 commercial tools you absolutely must keep (payroll? payment processing?), and how do you integrate them with Beancount?

  5. Have you done this? Anyone here actually running a zero-subscription practice? How’s it going?

I’m not trying to be anti-commercial-software. QuickBooks is great at what it does. But I’m watching my costs creep up year after year, and I’m wondering if there’s a different path—one where I own my tools, control my data, and keep more of what I earn.

What am I missing? Talk me into it—or talk me out of it.

2

I’ve been down this exact road, so I can share what actually happened versus what I hoped would happen.

The short answer: It’s not zero-integration, but it’s minimal-integration, and yes, it works.

What I Actually Did

Four years ago I was running my personal finances and two rental properties across QuickBooks Online, Mint, Personal Capital, and a hodgepodge of spreadsheets. Total cost: ~$300/month when you include QuickBooks, data aggregation services, and premium features. I went cold turkey on all of it and moved everything to Beancount.

What I kept:

  • My bank and credit card accounts (obviously)
  • A simple invoicing tool for the rentals ($8/month—honestly could eliminate this but tenants are used to it)
  • That’s it

What I built:

  • CSV importers for each bank/card (took a weekend to get 80% working, another month to get to 95%)
  • A custom Fava setup that I self-host on a $5/month VPS
  • Python scripts for recurring transactions (rent income, mortgage payments, subscriptions)
  • Balance assertions on the first of every month for every account

Total monthly cost: $13/month, down from $300+

What Works Better Than Expected

  1. Version control is incredible. I can see exactly when I made a mistake, revert it, and understand my reasoning (via commit messages). Try doing that in QuickBooks.

  2. Client education is easier than you’d think. I don’t explain plain text accounting. I just say “I use a custom system that gives you a secure web portal.” They log into Fava, see their numbers, and that’s all they care about. The tech stack is invisible to them.

  3. No vendor lock-in anxiety. I sleep better knowing my financial data isn’t trapped in some company’s database. If Intuit decided to triple their prices tomorrow, it wouldn’t affect me at all.

  4. Customization is limitless. Need a custom report? Write a query. Need to track something unusual? Add metadata. Try doing that in QuickBooks without paying for “Advanced” tier.

What’s Harder Than Expected

  1. Payroll is the killer. I don’t run payroll for clients (I refer them to specialists), but if you need to handle payroll, you WILL need a service like Gusto or ADP. The compliance burden is too high to DIY. Import their data, but you can’t avoid the subscription.

  2. Bank feeds aren’t automatic. I spend 20 minutes on the first Monday of every month downloading CSVs and running importers. It’s not painful, but it’s not zero-touch either. For client work, this becomes part of your monthly close process.

  3. The upfront time investment is real. It took me probably 40 hours spread over three months to build a robust setup. That’s 40 hours I wasn’t earning. But now? I save 5-10 hours a month not fighting with software, so ROI was around 6 months.

  4. You become the support line. When something breaks—and it will—you’re debugging Python scripts at 11pm. There’s no phone number to call. Community forums are helpful, but you need baseline technical comfort.

Client Types That Work

Great fits:

  • Tech founders who value data ownership
  • Privacy-conscious professionals (lawyers, therapists, consultants)
  • Cost-conscious small businesses (they appreciate saving money on software)
  • Anyone with straightforward accounting needs (simple P&L, basic balance sheet)

Bad fits:

  • Businesses with complex inventory (stick with QuickBooks or NetSuite)
  • Companies needing integrated payroll + benefits + time tracking
  • Clients who demand “industry standard” software for audit/lending purposes
  • Anyone expecting 24/7 phone support

My Honest Take

Is it realistic? Yes, but not for everyone.

If you’re technical enough to write Python scripts, comfortable self-hosting, and serve clients with straightforward needs, this absolutely works. The cost savings are real, the control is real, and the lack of subscription anxiety is wonderful.

But if you’re running payroll for 15 clients, dealing with complex inventory, or serving clients who need QuickBooks files for their bank, this probably isn’t the path.

For me, going minimal-integration was the best business decision I’ve made. I own my tools, control my data, and pocket an extra $3,500/year. But I’m also the person who enjoys writing importers on weekends, so YMMV.

The minimal tool set I recommend:

  • Beancount + Fava (core ledger + UI)
  • Payroll service IF needed (Gusto, ADP—no way around this)
  • Maybe a simple invoicing tool IF clients expect it (Wave is free, or $10/month for features)
  • Everything else: build it or skip it

Happy to share my importer scripts or Fava setup if anyone’s interested. The initial investment is front-loaded, but the ongoing freedom is worth it.

3

This resonates hard. I’m not running a bookkeeping practice, but I’m obsessively tracking my FIRE journey, and I’ve been experimenting with a similar “zero-subscription” philosophy for personal finance.

My stack: Beancount for everything, Fava for dashboards, Python for importers and custom reports, GitHub for backups. Total monthly cost: $0 (I use free GitHub private repos and don’t even self-host Fava—I just run it locally when I need to review).

The FIRE Angle: Why This Matters More Than You’d Think

Here’s the thing about subscriptions that I don’t think gets discussed enough: they’re invisible lifestyle inflation.

Every $100/month in recurring costs = $30,000 in extra retirement savings needed (at 4% withdrawal rate). Your $400/month in SaaS? That’s $120,000 you need in your portfolio just to cover software subscriptions in retirement.

For someone pursuing FIRE, eliminating subscriptions isn’t just about saving money today—it’s about permanently reducing your cost of living, which directly accelerates your retirement timeline.

I went through a brutal subscription audit last year using Beancount queries. I was paying for:

  • Three budgeting apps (YNAB, Mint Premium, Personal Capital paid features)
  • Two investment trackers (Morningstar Premium, some portfolio analyzer I forgot about)
  • Cloud storage for financial documents
  • Password manager
  • Receipt scanning app
  • Plus miscellaneous “productivity” tools I barely used

Total: $230/month = $2,760/year

Switching to Beancount + self-hosted/local-first tools: I cut that to under $50/year (just my domain name and occasional tool purchases).

The Professional Practice Calculation

For your bookkeeping practice, Bob, let me run the FIRE math:

Scenario A: Keep subscriptions

  • Software costs: $400/month = $4,800/year
  • After-tax cost (assuming 25% tax bracket): ~$6,400 in gross revenue needed
  • Retirement impact: Need $120,000 extra in portfolio to sustain this expense at 4% rule

Scenario B: Go Beancount-minimal ($40/month)

  • Software costs: $40/month = $480/year
  • After-tax cost: ~$640 in gross revenue needed
  • Retirement impact: Need $12,000 in portfolio

Difference: You’d need $108,000 LESS in retirement savings by eliminating subscriptions. Or alternatively, that’s 1.5-2 years earlier retirement if you invest the savings.

But here’s the kicker: what if you ALSO raised your rates because you’re offering “premium data ownership and privacy-focused bookkeeping”? Now you’re both reducing costs AND increasing revenue.

What I’ve Learned from Zero-Subscription Personal Finance

What works:

  1. Beancount + Fava is shockingly powerful. I built custom dashboards for everything YNAB and Mint did, plus stuff they couldn’t (net worth tracking across 15+ accounts, investment performance attribution, savings rate by category, FIRE projection modeling).

  2. Python is your friend. I’m not a developer—I learned Python specifically for Beancount. After building importers for my 8 banks/brokerages, I can now build custom analyses in minutes. This is a superpower.

  3. Local-first is faster and more private. No syncing delays, no “service unavailable” messages, no worrying about Mint selling my transaction data to advertisers.

  4. You learn your data intimately. When you’re writing queries instead of clicking dashboard buttons, you understand your finances at a deeper level. This makes better financial decisions.

What’s hard:

  1. No mobile app. I can’t check my net worth on my phone unless I VPN into my home network. For most people, this is a dealbreaker. For me, it’s actually a feature (reduces compulsive checking).

  2. Partner buy-in is tough. My spouse wanted “pretty charts” and “mobile access.” I had to build custom Fava dashboards and compromise on some things. If you’re serving clients, expect similar friction.

  3. Time investment upfront. I probably spent 60 hours learning Beancount, building importers, and setting up my workflows. But I’ve saved 2-3 hours per month for two years now (48-72 hours saved), so I’m approaching break-even—and the learning compounds forever.

For Your Practice

Some questions to consider:

  1. Can you charge more for data ownership? Frame it as “privacy-focused bookkeeping” or “custom financial infrastructure” and charge a premium. Clients who care about data sovereignty will pay extra.

  2. What’s the break-even timeline? If you save $360/month but invest 100 hours upfront, your break-even is when ($360 × months) > (100 hours × your hourly rate). For me, that was about 8 months. For you, it might be different.

  3. Could you open-source your setup? Build the Beancount bookkeeping practice toolkit, document it, share it. Suddenly you’re not just a bookkeeper—you’re a thought leader in plain text professional accounting. That’s differentiation.

  4. What about partial transition? Maybe start with your own firm’s books in Beancount, build confidence, then offer it as a premium tier to interested clients. “Standard” clients stay in QuickBooks, “premium” clients get Beancount + data ownership + lower monthly fees (because your costs are lower).

My Take

For personal finance, going zero-subscription has been one of the best decisions of my FIRE journey. The cost savings are real, but more importantly, I have complete control and deep understanding of my financial data.

For professional practice, I think there’s a real market opportunity here. Not everyone will want it, but the clients who DO want it (privacy-conscious, cost-conscious, technically-inclined) will be loyal and willing to pay for the differentiation.

The key is framing it right: you’re not “the cheap bookkeeper who can’t afford real software.” You’re “the boutique firm offering data ownership and custom financial infrastructure.”

Start with one client who fits the profile (tech founder, privacy-conscious, straightforward needs) and pilot it. See if the economics work. Then scale if it does.

Would love to hear how you’re thinking about pricing this if you go forward. Is it cheaper than traditional bookkeeping (pass savings to clients), same price (keep the margin), or premium (charge more for data ownership)?

4

I need to offer the CPA perspective here, because there are some professional liability and compliance considerations that go beyond just “can I technically do this?”

Disclaimer: I run a small CPA firm, not a bookkeeping practice, so my risk profile is different. But I work with bookkeepers regularly, and I see where things go wrong.

The Professional Standards Issue

When you’re providing professional bookkeeping services, you have certain obligations:

  1. Reasonable assurance that records are accurate and complete. Can you provide this with a custom Beancount setup? Absolutely—in fact, arguably better than QuickBooks because of Git audit trails and balance assertions forcing reconciliation.

  2. Protection of client data. Self-hosting on a VPS means YOU are responsible for security. Are you running backups? Encryption at rest? SSL certificates? Access controls? Intrusion detection? When you use QuickBooks Online, Intuit handles this (and has insurance if they screw up). When you self-host, it’s all on you.

  3. Business continuity. What happens if you get hit by a bus? Can your clients access their data? Can another bookkeeper take over? With QuickBooks, the answer is straightforward. With your custom Beancount setup… how do you handle succession?

  4. Professional liability insurance. Does your E&O policy cover self-hosted solutions? I’d check your policy language. Some insurers expect you to use “industry-standard” software and might balk at custom setups.

The “Industry Standard” Problem

Here’s an uncomfortable truth: banks, auditors, lenders, and investors often expect QuickBooks or Xero files.

Real scenario from last month: A client needed a business loan. The bank asked for “QuickBooks file export for the last two years.” Not PDFs of financial statements—the actual QBX file so their analyst could dig into transaction details.

If you’re on Beancount, you can’t provide that. You can provide:

  • PDF financial statements
  • CSV transaction exports
  • Fava web portal access
  • Plain text ledger files (good luck explaining that to a bank loan officer)

For some clients, this isn’t an issue. For others, it’s a dealbreaker. Make sure you’re screening for this upfront.

The Tax Preparer Handoff

I receive files from bookkeepers every tax season. Here’s what I typically get:

  • QuickBooks backup or export
  • Xero trial balance export
  • Excel P&L and balance sheet (for the spreadsheet warriors)

If a bookkeeper hands me Beancount files, I have two options:

  1. Learn Beancount (unlikely—I’m slammed in tax season)
  2. Ask for CSV exports and rebuild everything in my system (time-consuming, error-prone)

This adds friction. Not insurmountable, but real. You might need to provide “QuickBooks-compatible” exports (CSV files structured like QB reports) to make tax preparers’ lives easier.

What I Think Could Work

Despite the concerns above, I think there IS a viable model here, but it’s not “zero-integration”—it’s strategic minimalism:

Core principle: Use Beancount as your system of record, but provide standard outputs.

The stack:

  1. Beancount + Fava as your internal system
  2. CSV export scripts that generate QuickBooks-compatible transaction files
  3. PDF report generator for standard financial statements (so clients can share with banks/investors)
  4. Payroll service (non-negotiable—use Gusto, ADP, or similar)
  5. Payment processing partner (Stripe, Square—import their data into Beancount)

Client communication:

  • Never mention “plain text accounting” or “custom system” unless they’re technical
  • Say “we use a secure, cloud-based financial portal” (Fava is web-based, so this is true)
  • Emphasize data ownership: “You can export your complete financial history anytime, in any format”
  • Highlight audit trails: “Every transaction is timestamped and versioned”

Risk mitigation:

  1. Get professional liability insurance and confirm your setup is covered
  2. Document your workflows thoroughly (BPM, procedures manual, disaster recovery plan)
  3. Maintain succession plan (if you’re unavailable, who can access client data and hand off?)
  4. Regular backups with offsite storage (not just Git pushes—actual encrypted backups in a different location)
  5. Security audit (hire someone to review your VPS setup, or use a managed hosting service)

The Client Segmentation Strategy

Rather than “all clients on Beancount” or “all clients on QuickBooks,” consider segmentation:

Tier 1 - Beancount Premium (15-20% of clients)

  • Tech founders, privacy-conscious professionals, cost-conscious small businesses
  • Simple accounting needs, no inventory, no complex multi-entity structures
  • Comfortable with “custom” solutions
  • Premium pricing (charge MORE because of data ownership + customization)

Tier 2 - QuickBooks Standard (70-80% of clients)

  • Traditional small businesses
  • Need “industry standard” software for lending, audits, investor relations
  • Want phone support and established vendor ecosystem
  • Standard pricing

Tier 3 - Complex/Enterprise (5-10% of clients)

  • Refer to specialized firms using NetSuite, Sage Intacct, or similar
  • Out of scope for solo/small practices anyway

This way, you’re not all-in on Beancount, but you’re offering it as a differentiated premium service for the right clients.

My Honest Recommendation

Don’t go zero-subscription. Go selective subscription:

  • Use Beancount for clients who fit the profile
  • Keep QuickBooks Online for clients who need it (and charge them more to cover your software costs)
  • Use Gusto/ADP for payroll (charge markup to clients)
  • Self-host Fava but also maintain security, backups, and business continuity plans

You’ll still save significant money compared to “all clients on QuickBooks.” But you won’t box yourself into a corner where you can’t serve clients who need standard software.

And honestly? The $400/month in software costs is a business expense. If you’re charging clients appropriately, this should be covered in your pricing. If you’re eating the cost yourself, your pricing is too low—that’s a different problem.

Frame it as an investment, not a cost: The software enables you to serve clients efficiently. The question isn’t “how do I eliminate this expense?” It’s “how do I price my services to profitably cover my tools?”

That said, Fred’s point about FIRE math is valid. If you’re planning to retire early and want to minimize recurring costs in retirement, then building a Beancount-centric practice makes sense. Just do it with eyes open to the professional liability and client service implications.

Happy to discuss offline if you want to talk through specific scenarios or client situations.

5

I’m coming at this from a totally different angle—I’m NOT a bookkeeper (yet?), but I’m a former software engineer who’s been lurking here while considering a career transition into accounting.

Reading this thread is fascinating because it highlights something I’ve been wrestling with: the accounting profession seems to be at a weird inflection point where technical skills are becoming as valuable as accounting knowledge.

The Software Engineer’s Perspective

When I read Bob’s original post, my first thought was: “Wait, $400/month in SaaS costs is the PROBLEM? That’s incredibly cheap compared to software development tools!”

In tech, we routinely pay:

  • $25-50/month per seat for project management (Jira, Linear, etc.)
  • $50-100/month for CI/CD and deployment tools
  • $100-300/month for monitoring and observability
  • $50-150/month for various dev tools
  • Plus infrastructure costs (AWS, GCP, etc.)

A small dev team easily spends $2000-5000/month on tooling, and nobody blinks. But for bookkeepers, $400/month is a crisis. Why?

Because the revenue per client is completely different.

A software team generating $500K-2M in revenue can absorb $60K/year in tools (3-6% of revenue). But a bookkeeper charging $1000-2000/month per client can’t absorb those same costs—$400/month in tools is 20-40% of single-client revenue.

This is why Bob’s idea resonates: the economics of small-scale professional services DEMAND low-cost tools.

What Tech Can Teach Bookkeeping

In software development, we’ve been through this exact evolution:

2000s: Everyone used expensive commercial tools (Microsoft Visual Studio, Oracle databases, proprietary everything). Costs were insane.

2010s: Open source alternatives matured (VS Code, PostgreSQL, Linux, Git). Suddenly teams could run entire dev operations for near-zero software costs.

2020s: The best developers use open source as their foundation and only pay for services that genuinely save time (cloud hosting, specialized SaaS for non-core functions).

The pattern: Commoditize the core stack, pay for differentiated services.

For bookkeeping, this might mean:

  • Core ledger: Beancount (free, open source, you control it)
  • Differentiated services: Payroll (Gusto—compliance is too complex to DIY), payment processing (Stripe—network effects matter), specialized reporting (if needed)

The Technical Moat Advantage

Here’s what excites me about this discussion: if you’re a bookkeeper who can code, you have a MASSIVE competitive advantage.

Most bookkeepers can’t build custom importers, write Python scripts, or self-host web applications. If you can, you can:

  1. Offer services others can’t: Custom reporting, specialized workflows, data integrations that don’t exist in QuickBooks
  2. Operate at lower cost: No subscription overhead means better margins
  3. Scale differently: One VPS can host 50 clients for $100/month; QuickBooks would cost $50/seat × 50 = $2500/month
  4. Attract technical clients: SaaS founders, developers, tech workers—they WANT someone who speaks their language

This is the future, in my opinion. The bookkeepers who survive and thrive in 2030 will be the ones who are technologically fluent.

The Counterargument: Opportunity Cost

But here’s where Alice’s point lands hard: is building custom tools the best use of your time?

As a developer, I can build a bank CSV importer in 2-3 hours. For someone learning Python from scratch, that’s 20-30 hours (learning Python basics, understanding CSV parsing, debugging, testing).

If your hourly rate is $100/hour, that’s $2000-3000 in opportunity cost to build something that QuickBooks does automatically. It takes a LONG time to recover that investment through subscription savings.

However: The learning compounds. Once you can build one importer, you can build ten. Once you understand Beancount internals, you can solve client problems that QuickBooks can’t touch.

So maybe the right mental model is: invest in technical skills early in your career, reap compounding returns over decades.

What I’d Do (If I Were Starting a Bookkeeping Practice Today)

Phase 1 (Year 0-1): Hybrid approach

  • Use QuickBooks for most clients (proven, they expect it, less friction)
  • Use Beancount for YOUR OWN books (learn deeply, no client risk)
  • Build importers and tools in your spare time (nights/weekends)
  • Document everything (blog posts, GitHub repos—build your brand)

Phase 2 (Year 1-2): Selective Beancount offering

  • Offer “Premium Data Ownership Tier” to interested clients
  • Charge MORE, not less (this is specialized service)
  • Target tech founders, privacy-conscious professionals
  • Keep QuickBooks as standard tier

Phase 3 (Year 2-3): Open source your stack

  • Release “Beancount Bookkeeping Toolkit” as open source
  • Write guides, tutorials, YouTube videos
  • Position yourself as thought leader in plain text professional accounting
  • Attract clients BECAUSE of your technical reputation

Phase 4 (Year 3+): Scale or specialize

  • Either: Scale the practice (hire bookkeepers, train them on your system)
  • Or: Specialize in high-value Beancount consulting (enterprises, complex setups)

My Outsider Take

You’re not just deciding whether to eliminate $400/month in subscriptions. You’re deciding what KIND of bookkeeper you want to be:

Option A: Traditional bookkeeper using standard tools, competing on service quality and relationships (totally valid!)

Option B: Technical bookkeeper using custom tools, competing on specialization and capabilities (also valid!)

Both can succeed. But they’re different businesses with different economics, different ideal clients, and different skill requirements.

The exciting thing? You get to choose. And in 2026, with tools like Beancount mature and accessible, Option B is actually viable in a way it wasn’t 10 years ago.

Whatever you decide, I’m here learning from all of you. This community is gold for someone like me trying to understand where accounting meets technology.

(And if anyone needs help building Python importers or setting up self-hosted infrastructure, happy to help—consider it payment for all the accounting knowledge I’m absorbing here!)