SEO SOP: Validate Keywords With Three Weeks of Google Ads

Before you commit six months of SEO to a keyword set, buy three weeks of ads on it and let the real data settle the argument. This is the test Gert Mellak of gertmellak.com runs when a client insists on keywords he thinks are wrong, from his Unscripted SEO interview. It costs three weeks and an ad budget, and it answers the one question a keyword tool cannot: do the people who type this actually buy.

Objective

Replace a keyword argument with a three-week purchase test. You end up with two numbers a search volume tool cannot give you: how many people really typed the exact query in your market, and how many of them converted after landing on your page. Those two numbers decide whether the SEO project gets built on those keywords or on different ones. Setup is about two hours, the test runs three weeks, and reading the result takes an hour.

Key Steps

1. Ask the client to name the three keywords they believe will make or break the business. Their three, not yours. Gert’s framing is exact: “pick the three keywords that you think are going to make or break your business.” Writing them down is what makes the test binding later, because both sides agreed on the terms before the data arrived.
2. Write down what each side expects, before the ads run. Record the tool’s monthly search volume for each keyword and the client’s expected conversion count. This is the prediction the test is scored against. Without it the result gets reinterpreted after the fact.
3. Pause the SEO project on those keywords. Gert stops the project, runs the ads, and restarts it after: “run ads for them over three weeks. And then you come back, and then restart the SEO project.” Do not build content on a term you are about to test. If the test kills the keyword you have thrown away the work.
4. Run Google Ads on exactly those three keywords, in exact match, for three weeks. No broad match, no expansion, no other terms in the campaign. The whole point is measuring the people who typed that specific query. Send them to the page you would have built the SEO around, or the closest live equivalent. (inferred: he specifies three keywords and three weeks; exact match is what makes “people who actually searched for this exact keyword” measurable.)
5. Budget for enough clicks to read a zero. A zero-conversion result only means something if enough people clicked. Set the daily budget so each keyword can accumulate a meaningful click count over the three weeks rather than spending out in the first four days. (inferred: Gert does not name a budget; he does rely on “so many people who actually searched for this exact keyword” having had the chance to convert.)
6. Change nothing else for three weeks. No landing page edits, no offer changes, no other campaigns pointed at the same page. The three weeks exist so people have time to decide. Gert makes that explicit: “in a space of three weeks, so they even had time to decide.”
7. Read two numbers, not the whole report. First, real impressions and clicks against the tool’s search volume. Gert’s example: “the 2,500 search volume actually is 120.” Second, conversions from people who typed the exact query. Everything else in the ads dashboard is noise for this decision.
8. Apply the decision rule and restart the project on what survived. Conversions at an acceptable cost means the keyword is real, build the SEO on it. Real click volume with zero conversions means the keyword is not the business, drop it no matter what the volume column says. Real volume far below the tool number means recalculate the case on the true number, and keep the keyword if the traffic still converts. Gert’s own 2,500-to-120 case survived on that basis: the click-through rate he expected was “decent enough to make this work.”

Cautionary Notes

• This is a test of purchase intent, not of your SEO. An ad result and an organic result are not the same click. Use it to decide whether the demand and the conversion exist at all, not to forecast organic traffic.
• Broad match destroys the test. The moment other queries enter the campaign you are no longer measuring the keyword the client named, and the result is arguable again.
• Do not let the tool’s volume number stand in for the answer. Gert’s whole objection is that the volume column is not what people think it is. Search Console does not rescue it either: “We know now that Search Console is heavily impacted by bots. Lots of bots produce impressions.”
• Three weeks is short for a long sales cycle. If the buying decision takes two months, a zero-conversion result may be a timing artifact. Track soft conversions and key-page landings in that case rather than closed deals.
• Do not run it as a gotcha. Gert describes it as “one of the best methods I have come up with to just show them when I think they are wrong,” and it only works because both sides agreed to the terms in step one. Run it as a shared question and the client owns the answer with you.
• A surviving keyword is not a strategy. It is one validated term. The SEO project still needs the rest of the work built around it.

Tips for Efficiency

• If the client already runs paid, skip the test and pull their converting-keyword report instead. Gert asks the PPC team for exactly that on every account. The three-week test is for when nobody has the data.
• Run all three keywords in one campaign with three exact-match ad groups, so budget pacing and reporting stay in one place.
• Book the read-out meeting when you launch the test, three weeks out. It stops the test from quietly running for two months.
• Save the before-and-after table. Tool volume next to real volume, expected conversions next to actual, on one page. It is the artifact that changes the next keyword conversation before you have to run another test.
• Feed the result back to the paid team either way. A term that converts in ads is a term worth ranking for, and a term that does not is worth pausing on both sides.

Why this one matters

Gert’s argument on the episode is that the standard SEO business case is built on numbers that cannot carry it. Volume times click-through rate times conversion rate times lifetime value, with 68% of searches never producing a click at all. His verdict: “Thinking that you still can calculate an ROI based on a traffic metric or a search volume metric, I think is insane.”

The damage shows up in keyword selection. He described a client who picked broader keywords than their niche warranted, for one reason:

“They had to pick those keywords because they had more volume, which actually made the SEO project work. Can we take a moment to think about this, right? We go for keywords in SEO that we wouldn’t go for, just in order to make sure the spreadsheet shows a positive expected ROI. Which is not going to happen after six months, because the keywords are useless.”

Three weeks of ads is the cheapest way out of that loop. It replaces an assumption with a receipt.

Sources & Relevant Episodes

• Gert Mellak, gertmellak.com – he described this as the method he reaches for when a client’s keyword choice is wrong and the argument is going nowhere. Full interview: Gert Mellak: Your Search Console Impressions Are Bots.
• Listen: Gert Mellak: Bot Impressions and the SEO ROI Myth.
• More from Gert: gertmellak.com · his writeup on why precise attribution is a myth.
• Related reading: SEO SOP: Find the Traffic You’re Leaving on the Table with the Expected-CTR Model – the forecasting math Gert argues against, worth reading next to this one.
• Part of the SEO SOP library.