Not All Searches Are Equal, and Pretending They Are Will Cost You (Part 2)

This is the second part of a three-part series. You can read the first part "SparkToro Says Google Has 73% of Search And I Have Questions (Part 1)" here

In the first part of this series, I looked at SparkToro's headline finding that Google holds 73.7% of desktop search share across 41 major websites, and why that number needs serious context. Now I want to dig into the bigger conceptual problem: the idea that all searches are created equal.

SparkToro's research explicitly states: "This report treats all searches and prompt sessions equally." An Amazon search for "wireless earbuds under $50" is counted the same as a Google search for "how to fix Bluetooth connectivity issues" or a ChatGPT prompt asking "explain Bluetooth Low Energy protocol to me like I'm five." Same unit. Same weight.

I understand why they did this. If you want to map the total landscape of search behaviour, you need a common denominator. But the moment you start drawing strategic conclusions from that map, the equivalence falls apart.

The Intent Problem

I have written before about understanding search intent as the key to effective SEO, and this is where that thinking becomes critical.

Search intent is not a nice-to-know academic concept. It is the thing that determines whether your content, your product page, or your ad actually connects with a real human being at the moment they need it. And different platforms serve fundamentally different intents.

Amazon search is almost entirely transactional. The person has already decided they want to buy something. They are narrowing options. The conversion funnel is short. The question is "which one?" not "should I?"

YouTube search is a mix of entertainment, education, and how-to. The person wants to watch something. They might learn, they might buy something later, but the immediate intent is consumption of video content.

Google search spans the entire intent spectrum: informational, navigational, commercial investigation, and transactional. It is the broadest intent platform by far.

ChatGPT prompts are something else entirely. Many prompts are not searches at all. They are content creation tasks, coding assistance, brainstorming sessions, or conversation. SparkToro's own research notes that "AI tool prompts can lead to long, complex conversations, many with goals orthogonal to search (programming, image generation, content creation, etc.)" and then proceeds to count them as searches anyway.

Putting these in the same bucket and saying "Amazon has more search activity than ChatGPT" is technically a statement you can make with the data, but what does it mean for your business? Not much, unless you understand the completely different user states behind those searches.

What SparkToro Gets Right (And It Matters)

Here is where I want to be fair, because the underlying insight is genuinely useful.

SparkToro's core thesis, which Rand articulates as "search is a behavior, not a channel," has real merit. For years, the SEO industry has been overly fixated on Google rankings as the primary metric of online visibility. SparkToro's data, even with its methodological issues, demonstrates that significant search behaviour happens on platforms that many marketers ignore.

This tracks with what I have been observing. When we covered the zero-click era and how it is reshaping search marketing, one of the key points was that younger generations do not consider "Google search" as the default. They search on Instagram, TikTok, YouTube, and increasingly through AI tools. The user journey has fragmented.

And in our research on AI curating buying decisions, we found that 43.5% of B2B product selection stakeholders in Japan are already using AI search, while 90.6% still use traditional web search. The channels are multiplying, not replacing each other.

So yes, search happens everywhere. That part is right.

Where The Equivalence Breaks Down For Marketers

The problem comes when you try to use SparkToro's market share percentages to make budget decisions.

Consider their finding that Amazon, Bing, and YouTube each receive more desktop search activity than ChatGPT. SparkToro presents this as evidence that marketers are over-investing in AI search optimisation and under-investing in these platforms.

But think about what that means in practice:

If you sell physical products, Amazon search share is incredibly relevant to you. You should absolutely be optimising your Amazon listings. But this was true before SparkToro's research and will be true regardless of what percentage of total "search" Amazon represents.

If you sell B2B services, Amazon's search share is irrelevant. Nobody is searching Amazon for "enterprise CRM implementation partner." Your search landscape is Google, LinkedIn (which SparkToro's data shows has surprisingly low search activity), and increasingly AI tools.

If you are a content publisher, YouTube's search share matters, but YouTube SEO is a completely different discipline from Google SEO or Amazon listing optimisation. You cannot simply "redirect" effort from one to the other.

The percentage of total search each platform holds tells you nothing about whether that platform is relevant to your specific business. This is something I keep coming back to: as I argued in our article on the Google Zero myth, the right approach is platform diversification based on where your actual audience spends time, not based on aggregate market share numbers.

The "Search Everywhere Optimisation" Trap

SparkToro has been pushing the idea that SEO should mean "Search Everywhere Optimisation". It is a catchy rebrand. But it risks creating a new kind of confusion.

Optimising for Google is a discipline with decades of established practice, clear metrics, and well-understood mechanics. Optimising for Amazon is a different discipline. YouTube is another. Pinterest is yet another. Each has different algorithms, different ranking factors, different content formats, and different user expectations.

I will give a personal example. When we at Kafkai started thinking seriously about multi-platform visibility, the instinct was to try to be everywhere. Optimise the website for Google SEO. Post on LinkedIn. Create YouTube content. Show up in AI citations. The result was predictable: we spread ourselves thin and did none of them particularly well. What actually moved the needle was narrowing down to the 2-3 channels where our specific audience (SME marketers and business owners interested in AI and content strategy) actually spends time, and going deep.

Telling a small business owner "you need to optimise for search everywhere" without acknowledging that each platform requires different expertise, different content, and different resources is not helpful. It is overwhelming. An SME with a two-person marketing team just cannot simultaneously master Google SEO, Amazon listings, YouTube optimisation, Pinterest strategy, and AI citation building. Pretending otherwise sets them up for failure.

What is actually useful is helping businesses identify which 2-3 platforms matter most for their specific audience and going deep on those, rather than spreading thin across all 41 sites in SparkToro's study.

This is essentially what we recommended in Part 2 of the AJSA seminar report: focus on doing a few things well rather than trying to be everywhere at once. The four-channel strategy we outlined in that article (Google SEO, Map Engine Optimisation, Social Media, and Answer Engine Optimisation) is a practical framework that acknowledges multi-platform search without pretending you can master everything simultaneously.

What To Actually Do With This Information

1. Stop treating aggregate search data as a strategy. SparkToro's 73.7% figure is interesting for industry analysis. It is not a blueprint for your marketing budget. Your customers are not an average of all internet users. Find where your specific audience searches and focus there. Differentiation starts with competitive intelligence: knowing which keywords, content gaps, and platforms your competitors have missed so you can focus where it actually matters. That is the problem Kafkai solves, turning the noise of multi-platform search data into actionable shortlists you can act on this week.

2. Recognise the intent behind each platform. If your business depends on informational queries that lead to trust-building and eventual conversion, Google (and increasingly AI tools) is your primary battlefield. If your business depends on product discovery, Amazon and social platforms might matter more. Match the platform to the intent your business serves.

3. Invest in understanding AI search, but proportionally. SparkToro's data shows AI tools at 3.2% of desktop search. That is small. But as I wrote in our article on GEO not being optional, the structural shift matters more than the current number. AI search is growing, and the businesses that prepare now will have an advantage. But "prepare" does not mean "panic and redirect your entire budget."

4. Build content that works across platforms. Rather than trying to optimise separately for every platform, focus on creating genuinely useful content backed by original data, real experience, and clear structure. This is the one strategy that works everywhere, because every platform (and every AI engine) is converging on the same principle: reward content that actually helps the user.

The Meta-Lesson

SparkToro's research is a good example of something I see a lot in the marketing industry: data that is directionally correct but presented in a way that overstates the actionable implications. The underlying observation (search is diversifying) is valid. The specific numbers (73.7%, 3.2% for AI, etc.) are methodology-dependent artifacts that should not drive strategy.

The best response is to acknowledge the trend, understand the limitations of the data, and make proportional investments based on where your specific audience actually spends their time.

In the final part of this series, I will look at what SparkToro's research tells us (and does not tell us) about AI search specifically, and why the current small numbers might be hiding a much bigger shift.