Recommendations for how to conduct keyword research?

[joemost]

I'm a little overwhelmed by all the tools out there for keyword research.
I am looking to research from tool ideas, my biggest gap is understanding where there's real opportunity in a niche, and understanding what the upside of that niche could be worth. Is there a rule of thumb for understanding this?

 

[Ryuzaki]

A lot of time people get lost at the very start because they're trying to research and find "any niche that has an opportunity" rather than choosing a niche for a good reason first and then determining if it has an opening. This is not seeing the forest for the trees. It's looking at the details before having a grasp on the big picture.

That may not be you, but you can still refine the process before ever starting by getting clear about things like:

• What's my vertical? What's my niche?
• Will I expand into other niches in the vertical?
• Am I going into a sub-niche? Will I expand into other sub-niches?
• Am I a publisher who will publish a ton of content?
• Or do I have a product or service that only needs a 20 page site?
• Who's my demographic? What's the annual "GDP" of the industry?

The more you're going to publish and scale in that direction, the less it matters. Start at the low competition keywords with respectable volume (100+ at least) and move up as you grow.

If you're going to have a 25 page static site, it's even simpler. You know the topics you need to cover. And if you're serious about the business the rest doesn't matter. Just find out what the keywords are with the volume, competition-levels-be-damned.

There's always opportunity anyways. If it's a huge vertical, then there's gobs of keywords to target that the big guys don't and tons of little angles to attack. There's tons of money flowing, so no question there.

If it's a small sub-niche, then the competition isn't going to be fierce anyways. The real question is can you make enough money for it to be worth it? Often, the same amount of work is required to make $1,000 in a tiny niche as it is to make $10,000 in a big one. Going small doesn't necessarily mean it's easier, when it comes to SEO.

The way to deal with the potential upside is to choose something that's simply way beyond your scope or ability to influence. If a niche has 5,000,000 keywords over 200 search volume per month, you don't need to worry about upside. It's bigger than you can dream.

If you choke yourself off by deciding you want to "knit fake glow-in-the-dark dreads for cyber-punk dancers" then your upside is pretty much "dead in the water".

You don't really need fancy mathematical formulas. But you can easily find data on how much money an industry generates every year if you need to dig that deep for whatever reason. But you can know in a 1 second glance if something is worth pursuing or not, in terms of SEO, which is what I'm assuming you're talking about.

 

[secretagentdad]

There really isnt much to look at that generalistically applies to diverse niches other than, Volume, Value, and Competition.
Most people seem to agree you're looking for a good ratio of volume and value relative to competition when doing keyword research. How you create this ratio doesnt really matter as the end result is you have keywords with relatively higher or lower numbers and you min max based on your priorities. The Keyword Golden Ratio is a popular industry label for when you compare search volume to the quantity of documents with the title in google via an all in title query and adwords data set.

You get volume by looking at adwords or search console. If you want to be truely accurate you gotta rank and see what there is in search console. Adwords volume is the best pre ranking proxy estimate at this time. You can also get volume from other platforms or smaller search engines like bing, but its not particularly accurate or useful for anything that's not on the higher end of the volume chart.
There are a lot of 3rd party tools that spout off about MAH Machine learning estimates that they create with old adwords data and questionably permissioned app data. I've found these to be nothing but deceptive and useless.

Value is a bit more subjective to measure but generally comes down to, how much do things cost? Is there recurring demand or is it a one off line item. Are their free or low cost substitutes? CPC is somewhat helpful as an indicator here but nowhere near as useful as the folks that peddle advice would lead you to believe. Counting ads, looking at shopping results or flagging value at the individual level can be indicative but its very hard to incorporate into a tool so most platforms don't. Humans can glance at this kinda stuff and jump to conclusions very accurately with a bit of experience.

Competition the 2 most common generalistic ways to measure are Document Volume and Top Cohort analysis. Everything else I've ever seen touted for measuring competition other than looking at the number of registered domain tlds for your keyword set is basically worthless hogwash sold by marketing people.

Document volume is looking at how many documents are targeting the words whether its mentions or in their document titles. More = Generally harder. Titles imply more competition than mentions alone. Short tail keywords with a wide variety of meanings can appear deceptively difficult when using these metrics.
You can get information that you can use in this way directly on google by running All In Title or "keyword here" type queries and looking at the number of results returned. You can also look at the volume of results in third party indexes that are not google such as the common crawl.

Top cohort analysis looks at volume and or velocity of new content and links by the top players for your keyword. It's generally only useful for evaluating the very top cohort group of keywords as it ends up being deceptive noise on lower difficulty keywords due to signals getting crossed and google just ranking big sites (with impressive metrics) due to lack of real topical relevancy.

You can also run queries to look at the number of registered domains in large ideas lists and make comparisons. However, the correlation ratios you can extract are sometimes very deceptive due to the domains market having some weird idiosyncracies and trends.

 

[Pagey]

I like to focus on a niche, identify its main entities. Get competitors keywords from ahrefs (even kws with 0 searches). Then I put them in a keyword clustering tools such as kw cupid. I check manually the clusters and create briefs from those that I hand out to my writers.