Most keyword lists I review are about 70% noise. Not what I'd call 'slightly off-target,' either. These are queries that should never make it into a content brief. When they do, the article fails before a single word gets written. Good writing can't rescue a brief built on the wrong intent.
I'm the quality/compliance manager at Valor, a building-products company that makes roofing systems and gas fireplaces. I review everything that carries our name: blog posts, spec sheets, installation guides, brochures for our exterior partners network. About 200+ assets a year. When we approve printed partner materials, I verify that images are 300 DPI and the logo color falls within our Pantone tolerance. Yes, I'm that person. So I notice patterns.
In our Q1 2024 quality audit, content built from unfiltered keyword lists failed first review 19% of the time. Content whose keywords passed a simple intent check failed only 6% of the time. Comparable writers. Same review process. Different starting point. That gap is why I review keywords before we commission any content. Most content quality problems start before the writer opens a document.
Why raw keyword lists fill up with junk
Because SEO tools count queries. They don't judge intent. If you export a list for our brand name, the platform might also include the motor yacht that shares the name, a screen-protector brand, a massage franchise, and a Windows tutorial—and it won't flag any of it as irrelevant. To the tool, they're all just keyword strings.
So the export fills with everything: brand queries, brand collisions, comparison shopping, tutorial requests, typos, and one-off questions from people looking for something else entirely. Raw data is not strategy. Sorting it is.
Six keywords, reviewed the way I'd review yours
Here's a set from a recent export. See how you'd sort them:
- valor
- valor mold removal
- motor yacht valor
- magic john screen protector
- hand and stone
- how to snip on windows
'Valor' is an easy yes. It's our brand, and we need pages that reinforce it.
'Valor mold removal' isn't an automatic yes. Some homeowners type that when moisture appears near a Valor gas fireplace; that's a real maintenance topic—condensation, venting, clearances. But 'Valor' is also a motor yacht, and boats get mold too. So I'd check the actual search results before approving a brief. That five-minute check has prevented many wrong turns.
'Motor yacht valor' — we don't build boats. Drop.
'Magic john screen protector' is consumer electronics. Drop.
'Hand and stone' is a massage franchise, not a building product. Drop.
'How to snip on windows' tricks people in our industry. Roofers use snips, and windows are literally in the product name, so it looks close. The query is really about taking screenshots in Microsoft Windows. There's no roofing angle. Drop.
So out of six terms, one is clearly ours, one needs investigation, and four are noise. That's about two-thirds junk, right in the range I mentioned.
Here's the counterintuitive part. Dropping those terms feels like shrinking your strategy. It actually strengthens it. Publishing a useful article about Windows screenshots on a roof-and-fireplace website tells search engines nothing, and it tells human readers we don't know who we are. Saying no to noise is a positioning decision, not a loss.
What it cost us before we had a process
In 2022, I assumed a raw keyword export had already been filtered for intent. It hadn't. We sent it to an agency with a short note: 'write articles for these terms.' Among them was 'how to snip on windows.'
The agency delivered a clean, well-written tutorial about the Windows Snipping Tool. Technically, it was exactly what the keyword asked for. It was also useless to a company that sells roofing systems and gas fireplaces. We couldn't publish it. We'd already paid for it.
The surprise wasn't that the writer followed the keyword list—that's what we asked them to do. It was that the article was well done: a quality piece for an audience we don't have. I do not blame the writer. I blame the brief.
That detour cost about $22,000 in agency fees and delayed a content launch by six weeks. We didn't have a formal keyword vetting process yet. It took a third near-miss before I built a checklist. It should have been after the first one.
The five-question vet we use now
It takes about 15 minutes per batch. Each term has to get five yes answers, or it stays off the brief:
- Does the search point to a problem our customers actually have?
- Can we answer from experience, or would we be guessing?
- Does it fit roofing systems, gas fireplaces, or exterior building products?
- Would a contractor, builder, or homeowner trust a manufacturer on this topic?
- Is this 'Valor' search about us—or about the yacht or another brand with the same name?
Question five isn't a joke. Brand-name collisions are real. 'Motor yacht valor' can show healthy search volume, but none of those searchers are buying a fireplace insert. If our name keeps appearing next to unrelated terms, we flag it for the brand team. That's a reputation issue, not an article idea.
What this checklist won't do
It won't replace keyword research. You still need raw lists to hear what people ask. But a list is raw material, not a finished plan. A low-volume phrase like 'Valor fireplace condensation' often beats a high-volume term that doesn't fit the business.
I have mixed feelings about keyword tools, honestly. On one hand, they surface questions we wouldn't think to answer. On the other, they have no sense of identity—they can't tell the difference between a fireplace manufacturer and a yacht charter company. Part of me wants to trust the numbers. Another part remembers the $22,000 Windows tutorial. Our compromise: use the tools for discovery, apply human judgment before commissioning.
The vet also depends on the person running it. If someone doesn't know what we sell, the checklist won't help. Each content brief now includes a short brand glossary and audience definition, because context matters as much as the questions.
One exception: sometimes unrelated terms deserve attention, just not on the editorial calendar. If 'motor yacht valor' suddenly spikes, that's a brand-confusion signal. It goes to the trademark and brand team, not to a writer.
Bottom line: content quality starts before the first draft. You can hire better writers, but if the keyword list points them at the wrong topic, they'll produce excellent articles nobody asked for. Five minutes of keyword review is cheaper than five weeks of rewriting. It's the closest thing to a no-brainer in content operations.
