SEO or Google Ads First? The Three-Question Budget Test

Four neighbouring questions are assumed here rather than re-argued, and there is no search pricing below. What search work costs is in how much SEO costs for a small business. How long it takes to pay off is in how long SEO takes to work. Why a site has no visitors at all is a different diagnosis, worked through in why your website gets no traffic. And what to do when visitors arrive and nothing happens is in my website gets traffic but no leads.

This piece is about one decision: you have a marketing budget that is smaller than the advice you keep receiving, and two salespeople have each told you their channel is the obvious starting point. Both of them were describing their own business model. The test below is what we actually run before quoting either one.

The Question Is Almost Always Asked Backwards

Owners frame this as which channel performs better. Framed that way it has no answer, because the two are not competing to do the same job. Paid placement converts demand that already exists, today, in whatever volume you are willing to pay for. Organic search accumulates a position you keep, slowly, at a cost that does not scale with the number of people who arrive.

The useful framing is a question about risk tolerance rather than performance: which of these can you afford to be wrong about? A failed ad test costs you the spend and ends cleanly in four to eight weeks. A failed year of content costs you the year, and you will not know it failed until most of it is gone. Those are genuinely different kinds of exposure, and a business with six weeks of payroll in the bank should treat them differently from a business with a profitable year behind it.

Rent Versus Asset, and Why It Changes the Arithmetic

Ad spend is rent. It buys visibility for exactly as long as the card is being charged, and the day you pause, the leads stop that afternoon. That is not a criticism — rent is the correct instrument when you need the room this month — but it means the spend never finishes. Year three of a campaign starts from zero the same way year one did, usually against more bidders paying more per click.

Search investment behaves like construction. The pages you publish and the technical debt you clear keep working after the invoice stops, which is why the cost per lead tends to fall as the body of work grows while an ad campaign's tends to drift upward. The catch is that construction has a period where you have spent money and own nothing usable, and in this category that period is measured in months. Most people who quit on organic search quit inside it.

Real Numbers, So You Can Do the Arithmetic Yourself

On the paid side you are buying clicks at a price other bidders set. Quiet local categories can sit in the low single digits per click. Categories where a single job is worth thousands — legal, insurance, water and fire restoration, elective medical, emergency trades — commonly run $20 to $50 per click and sometimes well past it. Management sits on top of the spend, typically ten to twenty percent of it or a flat monthly fee somewhere between several hundred and a few thousand dollars.

Now do the multiplication that most campaign proposals leave to you. At $25 a click, a hundred clicks is $2,500 and a hundred clicks is a small sample. If one in twenty of those becomes an enquiry and one in three enquiries becomes work, $2,500 bought you roughly one or two jobs. That is excellent if a job is worth $8,000 and fatal if a job is worth $400. Run that arithmetic against your own average job value before anyone shows you a dashboard, because it determines whether this channel can work for you at all.

This is also where the common budget failure becomes visible. A campaign needs enough spend to produce enough conversions to tell signal from noise, and in most local service categories that threshold is somewhere around $1,500 to $5,000 a month sustained for two or three months. A $400 monthly test in a $25-per-click category buys sixteen clicks. Sixteen clicks tells you nothing, and the conclusion people draw from it — that ads do not work in their industry — was never supported by the data.

The Three-Question Test

Three inputs decide this, and they are all about your business rather than the channels.

One: how many months of runway does the next lead have to arrive in?

If the answer is under three months, the decision is already made and the rest of this article is background reading. Organic search cannot be compressed into a quarter no matter who you hire, so a business that needs booked work by November is choosing between paid placement and something other than marketing. If the answer is a year or more of adequate current lead flow, the calculus flips entirely, because you are in the rare position of being able to fund an asset rather than rent a room.

Two: does anybody actually type your service into a search box?

Emergency and high-urgency services live in the search box. Somebody with water coming through a ceiling searches, clicks the first credible result and calls. Both channels work there, and paid placement works immediately. But plenty of good businesses sell something nobody knows to search for, or something bought on referral and reputation rather than query. Pouring money into either channel to capture demand that is not being expressed is the most common way a marketing budget disappears without a trace. Some of those businesses belong in paid social advertising instead, where you can reach people who are not looking yet.

Three: can the page you would send them to actually convert?

This is the question that saves the most money and gets asked the least. Both channels deliver a human being to a page. If that page loads slowly, buries what the business does, or demands a form from someone who would rather tap a phone number, the channel is not being tested — the page is, and it is losing. If you have never verified that your own enquiry form arrives in a real inbox, do that before you spend anything at all.

Where Paid Placement Wins Outright

Four situations where the answer is ads and there is no real argument. You are new, with no domain history, and need revenue before you need equity. You have a genuine deadline — a season, a lease, a new location opening in six weeks. You are testing whether a service line has demand at all, which paid clicks answer in days for the price of the clicks. Or your category is dominated by established competitors whose organic position will take a year or more to approach, which is a reason to rent while you build rather than to skip building.

There is a fifth and it is underrated: a short paid campaign is the cheapest keyword research available. Four weeks of real click and conversion data tells you which phrases produce paying customers rather than which phrases have volume, and that is exactly the input that makes the organic work that follows far less speculative.

Where Organic Search Wins Outright

The clearest case is a long tail of specific questions. Somebody searching a narrow version of your service — a particular material, a specific regulation, one neighbourhood, an unusual timeline — represents small individual volume and high intent, and there are hundreds of those phrases in any trade. Bidding on all of them is impractical. Answering them is precisely what content does, and each answer keeps earning after it is published.

It also wins when your margins cannot absorb a click price. If an average job is worth a few hundred dollars in a category where clicks cost $20, the arithmetic above never closes, and the only viable route to volume is traffic you do not pay for per visit. Our SEO services exist for exactly that shape of business, and the vertical guides — contractors and HVAC companies among them — show what that tail looks like in practice.

The Third Line Item Missing From Most Budgets

A share of your buyers no longer reach a results page at all. They ask an assistant, receive three recommendations in a paragraph, and contact one. There is no auction in that exchange. Whether your business is named depends on whether the assistant has something specific and citable to name, which puts answer engine optimization on the asset side of this decision rather than the rented side.

We treat it as its own line item rather than a bonus feature of content work, and it is currently the cheapest place to be early, because most local competitors have not started. The mechanics are in what answer engine optimization is for a small business, and it is delivered through our AEO work. It does not replace either channel above. It is the third column in the spreadsheet.

A Split That Works Once the Site Converts

The pattern that holds up across small budgets is not an even division. It is one channel funded to its threshold and the other funded minimally but continuously. Concretely: enough paid spend on your two or three highest-value services to keep work booking now, and the remainder going into the pages and technical fixes that compound. Never a thin campaign spread across every service line plus a token content retainer, which is the configuration where neither side reaches the point of producing evidence and both get cancelled in month four.

Review the split on a fixed cadence against one number. Not impressions, not clicks, not average position — booked work, tagged by where it came from. If paid is delivering jobs at an acceptable cost, keep renting while the asset builds underneath it. When organic starts producing the same jobs at a lower cost, shift the money deliberately rather than by feel. Businesses that do this for a year usually end up spending less in total than the ones that picked a side at the start.

Four Questions to Ask Whoever Is Selling You Either One

First: what is the minimum monthly spend at which this test produces a conclusion, and what happens if I fund half of it? An honest answer includes a floor and a refusal below it. Second: what do you need from my side, and what is the plan if the page they land on is the problem? Third: how will a booked job be attributed back to a channel — specifically, what happens to the phone calls? Fourth: at what point would you tell me to stop?

The fourth question is the one that sorts people. Anyone who cannot describe the conditions under which their own service should be cancelled is selling a subscription rather than an outcome. We build the site, the content and the citation layer with Claude Code, which is what lets us quote the whole picture rather than whichever piece we happen to sell — including the times the correct recommendation is to spend nothing on either channel this month and repair the destination first. What that repair involves is covered in why your website is slow and across our AI website building work.