But:
Mixpanel holds every click your customers make. Every signup. Every drop-off.
Your funnels, your cohorts, your years of history all live on their servers.
And the bill grows every time your product gets more popular.












But:
Mixpanel holds every click your customers make. Every signup. Every drop-off.
Your funnels, your cohorts, your years of history all live on their servers.
And the bill grows every time your product gets more popular.

Mixpanel Growth at 20 million events a month is $1,635 a month, billed annually.

Then a team built the open source version of the whole product toolkit.
PostHog is the open source product analytics platform you can host yourself.

Now here is the part that actually matters. It is not just charts. You see the number drop, then you click and watch the real person's session. Same place. Then you flip a feature on for ten percent of users, or run a test, or ask a survey. Mixpanel now does some of this too. But PostHog was built as one box for all of it, and 97 percent of its users pay nothing.

Be honest about self-hosting. PostHog says its free self-hosted version is for roughly 100,000 events a month, with no support. For bigger products, they point you to their cloud. It is free for the first million events a month, then you pay by usage.

Here is what PostHog ships:

Worth Noticing:
This is open-core. Almost everything is MIT, which means you can do anything with it, including sell it. But one folder, called "ee", is under PostHog's own enterprise license. Using that part in production needs a paid PostHog plan. If you want zero strings, PostHog keeps a separate copy with that folder removed. Self-hosting is for small volume. Big products end up on the cloud bill.
Your users. Your events. Your replays. Your server.

Online stores pay you to wire up PostHog, build a checkout funnel dashboard, and review the session replays of abandoned carts every month. A realistic anchor is $1,500 setup plus $300 a month for the monthly "here is where you lost sales" report.
Host the fully open source edition of PostHog on your own servers and sell it to small clinics (dental, physio, med spas) that don't want visitor data going to a big ad-tech company. Charge a flat monthly fee per clinic website, for example $99 a month, and keep each site small, since PostHog says self-hosting is meant for low volume.
Early-stage software companies watching their Mixpanel bill climb pay a fixed fee to move their events, rebuild their key dashboards, and check the numbers match. A fixed project price in the $3,000 to $5,000 range fits a job that takes a week or two.
Sell a ready-made pack for people running online courses and memberships: a student progress funnel, a "where students quit" dashboard, and churn and feedback survey templates, with a setup guide. Sell it once as a digital product, for example $149.

But:
Shopify is the storefront, the checkout, the customer list, and the bill.
Your theme only runs on Shopify. Your apps only run on Shopify. Take payments with anyone else and Shopify takes an extra cut of every sale.

Shopify Advanced is $299 a month, billed yearly.

Then a small team built an open source storefront.
Your Next Store is the open source online store you can change with AI.

Now here is the part that actually matters. The open source part is the storefront. That is the shop your customers see: the home page, the product pages, search, the cart. That code is yours. But the back office is not open source. Your products, orders, checkout, and customer accounts live on Your Next Store's own hosted platform. You need their account to make it run.

So this is not "own everything." It is "own the front of the store." The platform has its own monthly bill. What it does not take is a cut of your sales. You pay Stripe's normal card fees, and that is it.

Here is what Your Next Store ships:

Worth Noticing:
Open front, closed back. Your store's data sits on Your Next Store's servers, not yours. It is young. On GitHub since August 2024, with one official release. Shopify still wins on its app store, in-person selling, and the sheer size of its ecosystem.
Your storefront. Your code. Your Stripe. Your look.

Ceramic studios and independent potters pay you to launch a good-looking shop: theme, photos, products, Stripe hooked up. Something like $2,000 setup + $150/month for changes, with the client paying their own Your Next Store plan.
Take the MIT storefront, design a look for one kind of seller, like specialty coffee roasters with subscriptions up front, and sell copies. Store owners pay you once per copy, anchored to premium Shopify themes (the top three on Shopify's theme store today are $150, $220, and $420).
Brands paying for Shopify plus a stack of add-on apps pay you for a flat-fee audit: what they spend, what Your Next Store already covers, what they would lose, and the switch plan. Your Next Store says it moves the store data itself, so your value is the decision and the redesign, not the copy-paste.
Pick one niche, like independent podcasters, and run a store for each show: products, drops, and a monthly store update. Each show pays you a flat monthly fee or a share of merch profit, while every store runs on its own Your Next Store plan.

But:
QuickBooks holds your client list, every invoice you have ever sent, and the record of who paid.
Your customers pay you through their payment screen. Your history lives on their servers. When they change the plans, your whole billing setup changes with them.

QuickBooks Online Advanced is $340 a month.

Then a team built the invoicing app you can own.
Invoice Ninja is the source-available invoicing and payments app you can host yourself.

Now here is the part that actually matters. The job most small businesses really need is: send a quote, turn it into an invoice, get paid by card, chase the late ones. Invoice Ninja does that job. Full bookkeeping, payroll, and taxes are a different QuickBooks. Invoice Ninja does not win that fight.

Your accountant still lives in QuickBooks? Invoice Ninja can push your invoices and payments into it. And if you do not want to run a server, Invoice Ninja sells a hosted version too, with its own monthly bill.

Here is what Invoice Ninja ships:

Worth Noticing:
This is not open source in the strict sense. It uses the Elastic License, which means you can use it, change it, and even install it for paying clients. What you cannot do is turn it into your own invoicing service and resell it. Removing the Invoice Ninja branding from what your clients see costs $40 a year. QuickBooks still wins on bookkeeping and taxes.
Your invoices. Your clients. Your payments. Your server.

Plumbers, HVAC and landscaping companies pay you to install Invoice Ninja on their own server, connect Stripe, and set up quotes, maintenance-plan auto-billing, and late-payment reminders. A realistic anchor is $1,500 setup plus $150 a month to keep it updated and backed up.
Small agencies and consultants pay you a flat fee to import their clients, products, and open invoices from QuickBooks, then keep their accountant happy by pushing new invoices back into QuickBooks. Price it as a one-time project, around $1,000 to $2,500 depending on how messy the data is.
Sell ready-made invoice, quote, and proposal designs for one kind of business, like wedding photographers or event DJs, with the right line items, deposit terms, and payment schedules baked in. A design pack can sell for $49 to $99, and you can offer a $300 custom-branding version.
Belgium has required electronic B2B invoices over the Peppol network since January 1, 2026, and Invoice Ninja can send them. Freelancers and small firms pay you to set it up correctly and test it, for around $500 each; the Peppol sending credits are a separate cost from Invoice Ninja.

But:
Calendly is the link in your email signature. It is the page every client sees before they ever talk to you.
It reads your calendar. It holds every name, email, and time slot of everyone who ever booked you. Stop paying, and your booking page stops booking.

Calendly Standard is $12 a seat a month if you pay monthly.

Then a team built the open source booking page.
Cal.diy is the open source scheduling tool you host yourself.

Now here is the part that actually matters. You get the booking link. It checks your Google or Outlook calendar, so nobody books you twice. It drops in a Zoom or Google Meet link. It can charge people before they book. And it is MIT, which means you can do anything with it, including sell it. That is the Calendly job for one person. Round-robin for a sales team, routing forms, and automated reminder workflows are not in here. Calendly still wins that fight.

There is no hosted version. You run it on your own server. Its own docs say use it at your own risk and recommend it for personal use. If you want someone else to keep it running, that is Cal.com, the paid version from the same company.

Here is what Cal.diy ships:

Worth Noticing:
This is the one-person version. No teams, no round-robin, no routing forms, no automated workflows. The last tagged release was in March, though changes still land every week. You are the IT department. Calendly still wins for teams and for hands-off.
Your booking page. Your calendar. Your clients. Your server.

Set up a branded booking page on a tutor's or music teacher's own domain, with paid lessons through Stripe and their calendar connected. Charge something like $500 setup plus $40 a month to host it and keep it updated.
Run one server that gives solo tattoo artists their own booking page that takes payment up front before a slot is held. Sell it as a simple niche product, around $15 an artist a month.
Build a Cal.diy app or connector that a specific trade needs, like sending bookings into a salon's client-card spreadsheet or a coach's Notion workspace. Sell it one-time, around $49, or bundle it with your setups.
Move a solo consultant off Calendly: rebuild their meeting types, connect calendars and video, embed it on their site, and wire their Zapier steps back up. A flat fee around $750 is a reasonable anchor.

But:
Your prices live inside Chargebee. So do your plans, your trials, and every customer's renewal date.
Want to raise prices, or add credits for an AI feature? You change it on their system, on their terms.
And the fee grows every time your revenue does.

Chargebee's Flow plan is $99 a month plus 0.65 percent of everything you bill, which is $3,349 a month once you bill half a million.

Then a team built the open source billing layer.
Autumn is the open source pricing and billing platform that sits on top of Stripe.

Now here is the part that actually matters. Pricing becomes a setting, not a project. You want to raise the price of Pro? You change it in the dashboard. New customers get the new price. Your existing customers stay on the old one until you decide to move them. Free trials, usage limits, credits, all set the same way. Your developer does not rebuild the app every time you change your mind.

Autumn does not replace Stripe. Stripe still moves the money, and Stripe still takes its fees. Autumn is the brain on top that decides who gets what.

Here is what Autumn ships:

Worth Noticing:
This is young. It went on GitHub in January 2025. Self-hosting is real, but the team itself says it depends on a bunch of other services and to ask them for help. Most people will use their hosted cloud, which is free to start and $375 a month on Pro. Chargebee still wins on finance-team stuff like quotes, revenue recognition, and big multi-company setups.
Your prices. Your plans. Your customers. Your billing.

Small AI companies pay you to design and wire up their plans: a free tier, a credit balance, top-ups, and overage charges, all running on Autumn and Stripe. A fixed package like $2,500 to set it up plus $300 a month to run pricing changes is an easy sell to a founder who just wants it handled.
Run your own hosted copy of Autumn, under your own brand, for people shipping apps on tools like Bubble or Lovable who need paywalls and usage limits without a developer. Charge a simple monthly fee, for example $49 a month, below what a developer would cost them for a single afternoon.
Sell a ready-made starter kit: a pricing page, upgrade buttons, a credits meter, and a "you hit your limit" screen, pre-connected to Autumn. A one-time price like $149 fits what indie builders already pay for app templates.
Subscription companies stuck on hand-built Stripe billing or on Chargebee hire you to move them over and run their next price increase safely, keeping old customers on old prices. This is fixed-fee project work, for example $5,000 per migration.

But:
Typeform holds every answer your customers ever typed.
Your lead forms, your quizzes, your signups all live on their servers.
And every self-serve plan counts responses. Hit the cap, and you are shopping for the next plan.

Typeform's top plan, Growth Flow, is $266 a month billed yearly, and it still caps you at 10,000 responses a month.

Then a solo developer in France built the open source chatbot builder.
Typebot is the open source form and chatbot builder you can host yourself.

Now here is the part that actually matters. A Typebot is a Typeform that talks. One question at a time, like a chat. It branches on the answers. It takes payments. It files everything into Google Sheets or your own tools. And on your own server, there is no response meter. Typeform still wins on polish and its newer AI tools. Typebot wins on owning the whole thing.

One honest note on the license. It is "fair source," not classic open source. You can use it for your own business and build bots for clients. You cannot sell Typebot hosting to other people. Each version turns fully open, Apache 2.0, two years after it ships.

Here is what Typebot ships:

Worth Noticing:
Typebot is mostly one person's project. That is fast, but it is a small team to depend on. The license blocks you from reselling it as a hosted service for two years. Typebot also sells its own hosted cloud if you do not want to run a server, and that has a monthly bill.
Your forms. Your chats. Your answers. Your server.

Build a chat that qualifies home buyers and sellers (budget, timeline, pre-approval) and drops hot leads straight into the agent's inbox or CRM. Charge the brokerage something like $1,500 to build it and $150/month to tweak it and report on it, with the bot running on the brokerage's own Typebot account or server.
Coaches and course creators often run dozens of Typeforms; you rebuild them as Typebots and install Typebot on the client's own server. A flat project fee (say $2,000 for up to ten forms) pays for the rebuild, the install, and one training call.
Sell ready-made bot files people import into their own Typebot, like client-intake bots for law firms or new-patient bots for dental clinics. Price a pack at around $79, and name it after the niche, not the Typebot brand.
Run a local site like "get a roof quote in two minutes" where a Typebot collects the homeowner's details and photos, then sell those leads to local roofers at a per-lead price you agree with each one. You are selling leads, not software, so it is plain use of the bot in your own business.

But:
Salesforce holds your pipeline. Every lead, every deal, every note your reps ever wrote.
Your sales team lives in it all day. Leaving means moving years of history.
And every person you hire is one more seat.

Salesforce Sales Cloud Max is $550 a user a month, billed annually.

Then a team built the open source CRM.
Frappe CRM is the open source sales CRM you can host yourself, with unlimited users.

Now here is the part that actually matters. There is no seat price. Add a salesperson and the bill does not move. Every deal gets one page. The emails, the calls, the notes, the tasks, all in one place. That is the Salesforce job most small teams actually need. Salesforce still wins on AI agents and its giant add-on marketplace. Frappe CRM does not win that fight.

Frappe CRM is made by Frappe, a software company in India. They also sell a hosted cloud. It is priced by the server, not by the person. Self-hosting is free.

Here is what Frappe CRM ships:

Worth Noticing:
The license is AGPL. That means you can use it, change it, and even host it for paying clients. The catch: if you change it and let people use it over the web, you must share your changes. There is no locked enterprise folder. It is young, on GitHub since July 2023. And custom work means finding someone who knows Frappe.
Your leads. Your deals. Your team. Your server.

Set it up for small real-estate brokerages so Facebook and Instagram leads land straight in the CRM and agents follow up by WhatsApp and phone. Charge the broker for setup and monthly upkeep, e.g. $2,000 setup + $250/month.
Run a managed, private copy for independent insurance agencies: you host it, back it up, and update it, and the whole office pays one flat fee. A price like $149/month per agency works because there is no per-seat cost underneath you.
Move small sales teams off Salesforce: export their leads, deals, contacts, and history, rebuild their stages, and train the reps. Sell it as a fixed-price project, e.g. $3,000–$8,000 depending on how much data and custom fields they have.
Build a ready-made setup for solar installers: site-survey fields, quote-to-install stages, email templates, and web forms for inbound leads. Sell the pack plus install as a product, e.g. $499 one-time with optional paid support; the code stays AGPL, so you are selling the work and the updates, not a locked license.

But:
Your contact list lives in their database. So do your automations.
If you are an agency, it is worse. Every client account you sell sits inside GoHighLevel. Your whole business is a rebrand of theirs.

GoHighLevel's Agency Pro plan is $497 a month.

Then a team built the open source marketing automation engine.
Mautic is the open source marketing automation platform you can host yourself.

Now here is the part that actually matters. The piece most people use GoHighLevel for is the follow-up. The email sequences. The "they clicked, so send this next" logic. That is exactly what Mautic does, with unlimited contacts and no per-contact bill. But GoHighLevel bundles a lot more. Sales pipelines. Funnel builder. Phone calls. Review requests. A branded SaaS mode that bills your clients for you. Mautic does not ship those. It is the email and automation core, not the whole building.

One more honest thing. Mautic does not send the email itself. You plug in a sending service, like Amazon SES or SendGrid, and you pay for that. And for an agency, each client usually gets their own copy of Mautic. There is no one-click sub-account button.

Here is what Mautic ships:

Worth Noticing:
The license is GPL version 3. That means you can use it, change it, rebrand it, and sell services on it. If you hand someone a copy of the software, you must hand them the source code too. The name Mautic is a trademark, so you cannot put it in your company or product name without permission. GoHighLevel still wins on everything-in-one-box.
Your contacts. Your campaigns. Your clients. Your server.

Dental and physio clinics pay you to set up and run their appointment reminders, recall emails, and review nudges on a Mautic you host for them. A reasonable anchor is $1,500 setup plus $300 a month for management.
Host a separate, rebranded Mautic for each client in one niche, say real estate brokerages, and sell it as your own email-automation product. Something like $149 a month per office, undercutting a $497 GoHighLevel agency plan once you have a few clients.
Build add-ons Mautic users want, like a booking-tool integration or an industry pack of ready-made campaigns and email templates, and sell them with support. Price it like a product, $99 to $299 each, and remember GPL buyers are allowed to share the code, so you are really selling updates and support.
Companies tired of per-contact bills pay you a fixed fee to move their lists, forms, and sequences off HubSpot, ActiveCampaign, or GoHighLevel email, and set up their sending domain so mail lands in the inbox. A fixed-price project of $3,000 to $5,000 is a realistic anchor.

But:
Fin is the AI that answers your customers before any human does.
It learns from your help docs, your old conversations, your policies.
That front door, trained on your knowledge and talking in your name, now belongs to Salesforce.

Intercom Expert is $132 a seat a month billed annually, plus 99 cents and up for every conversation Fin resolves.

Then a team built the open source support inbox, with its own AI agent.
Chatwoot is the open source Intercom alternative you can host yourself.

Now here is the part that actually matters. Chatwoot's AI agent is called Captain. You point it at your help center, your website, your PDFs. Link it to an inbox and it talks to customers. When it gets stuck, it hands off to a person. It can even check things in your own systems, like a warranty or an order. But Captain is not in the free part. It lives in Chatwoot's enterprise folder, not the MIT part. MIT is the license that lets you do anything, including sell it. The enterprise folder needs a paid Chatwoot plan.

So if you self-host and want Captain, you pay Chatwoot and bring your own OpenAI key. If you stay on the free core, you still get small AI helpers. Rewrite a reply. Summarize a thread. Suggest an answer. And there is a door to plug in any bot you build yourself.

Here is what Chatwoot ships:

Worth Noticing:
Fin is still the stronger AI. Fin says it resolves 76% of questions on average, across more than 12,000 customers. Chatwoot publishes no number like that. And Captain sits in the enterprise folder. Delete that folder and you have the free MIT product, minus the AI agent.
Your inbox. Your AI. Your knowledge. Your server.

Set up Chatwoot plus Captain for dental or physio clinics so patients get instant answers on WhatsApp and the website, and staff take over for bookings. Charge a setup fee plus a monthly retainer (for example $1,500 setup + $200/month), with the clinic paying Chatwoot for its Captain plan directly.
Run the free MIT core on your own server as a managed, branded support inbox for small property management firms: tenant chat, email, and SMS in one place. Charge a flat monthly fee per firm (for example $99/month) that covers hosting, updates, and backups.
Build an AI answer bot that plugs into Chatwoot's built-in bot connection, tuned for one niche like Shopify stores (order status, returns, sizing). Sell it as a monthly add-on to anyone running Chatwoot, with no Chatwoot enterprise code inside it.
Help companies rethinking Intercom now that Salesforce owns it: move help articles, set up channels, and train the team. Sell it as a fixed-price project (for example $3,000 to $5,000 per company).

But:
Semrush holds your keyword lists, your rank history, and your site audits.
Your agency reports live inside its dashboard. Stop paying, and you lose the dashboard.
Want to track more websites or more keywords? That means a bigger plan.

Semrush Advanced is $549 a month, or $455.67 a month billed annually.

Then a small team built the open source SEO tool.
OpenSEO is the open source alternative to Semrush and Ahrefs that you can run yourself.

Now here is the part that actually matters. OpenSEO does not have its own map of the internet. The keyword numbers and backlink data come from DataForSEO, a paid data company. You bring your own DataForSEO key and pay them per search. The README gives examples: about five cents for a keyword search and about eight cents for a backlink check. Most small sites use a few dollars a month, not hundreds.

If you do not want to set it up, the hosted version is $10 a month with $10 of usage included. That is how they make money: the README says they add 28% on top of every DataForSEO request. Self-host and you skip that markup. DataForSEO asks for a $50 minimum top-up.

Here is what OpenSEO ships:

Worth Noticing:
The software is free. The data is not. Every search runs through DataForSEO, and you pay them for it. Semrush still wins on its own giant database, years of history, and one tool that does everything. OpenSEO is young. It went public in February, and it is only at version 0.1.9.
Your keywords. Your rankings. Your reports. Your server.

Sell monthly SEO to local businesses like dentists or roofers, running OpenSEO for rank tracking and audits. Your data bill is a few dollars a client, so a price like $300 to $500 a month is mostly your time.
Host your own branded copy for small marketing agencies, so their clients log in to "their" SEO tool instead of a Semrush seat. Charge each agency a flat monthly fee, with the DataForSEO usage passed through at cost plus a margin.
Charge founders and marketing teams a one-time fee to wire OpenSEO into Claude or ChatGPT and build custom skills for their weekly SEO tasks. Think $1,000 to $2,000 per setup, plus an optional monthly check-in.
Fork it into an SEO tool for one industry, like Shopify stores or real estate agents, with built-in keyword lists and audit checks for that market. Sell it as its own subscription; the MIT license allows it.