Search, paid ads, or both
One rents attention and stops when you stop. The other compounds and takes months to arrive. Most arguments about the two are really arguments about how long the business can wait.
In short
Paid ads answer whether demand exists and what it costs to reach. Search answers what happens once it compounds. If you do not yet know your conversion rate or which message works, run paid first, because it buys that answer in weeks instead of quarters. If you already know both and the cost of acquiring a customer is climbing, that is the signal to build search. Running both badly is the most common outcome and the most expensive one.
The six real options
Paid search
Buys attention immediately and stops the day you stop. Excellent at testing a message, filling a pipeline this quarter, and finding out fast whether anybody wants the thing.
Search and content
Takes months to arrive and then keeps working. What you are building is not traffic, it is the position of being the answer when somebody goes looking.
Both, in that order
Paid first to learn which message converts and what a customer is worth, then search built on the language that paid proved. This is the sequence that wastes the least.
Neither, fix conversion first
If the page you already have converts poorly, both channels multiply the same leak. Doubling a two percent conversion rate is cheaper than doubling traffic and it improves every channel at once.
Neither, fix the offer
Sometimes the traffic is fine and the offer is not. No channel survives that, and no amount of spending diagnoses it. Ten conversations with people who did not buy will.
Being the cited answer
A newer third option. Assistants answer questions by quoting sources. Content written to be quotable earns a kind of visibility that neither an ad auction nor a ranking position describes.
The version of this argument that is actually useful
The argument is usually framed as which channel is better, which has no answer, because they are not the same kind of thing. One rents attention and one builds an asset. Asking which is better is like asking whether renting or buying is better without mentioning how long you plan to stay.
The useful frame is time. Paid buys speed and buys information. Search buys durability and buys margin. A business that needs to know within six weeks whether anybody wants the thing should not be writing articles, and a business with a working offer and a rising cost of acquisition should not be spending another year renting.
What makes this expensive is running both at half attention. Two channels each getting a third of the effort they need produce less than either one done properly, and the reporting from each obscures the failure of the other for a couple of quarters.
There is also a quieter failure that shows up in almost every audit: traffic was never the problem. The page converts poorly, the offer is unclear, or the follow up takes three days. Every visitor bought at any price walks into the same leak, and fixing that leak improves both channels at once at no media cost.
The newer wrinkle is that assistants now answer a large share of the questions people used to type into a search box. That does not kill search, it changes what winning looks like. The prize is being the source that gets quoted, which rewards content that is specific, structured and genuinely useful, and offers nothing at all to content written to hit a word count.
Practically, that means the two channels have drifted further apart rather than closer. Paid is still an auction and behaves like one. Content is now competing to be a citation rather than a blue link, and the writing that wins is the writing somebody would have wanted to read anyway.
What each channel is actually for
Nine factors, read across. The rows that decide it are the first, the third and the last: how fast you need an answer, how the cost behaves as you scale, and what winning now looks like.
| Factor | Paid ads | Search and content |
|---|---|---|
| Time to first result | Hours. Traffic arrives as soon as the campaign is approved. | Three to nine months for anything competitive, longer on a new domain. |
| When you stop | It stops the same day. Nothing accrues. | It decays slowly. Pages keep earning for years with light maintenance. |
| Cost behaviour | Rises as you scale and as competitors bid against you. | Front loaded. Cost per visit falls the longer it runs. |
| What it teaches you | Which message and which audience convert, quickly and unambiguously. | Which questions your market actually asks, in their words rather than yours. |
| Best at | Testing, launches, seasonal demand, filling a pipeline this quarter. | Compounding demand, credibility, and being the answer people arrive at. |
| How it fails | You buy traffic that was never going to convert and find out after the budget is gone. | You publish for a year against terms nobody with a budget is searching for. |
| Who it needs | Somebody who reads the numbers weekly and is willing to switch things off. | Somebody who knows the subject well enough to write something worth citing. |
| Attribution | Clean, immediate, and slightly flattering to itself. | Messy and almost always undercounted, especially on longer sales cycles. |
| Effect of AI assistants | Largely unchanged. It is still an auction for attention. | Changed materially. Being the source an assistant quotes is the new first position. |
The two are not substitutes and the comparison is not really about marketing. It is about how long the business can wait for an answer, which is a finance question wearing a marketing hat.
Five questions before you spend anything
The first two are usually enough. Companies that answer all five almost never argue about channel mix again, because the answer stops being a preference.
- Do you know what a customer is worth?Not revenue on the first order. Gross margin across the whole relationship. Without that number every channel argument is aesthetic, because there is nothing to compare a cost of acquisition against.
- Does your page convert the traffic it already gets?Look at the last hundred visits and what they did. If the page converts poorly, both channels are pouring into the same leak and the cheapest available win is sitting on the page you already have.
- How long is your runway?Search is a six to twelve month commitment before it repays anything. If the business needs pipeline inside a quarter, that decision is already made, and the honest move is paid now and search started quietly alongside it.
- Does anybody actually search for this?Check that real search demand exists for the words your buyers use, not the words your industry uses internally. Whole content programmes have been built against terminology that only exists inside the company.
- Who owns it in month four?Both channels fail the same way, through inattention. Paid drifts and overspends, content stops being published. Name the person before the budget, and if there is not one, pick the channel that survives neglect better, which is search.
Questions we get
Which should a small company start with?
Usually paid, in a small and disciplined test, because it answers the questions that make everything else cheaper: does demand exist, which message lands, and what a customer costs to acquire.
Start search in parallel but quietly, on the strength of what paid teaches you about the words people actually use.
How long before search pays back?
On an established domain with existing links, a few months. On a new domain in a competitive category, six to twelve, and the first six can look like nothing is happening while indexing and authority build.
If that timeline does not fit the business, that is a real constraint and it should change the plan rather than be argued with.
Can we just do content and skip ads entirely?
Yes, and plenty of good businesses have. It is slower and it demands genuine subject knowledge, because thin content in a crowded category now competes with everything an assistant can summarise for free.
What you give up is the fast feedback loop. Without it you may write for a year before finding out the message was wrong.
Is SEO dead because of AI answers?
No, but the shape of the prize changed. Fewer people click through from an answer they already received, and being the source inside that answer matters more than being a ranking position.
What still works is the same thing that always worked: specific, structured, useful writing on questions people genuinely have. What has stopped working is volume for its own sake.
How much of the budget should go to each?
There is no ratio worth trusting. Decide by runway and by what you do not yet know. If you cannot state your cost of acquiring a customer, spend on the channel that will tell you.
Once you can state it and it is rising, shift toward the channel with a falling marginal cost, which is search.
What is the fastest way to waste money on ads?
Sending paid traffic to a page that was written for somebody who already knows you. Cold traffic and warm traffic need different pages, and the same page rarely serves both.
The second fastest is leaving campaigns running without anybody reviewing them weekly.
Should we hire an agency for this?
For paid media at scale, often yes, because it is a specialist discipline with a real learning curve. For content, only if they understand your subject well enough to write something a customer would respect.
In both cases agree at the start what would make you end it, and review against that rather than against the monthly report.
What should we do this week?
Work out what a customer is worth, and look at what the last hundred visitors to your site actually did. Those two numbers decide the channel question, and neither costs anything to find.
If the page converts badly, fix that first. It is the only work that improves every channel simultaneously.
More in the guides and every answer in one place.