OpenSEO on GitHub: turn the open-source Semrush alternative into a business and a SaaS

OpenSEO describes itself as an SEO tool for the people: if Semrush or Ahrefs are too expensive or bloated, it is a pay-as-you-go alternative you control. Instead of a fixed monthly seat, you connect your own DataForSEO account and pay only for the data you request, or you use the makers' hosted service.

The repository was created on 27 February 2026. When we checked on 18 October 2026 it had 23,995 stars, 3,218 forks, 20 contributors and about 780 commits, with the latest release, v0.1.13, published on 14 October 2026. The code is written in TypeScript and released under the MIT licence by Ben Senescu, under the every-app organisation.

It focuses on six workflows rather than trying to match every feature of the big suites: keyword research, rank tracking, competitor insights, backlinks, site audits and AI visibility, which checks whether your brand appears in AI assistants' answers. It also connects to Google Search Console and Google Analytics.

Its second big idea is that SEO now happens inside AI agents. OpenSEO exposes an MCP server, the standard way for AI assistants to use outside tools, plus ready-made agent skills, so you can ask ChatGPT, Claude, Codex or Cursor to run keyword research or an audit on your real data.

OpenSEO is a modern serverless web app. Knowing its parts tells you what running it as a business involves:

Docker is for trying it out on your own machine. It runs in a local mode with no login and a built-in admin user. The project warns you to expose it only behind your own authentication, a tunnel or a private network.

Cloudflare is the recommended route for anything internet-facing or shared with a team, and it works on Cloudflare's free plan. One deploy command creates the D1 database, KV namespaces, R2 bucket and Workers, applies migrations, and puts a Cloudflare Access login in front, allowing only the email addresses you list. R2 must be activated once in the dashboard, which needs a payment method on file even within the free tier.

Both routes need a DataForSEO API key. New DataForSEO accounts include $1 of free credit, and the minimum top-up is $50.

All of OpenSEO's SEO data comes from DataForSEO, a pay-as-you-go API company that is not affiliated with OpenSEO. There are no subscriptions, balances do not expire, and you can set daily spending limits. When we checked, its pricing page listed SERP results from $0.0006 each, Keywords Data from $0.0012 per task, DataForSEO Labs at $0.012 per task plus $0.00012 per item, and on-page crawling from $0.00015 per page. Backlinks and AI-optimisation data have their own pricing pages.

OpenSEO's pricing page translates that into everyday actions: about $0.05 for a keyword search at typical result limits, about $0.08 for a backlink overview of a domain with a year of history, and about $0.002 for one AI prompt check on one AI platform. Its example month of 100 keyword searches, 20 backlink checks and 217 rank checks comes to $7.52. The same page lists Ahrefs Lite at $129 a month for comparison.

The terms matter for a business. DataForSEO's terms of service say SERP data must not be used to compete with or adversely affect the search engines it comes from, and require you to indemnify DataForSEO if you break that rule. We found no clause in the terms that explicitly grants or forbids building a SaaS on the data. OpenSEO's own hosted service does exactly that, but before you charge customers, ask DataForSEO in writing to confirm that your planned use is covered.

OpenSEO is a clear example of an open-source project with a working business model, and the evidence is in the repository itself. The README says plainly that the hosted service makes money by charging 28% extra on every request it makes to DataForSEO, and the billing constants in the source code set that markup at 1.28.

The hosted plans, per the pricing page and source code, work like this:

Open source is the distribution. 24,000 stars and 3,200 forks spread the product further than advertising could. People try the free self-hosted version, and many decide $10 a month is cheaper than their own time.

The margin sits on usage, not seats. A $10 subscription covers the platform, and the 28% markup on data scales with customers' actual use. Heavy users pay more, and light users stay cheap and loyal.

Billing is engineered, not bolted on. The project's metering design document explains that every hosted data call first places a hold on the customer's credits for an estimated cost, then calls DataForSEO, then settles the hold at the real cost DataForSEO reports, or releases it if nothing was billed. The hold is atomic, so many simultaneous requests cannot overspend an account. Customers cannot run up costs you then fail to collect, which is the most common way usage-based products lose money.

AI agents are the moat. The MCP server and skills put OpenSEO inside the tools people already use, which a traditional SEO suite cannot easily copy.

What this means for you: the makers already sell the generic product at $10. A fork that only offers the same thing more cheaply has no advantage. Your business needs a reason to exist that they do not serve: a market, a payment method, a language, an industry, or a service layer.

The MIT licence allows a lot. You may use, copy, modify, publish, distribute, sublicense and sell the software, including running it as a hosted service, as long as the copyright notice and licence text stay in copies of the software.

The brand is not included. MIT licenses code, not names or logos. Rebrand your product completely: your own name, logo, domain and marketing. Do not imply that you are OpenSEO or endorsed by it. It is good practice, and good for trust, to say your product is built on the open-source OpenSEO project.

Data comes with its own terms. DataForSEO's terms apply to every request, including the rule against using SERP data to compete with search engines. Google's API terms apply to Search Console and Analytics data. Model providers through OpenRouter have their own policies.

Telemetry and privacy. Self-hosted installs send anonymised usage counts by default, never URLs, keywords, prompts or emails. Decide whether to keep it on, and tell your customers what data your service collects. In Kenya, customer data falls under the Data Protection Act, 2019, overseen by the Office of the Data Protection Commissioner; check whether your business must register as a data controller or processor.

Not every business has to be a SaaS. The fastest money usually comes from services that use the tool, and the SaaS grows out of what clients keep asking for. The table compares seven routes, and each is described in detail below. Effort and risk ratings are editorial judgements.

Who pays. Small and medium businesses that want more enquiries from Google: clinics, schools, hotels, law firms, online shops and contractors.

What you sell. Monthly SEO retainers: keyword research, technical audits, content plans, rank tracking and a plain-English monthly report. The client buys outcomes and expertise; OpenSEO is your internal tool.

How the repository helps. It replaces an expensive per-seat suite with pay-as-you-go data. Keyword research, site audits, rank tracking, backlinks and AI visibility cover most of a retainer's research work, and the MCP server lets you run repeatable analysis with an AI agent.

Pricing model. Illustrative: a one-off audit fee, then a monthly retainer per client. Your data cost per client is usually small compared with your time, so price your expertise, not the data.

Main risk. Selling rankings. No one can guarantee positions; sell a process, reporting and measured enquiries instead.

Who pays. Marketing and web agencies that report SEO results to many clients and want a branded dashboard instead of spreadsheets.

What you sell. A branded portal where each client logs in to see their rankings, audits and AI visibility, with monthly PDF reports generated automatically.

How the repository helps. Hosted mode already has accounts, organisations and invitations. Each agency client can be an organisation with its own projects. You add branding, report templates and per-agency billing.

Pricing model. Illustrative: a monthly platform fee per agency plus usage credits, or a fee per client workspace.

Main risk. Support load. Agencies expect fast fixes before their client meetings; set support hours and response times in writing.

Who pays. One type of business with shared SEO problems, for example hotels and lodges, private schools, real estate agents or online stores.

What you sell. A focused product with templates, checks and reports built for that industry: for hotels, booking-intent keywords, local rankings in tourist towns and competitor tracking against booking sites.

How the repository helps. The codebase is modular TypeScript with focused workflows. You keep the data engine and replace the generic onboarding, dashboards and reports with industry-specific ones.

Pricing model. Illustrative: a monthly subscription per business that includes a usage allowance, with top-ups for heavy users, the same shape as OpenSEO's own plan.

Main risk. Niche too small. Count the reachable businesses and their willingness to pay before building features.

Who pays. Kenyan and regional businesses and freelancers who find dollar card subscriptions hard to pay and English-only, US-centric tools hard to use.

What you sell. The same capabilities priced in shillings, paid by M-Pesa, with Kenyan locations set as defaults, local examples, and onboarding and support from people in the same time zone.

How the repository helps. OpenSEO's billing is a separate seam (a credit system with holds), so you can replace the payment provider while keeping the metering logic. DataForSEO lets you target locations and languages per request.

Pricing model. Illustrative: a low monthly shilling plan with included credits and M-Pesa top-ups. Check DataForSEO's keyword coverage for Kenyan locations first: volumes in smaller markets can be thin or missing.

Main risk. Competing with a $10 global product. Win on payment method, local relevance, language and support, not on price alone.

Who pays. Founders and marketers who already work in ChatGPT, Claude or Cursor and want SEO answers inside those tools.

What you sell. Set-up and custom agent skills that run the client's SEO workflows: weekly keyword gap reports, content briefs, technical fix lists and AI-visibility checks, delivered through the client's agent.

How the repository helps. OpenSEO exposes an MCP server and ships agent skills and plugins for Claude Code, Codex and Cursor. You build and maintain custom skills on top for each client's process.

Pricing model. Illustrative: a set-up fee per client and a monthly fee for maintaining skills, plus their data usage.

Main risk. Fast-changing agent platforms. Keep skills simple, documented and tested after each platform update.

Who pays. Agencies and in-house teams that want OpenSEO on their own Cloudflare account, with their own data and login, but no engineering time.

What you sell. Deployment on the client's account, Access login set up for their team, Search Console and Analytics connected, upgrades tested and applied monthly, and a support channel.

How the repository helps. One deploy command provisions the database, storage, Workers and the Cloudflare Access login. Upgrades are a pull and redeploy. The work you sell is doing it reliably and fixing problems.

Pricing model. Illustrative: a set-up fee and a monthly maintenance retainer. The client pays Cloudflare and DataForSEO directly.

Main risk. Being blamed for upstream bugs. Test each release on your own instance first and keep rollback steps ready.

Who pays. Students, career switchers, marketing teams and agencies training junior staff.

What you sell. Courses and workshops teaching keyword research, technical audits and reporting using real data, where learners run their own low-cost instance instead of buying an expensive suite.

How the repository helps. Low-cost, pay-as-you-go data makes hands-on practice affordable: the example month on OpenSEO's pricing page comes to about $7.52 of usage.

Pricing model. Illustrative: per learner per course, or per company workshop day.

Main risk. Out-of-date material as the tool changes. Teach SEO principles first and keep tool steps in an updatable appendix.

If you choose a SaaS (models 2, 3 or 4), the repository already contains most of the hard parts: accounts, organisations, invitations, metered billing, background jobs and an AI agent. What remains is mostly business work. Here is the order we would do it in.

Usage-based SEO products live or die on the gap between what customers pay and what data, payments and support cost. The example below uses OpenSEO's published markup and Paystack's listed Kenyan M-Pesa fee. Every figure is illustrative; replace it with your own numbers.

Assume a customer whose monthly activity costs $6.00 in DataForSEO data. With a 28% markup, that usage is billed at $7.68. If your base plan includes $10 of usage, this customer pays $10 and leaves $2.32 of allowance unused. Your gross margin before other costs is $10.00 minus $6.00 of data, or $4.00.

Now subtract payment costs. Paid by M-Pesa through Paystack at 1.5%, a $10 payment costs about $0.15. Infrastructure on Cloudflare is small per customer once you share a $5-a-month Workers Paid plan across many customers. That leaves about $3.80 per customer per month to cover support, marketing and your time.

The lesson: at $10 a month, a business needs many customers and very little support per customer. To make a smaller customer base work, you need a higher-priced plan that includes something valuable, such as industry-specific reports, a monthly review call, or done-for-you fixes. That is why service-led models often earn more per customer than a pure tool.

Most costs scale with usage, which keeps a small SaaS cheap to start. Prices below were listed on 18 October 2026.

Keyword data coverage. DataForSEO lets you set the location and language for each request. Check how much keyword volume data exists for Kenyan cities and for Swahili or mixed-language searches before you promise local keyword research. Small markets often show low or missing volumes, so pair keyword data with Search Console data from the customer's own site, which is free and reflects their real traffic.

Payments. Many Kenyan businesses prefer M-Pesa to cards, and dollar card subscriptions cause failed renewals. Shilling pricing with M-Pesa billing is a real advantage over global tools.

What customers actually buy. Most small businesses want enquiries, not rankings data. A dashboard plus a short monthly explanation of what changed and what to do next is usually worth more to them than more charts.

Agencies are a channel. Kenyan web and marketing agencies already serve the businesses you want to reach. A white-label portal or partner pricing can reach many customers at once.

These are the mistakes that most often sink open-source SaaS forks.

Days 1 to 5: self-host on Cloudflare with a $50 DataForSEO top-up. Run every workflow on your own site and two competitors, and record what each action actually costs.

Days 6 to 10: choose one business model and write a one-page offer: buyer, problem, deliverable, price, and what makes it better than the $10 hosted version for that buyer.

Days 11 to 18: sell before you build. If it is a service model, sign one paying client and deliver with stock OpenSEO. If it is a SaaS model, show a rebranded demo to ten target customers and collect pre-orders or a waitlist.

Days 19 to 26: fork, rebrand, turn on hosted mode and connect billing in a test account. Run the unit economics with your real usage numbers and adjust prices.

Days 27 to 30: onboard your first three customers by hand, watch where they get stuck, and fix those points first. Use the downloadable checklist below to track licence, billing, security and data-protection tasks.