The inaccuracy and flaws of search volume
The inaccuracy and flaws of search volume
Search volume is a double-edged sword. On one side, it guides our strategy. On the other side, it's so fundamentally flawed that we should handle it with extreme care. In this article, I explain why and how.
Search volume is a double-edged sword. On one side, it guides our strategy. On the other side, it's so fundamentally flawed that we should handle it with extreme care.
The keyword research process is broken
Let me start with a couple of examples.
This site ranks for the query "old christmas traditions", and got 145 clicks and 513 impressions in October 2020. That's a full month ranking on position #1, so I take it as the best source of truth for the keyword's search volume and traffic potential.
Ahrefs shows a search volume of 150, SEMrush of 140.
Had I research the keyword and then projected how much traffic I could get for it, I'd probably underestimate its potential. But here comes the kicker: that's just one keyword. The whole page ranked for 14 keywords in October and got 281 clicks and 2,840 impressions. Had I just looked at that one keyword, I would have underprojected traffic by 50%.
The oldschool keyword research process of creating a list of keywords and creating content for it based on the top keyword by search volume is flawed and has been for a while.
I'll give you two more examples.
The site I show above ranked for "dog poop colors" (formidable keyword) on #1 in October and got 114 clicks and 225 impressions in Google Search Console. Ahrefs shows a volume of 1,200, SEMrush of 2,400 monthly searches.
Lastly, the same site ranked #1 in October for "stripping sheets" and got 1,371 clicks and 2,794 impressions. Ahrefs returns a search volume of 0, SEMrush of 590.
The whole page gets 8,100 clicks and 166,000 impressions, though! That's a massive difference! None of the data from rank tracking tools would have pushed me to create content and optimize for this keyword.
That's not because these tools are bad, but because accurate search volume is hard to come by. The best solution would be to create a massive database of anonymized Search Console data.
But there is another problem.
Mobile vs. Desktop
When comparing GSC metrics by device, we see that clicks and impressions on Desktop are about 10% of mobile. In other words: people search for "stripping sheets" predominantly on smartphones.
That's another fundamental problem that search volume doesn't deliver: search volume on mobile and desktop is different. SEMrush does provide Desktop and Mobile search volume but the values aren't different from each other.
The many flaws of search volume
In summary, search volume is:
not available for many keywords, especially transactional keywords
often inaccurate
averaged over the year, which means that seasonality is not reflected at all
not differentiated for mobile or desktop
not indicating whether search volume is increase or decreasing over time
The last point is specifically important and under-discussed in SEO: what keywords gain or lose demand over time? Finding keywords that are growing in search volume means you can start ranking for them when they're less competitive and "grow with the market".
The solution?
What can we do? I don't have the perfect answer, but I have options.
We can ignore search volume and just create content based on expertise, market research, and product development. However, that comes with organizational challenges because, without an idea of how much traffic we get in return for creating content, we're on thin ice.
Alternatively, we can create content, see what it ranks for, and then optimize it. You would monitor the queries that the content gains traction for, and then adjust the content accordingly. That's an approach I personally like very much, but you also need to consider organizational challenges here. You basically need a blank cheque, and that's hard to come by.
We can also use several tools in combination that indicate search demand and take platforms outside of Google Search into account (I created a list). The challenge here is to scale keyword research for large sets across several tools, but it's not impossible.
I also suggest creating a database of keywords you ranked #1 on for a month. This allows you to compare clicks and impressions with keyword volume from rank trackers and apply the difference to other keywords you don't rank #1 for yet. That works best when you compare with keywords with the same syntax, e.g. "best {product}" or "cheap {product} for men".
Ahrefs' idea of page traffic as a metric is a good start, but we need to evolve further and beyond simple search volume.