I disagree with the concept present in other answers that search engines provide "curated" data that would be "safe". I however concede they have a point in that web search links:
- Are ranked, and the user will barely pass the first few results. Reputation of top results (accrued through time) is probably orders of magnitude bigger than most phishing pages (which are shortly lived).
- I expect <phishing inbox mails> / <total mail> will be higher than <phishing results> / <total indexed results> for most searches. (Proper) Web result sorting will make much more hard that it is clicked (a wrong sorting that put the phishing on the top would be really harmful, though), whereas likelihood of reaching the malicious link (prior to opening/filling the phishing form) will probably be equiprobable to the proportion of email in the INBOX/Spam folder (depending where it ends up) that the user reads. Up to 100% of being found by the user.
- Search engines can dynamically remove search results. If they had a result for the user query that would have been shown, but it is now on a blacklist [they pay attention to], or simply they received enough user feedback to reconsider it, they can skip it on future searches of that term. While email providers do sometimes move to spam folder emails that had already been delivered into the inbox, if they were not seen by the user (and some companies even completely remove already received phishing mails from user mailboxes) generally, once delivered, the mail stays there, with the classification it got at time of reception.
However, I don't think those would be as important as the context in which they are framed.
First of all, I'm not completely sold that an email link is more dangerous than a web search one. This would probably be interesting for a study. Ultimately, the one that ends up affecting you is the one that was most dangerous for you (this time). You need to keep an eye on all fronts.
Why do I talk about the contexts of links coming from emails vs web searches? Typical fraudulent emails have lures such as:
- New invoice
- Your mailbox is full
- Protected email
- Hello adamshakhabov5
- Bank account blocked
Phishing links flourish here, since receiving an email link from which you need to authenticate is a normal action. You need to enter your credentials to access your mail, your bank could send you notices by email (at least it would be conceivable that it did), and even the weird phishing pages where you need to log in with any email account to "download" the document have a fraction of plausibility (if you ignore their many telltale signs). on the the other hand, when were you last asked to provide your credentials (legitimately or not) as a result of a web search? It should raise much more suspicion.¹ They have no rationale to need your email password.
For malicious web search results or advertisements, you will mostly find things like fraudulent shops selling counterfeit goods. That's a model that fits much better the profile of a web search: the user wants <product> and gets to a web page claiming to sell it with a 80% discount ('surprisingly' they charge a different amount than advertised, the product is different to what was expected, it directly never arrives...).
Maybe it is related to the fact that links in an email may contain a more personal attack malicious to you or your company?
It depends. For instance, some companies would worried about attacks directed at them but not concerned about other malicious links not affecting them. An employee sharing their credentials to the company could lead to a leak of confidential data, access by an attacker to their systems, etc. but not consider an issue that an employee lost some money by buying a pair of fake shoes (in fact, they probably shouldn't have been buying shoes on company time).
Email would be an obvious point of entry for an attacker that tried to penetrate the company IT security, whereas getting their search engine results to lead to them would be harder, and it may be expected that you would have to be either the search engine or the ISP (not that you couldn't indirectly target specific people, though)
I would be wary of approaches that focused on specific threats and then neglected compromise sources not typically used by actors you are concerned about, though. Nowadays, nobody can be considered safe thinking they will be ignored. They will be compromised first, then sold to someone willing to pay for access to that company, not the other way around (while such focused targeting might still happen in some cases).
¹ Interestingly, this could change if "log in with Facebook/Google" options were much more widespread.