In most cases, 404

Large ecom website 2m plus pages.

We do not have a 404 page. We use logical fallback.

Examples:

That is valid.

If you enter

You get 301 redirect to:

You get 301 redirect to

Falls back to valid directory.

Falls back to homepage as no directory.

---

This could be argued is a good UX as it keeps customers in the right area (category).

The idea from the manager and dev was that if people use a typo or share a link with a typo they will get to where they need to be (roughly).

I've asked a few clever people over recent months about this and cannot get a definitive answer (which is fine, I get it, this is SEO of course!).

Putting out there to the wider community, is there any case studies or factual evidence which would warrant me opening a ticket to change this?

One downside which got us recently was that our manager changed a load of URLs (10000s) as he changed a type name in the database. This means that loads of the previous URLs are now 301 to the brand page instead of the new final URL.

We do not have htaccess file. There is an xml file we use on .net which has about 2500k redirects which I can edit with the dev help. But I asked him about the 10000s and he said we can't put all them in there.

I would love to learn more about this and get some data.

Happy to answer any questions if you need more context!

Seems like there may be security risks with this approach?

EG I’m generally fine with most brand misspellings redirecting - you might have taken the time to redirect those manually.

But auto redirecting *everything* could mean you are redirecting intentional spam URL variations to your URLs, if/when you are the recipient of a spam attack.

EG I have one enterprise client with search results that default to a 200/okay and indexed URL (I know I know, this is an ongoing topic of conversation, but it’s got it’s benefits in addition to it’s issues) - that means that when they got spammy backs from crappy sites, at scale, for - well I’m not sure to what purpose, but it’s clearly intentional as it started with Korean kanji and evolved to emojis and other special characters.

We obviously noticed - but this is not spam you generally want to 301 as then you are sort of saying it’s a valid URL - which it is surely not. We 410'd these, and are considering 403ing them, as the scale is MASSIVE and junking up indexing/our otherwise valid GSC reports.

But my point is - if you auto redirect everything you’d also validate the spam attack and there could be security/spam/malware concerns as an outcome.

@tory excellent information, thanks. I had not thought about that.

It has been recommended to me from a very clever SEO to check log files (which I don't have access to). My brain has clicked, maybe I'm wrong, that if we find a huge scale of these wrong URLs that would indicate spam attack.

In which case, yes, what you say makes sense re: unintentionally adding validation!

I'm assuming your site is not creating the redirected links, and suggest you check your logs and backlinks to see if it's necessary. Separately check if can add a Regex redirect in the xml redirect file for the changes type name.

This also sounds like a potential for someone to massively spike server load/DDOS if they wanted to and figured out your pattern.

Just make a useful programmatic content 404 page that gives visitors the most relevant next options to not fully dead end them.

Ask yourself what happens if you kill all those redirects. Ask yourself what technical debt you're creating for yourself with all those redirects.

To me, there are rational scenarios where covering bases with redirects makes total sense, but I don't see a lot of business rationale behind managing THIS MANY redirects. With the tech risks, and the level of effort.

If it were my site, I'd likely (obviously there are caveats) do the following:

1. Delete all redirects that are more than 1 year old. Yes, all. There are likely thousands there that some business person would say, "But we need that!" and you do not, in fact, need that.

2. Audit remaining redirects for business need. Do you need a 1:1 redirect? Do they REALLY need it?

3. If the answer is "yes, the user absolutely needs to end up on a specific page with no extra steps" then redirect. Examples of this might be customer service, pages attached to legal compliance, etc.

4. A lot of the time, the answer will be no. This is ESPECIALLY true in ecommerce with high inventory turnover and product seasonality. In these cases, a high quality 404 with user-friendly next steps (search bar, secondary menu options, etc.) captured most potential.

Thanks so much @bill and thanks as always @mika

Yes food for thought.

Great options. I'll review log files (I asked for them earlier and got one!).

It is 2.1gb FYI

Never analysed one before so looking forward to that next week!

You are in for a treat. It's the geekiest SEO geek shit you can do and I love that for you.

Thank you @kellystanze for the input.

We manage about 2500k in an XML.

They are old ones but some are external and lots were from our GSC indexing report which Google crawls frequently.

The redirects we do using the logic are not managed.

What sort of rule structure drives the logic-based redirects?

To me, logic-based redirects can be a great tool but I'd be very particular about them. And the downside to them is that they generally never get cleaned up after they're created, which to the others' points generates tech debt and security risk.

Most major ecommerce platforms generate dynamic redirects anytime a product is no longer in circulation. (I think this is basically what you mean by "logic driven redirects" but I could be wrong.)

And none of the ecommerce companies I've worked with have had a process for removing these old redirects when they've fallen from relevancy -- until I told them that they. needed to do housekeeping.

RP's site is home grown.

@mika I missed that part. Thank you!

I think functionally it's the same, though? Redirects are built based on logic, but likely never cleaned up?

(Can I just say how impressive it is to see a home grown ecomm site? Like...dang. That's a big undertaking.)

He may not have mentioned it, but I know some about his scenario from talking over time.

Yea, similar issues regardless.

Yeah, that is a pretty large lift. It's on MS as well!

That's helpful context as I'm sitting here thinking about how I'd manage it in Salesforce Commerce Cloud, Woocommerce, or Shopify. While the tasks I'd recommend and the risks I'd mention are the same, I have no idea how that would look to execute.

Now I'm going to get off-topic because my brain is all wrapped up in homebrew ecommerce.

Yes all bespoke using .net.

The logic is basically: "fallback to most likely area of website"

I could be wrong, but I think it uses some sort of "like" query.

You guys have been super helpful as always, thank you so much.

I will review the IIS web server logs and ask the dev exactly what the redirect logic is and I will let you know!