Rank and Rent: Local SEO Sites That Earn $500 to $3,000 a Month

Written with AI assistance. Figures without a traceable source were cut before publishing.

The same piece is posted as a thread on GitHub — that copy has a reply box under it, and this one does not.

The rank-and-rent model is old enough that most people have heard of it and specific enough that almost nobody runs the numbers before starting: you build a website for a local service category, you rank it in Google for the searches that category gets, and then you rent the finished site to one business in that town for somewhere between $500 and $3,000 a month. What AI changed is only one line of the cost structure, the content, and I think that single change is what moved this from a full-time grind to something a solo developer can run alongside other work. What it did not change is the timeline, which is where most attempts die.

The Selection Is the Business

Pick the wrong city and industry pair and no amount of content saves you.

Two conditions have to hold at once. The industry needs a high customer value, which is why plumbers and dentists come up in every version of this playbook: one converted lead is worth hundreds or thousands of dollars to them, so a monthly rent in the same range is an obvious trade rather than a negotiation. And the city needs low competition, because you are not going to outrank an entrenched agency-built site in a major metro on your first attempt, and you should not try. AI helps here, mostly by making it cheap to screen a lot of pairs instead of falling in love with the first one.

High value, low competition. Both, not either.

I’d argue the second condition is the one people compromise on, because big cities feel like bigger opportunities and the search volume charts agree with them. The charts are measuring the wrong thing. A site that ranks third in a town of 40,000 earns rent; a site that ranks fortieth in a city of 4 million earns nothing at all, and the second one took longer to build.

There is a sharper version of this available. One developer, Brian Donatiello, registered a product name and domain that were literally the phrase his users type into Google — “low content book AI” — so his product appears on the first screen of results for its own category. Naming the asset after the query is the cheapest ranking decision anyone makes, and it costs the price of a domain.

Content Depth Beats Content Volume

Here is the number I keep coming back to when people ask how many pages they need. Jordan’s entire blog around the keyword “Poshmark Automation” has 15 articles, and 5 of them carry most of the organic traffic.

Five. Out of fifteen.

That ratio is not a failure of the other ten; it is what a working content asset looks like. A third of the pages do nearly all the work, which means the useful question is not “how do I publish more” but “which 5 questions does a customer in this category actually type before they call someone.” AI is very good at producing the fifty pages nobody reads and only adequate at producing the five that matter, so the leverage stays where it always was, in choosing what to write about.

The keyword research that supports that choice works better when it goes narrow rather than broad. Start from the industry term, break it into core terms, scenario terms, pain-point terms and long-tail terms, and keep going until you land on the question the user has not finished asking. Comment sections are unusually good raw material for this, because that is where people describe their actual situation instead of a search-friendly summary of it. Honestly, the best local-service page I have read was structured entirely around objections, not services.

Emergency plumbing at 2 a.m. Root canal cost without insurance. Those are the pages that rent.

Budget for Months, Not Weeks

Local SEO ranking takes several months. That is the constraint, and it is not negotiable by working harder.

The failure mode that follows is predictable enough that the source material calls it out directly: people give up in the blank window between launching and ranking, when nothing is happening and no signal tells you whether the site is going to work. I suspect this is the single largest cause of abandoned rank-and-rent projects, larger than bad niche selection, because bad niche selection at least produces a visible answer. The blank window produces nothing, which the brain reads as failure.

Two things make that window survivable. Build more than one site, so you are watching a portfolio move rather than staring at one asset. And separate the milestones you can control, which are pages published and technical issues fixed, from the one you cannot, which is where Google puts you in month four.

For the same reason I would not sign a rental agreement before the ranking exists. You have nothing to sell until the phone rings, and a site that ranks is a very easy sell precisely because you can show the business owner the search result with their competitor’s name absent from it.

What the Model Rewards

The durable version of this business is not the website. It is the ownership of the lead channel, which is a different asset from the code.

BackPedal makes the point in an adjacent category: founder James Dunn pairs GPS trackers with an offline recovery team and runs about $55,000 in monthly recurring revenue, and the moat is not the tracker. It is being willing to do the last-mile work that nobody else wants. Rank-and-rent has the same shape. Anyone can generate a services page. Ranking a local site and holding that rank for two years is unglamorous maintenance work, and that is exactly why it keeps paying.

There is a second-order version of the same idea worth watching, and I’m not sure how fast it arrives. Discovery is drifting from ranked lists toward AI readability, where an assistant picks one service and executes it rather than handing the user ten links to sort through. In that world the thing that gets you chosen is an explicit capability statement — what this business does, what it does not do, stated plainly enough for a machine to match against a request. WeChat’s own sample project spells it out at roughly that level of specificity: search the store’s drinks by name or category, and order directly. Whether or not that particular platform matters to you, the underlying advice is cheap insurance. Say what the site covers in unambiguous terms, because vagueness costs you nothing today and costs you the match later.

Presentation is cheap to fix now, which removes one old excuse. A brand-visual template in Miora takes a product description and a logo and returns a consistent set of visual assets, which is enough polish for a local service site and takes minutes rather than a design budget.

And when you price the rent, price it against the client’s lead value rather than your hosting bill. A plumber comparing $800 a month to the cost of one lost emergency job is doing arithmetic that favors you. A plumber comparing $800 a month to what a website “should” cost is doing arithmetic you introduced by accident.

The Honest Caveats

Every figure in this article comes from operators describing their own results, and the $500 to $3,000 range is a report rather than a distribution — I have no idea what the median is, or how many sites never rent at all. That second number is the one I would most want before starting, and nobody publishes it.

What I would take from the model regardless is the ordering. Choose the pair before you build anything. Write the 5 pages that answer real questions before you write the other 10. Assume months of nothing, and build enough sites that the months of nothing are happening in parallel. The AI part is genuinely useful and it is also the least important decision you will make here.

Also readable on Telegraph.

Read next

All 49 write-ups

The 7 figures in this piece — each with the sentence it came from — are in the figures table, alongside 401 more, as JSON and CSV.

Topics: AI

Part of llm-api-pricing — field notes on AI coding agents.

Did this save you an afternoon? A star on the repository is the whole ask — it is what puts these in front of the next person looking; the data is CC BY and does not require starring.

One thing this piece could not settle: How many months did your local site take to reach the first page, if it got there at all? Reply with a number. The reply box is on the thread copy of this piece.

Want a figure that is not in here yet? Say which metric, which provider, which unit — in one line. One required field, and the page you came from is already filled in. Got a better number? Open an issue — that form knows which write-up you came from too; corrections and counter-data are the point.