You know what really frustrates me? People who talk about Keyword Research as if it’s an exact science. Let me give anyone who needs it a “heads up”. Keyword research is not an exact science and shouldn’t be discussed like it is. It’s more of a black art based on skewed and inaccurate datasets. As a result it’s very easy to screw up.

Yet there seems to be an underlying assumption in most articles written about the subject that keyword data is accurate and can be completely relied on. This is a dangerous assumption. Let me give you an example that demonstrates this.

Take a look at what the leading keyword research data companies were saying on what they consider to be the top 10 keywords for the phrase “fly fishing”.

Here’s what they came back with.

Word Tracker
fly fishing
fly fishing alaska
fly fishing equipment
fly fishing gear
alaska fly fishing
alaska fly fishing trip
fly fishing reels
fly fishing canada
alaska fly in fishing
fly fishing in alaska

Overture
fly fishing
fly in fishing
fly fishing trip
fly fishing gear
colorado fly fishing
fly fishing vacation
montana fly fishing
fly fishing in colorado
fly fishing equipment
fly fishing reel

Keyword Discovery
fly fishing
fly fishing equipment
fly fishing gear
michigan fly fishing
fly fishing reels
fly fishing rod
fly fishing alaska
fly fishing reel
fly fishing and tying journal
fly fishing tackle

Keyword Intelligence
fly fishing
fly fishing knots
fly fishing flies
fly fishing tips
oregon fly fishing guides
fly fishing gear
fly fishing supplies
fly fishing rods
oregon fly fishing guide
fly fishing reels

Web CEO
fly fishing
fly fishing vacation
fly fishing equipment
fly in fishing ontario
fly in fishing
fly fishing instruction
montana fly fishing
fly fishing gear
pennsylvania fly fishing
new zealand fly fishing

Notice anything?

That’s right; none of them are the same. In fact, none of them are even close to being the same. So who’s right and who’s wrong?

Answer: they’re all wrong.

Let me break it down for you. If one tool says that there are 1246 searches a month for a particular keyword, that doesn’t mean you can say that during the past month the entire World Wide Web had 1246 searches.

What this figure means is that in this particular chunk or sample of data, that keyword was found 1246 times. The only time we could ever be sure that there were 1246 would be if the sample was every single search engine on the Internet and included 100% of their search volume. No tool does this - not even close.

To highlight this, Wordtracker (one of the leading keyword research companies) samples around 2% of the total Internet search volume. This sample size will provide some relatively stable data for popular keywords, but will proportionally become less accurate as the keyword popularity decreases. In simple terms, the less popular the keyword, the more likely that keyword data source will be wrong.

In fact you could argue statistically that there’ll be millions of keywords that are being searched for up to 50 times a month that won’t even register in the Wordtracker database.

It’s not important for you to know why, you just need to understand that the mathematics behind it say it’s so.

Now I’m not saying that to discredit Wordtracker. They do a fantastic job of keeping a massive keyword database that is arguably the most accurate in the industry. It’s just that unless your sample includes every single search, statistically, missing data is unavoidable. This problem goes across the board for all keyword research services.

So should we all just abandon keyword research tools because they aren’t 100% accurate? Not at all! It just means that we need to be more aware of the limitations of the data we use; especially when we’re researching within niche markets.

Niche marketers need to pay attention to this because they are often targeting the fringe keyword phrases. More often than not deep within the statistical long tail.

So keep in mind these keyword research tool limitations next time you think you have found yourself a hot niche. Here are two good ways I double check my keyword assumptions before I invest too much time in a niche market.

Cross reference with multiple keyword tools to see if they are all saying the same things. NicheBOT V2 allows you to do this easily as it provides data from the major keyword data sources.

Run an AdWords campaign for a couple of days with your keywords to see how many searches they will receive. Just remember that unless there are less than 8 ads for that keyword, the impression number will be lower as your ad doesn’t necessarily get shown for every search.
Remember, always use keyword data with skepticism. It’s almost always not 100% accurate.

Do not put your faith in what statistics say until you have carefully considered what they do not say.