eCommerce site structure - best use of canonical tag?

[FranticFish]

eCommerce rebuild for a niche website with >1000 different SKUs with extremely little variation, and the following considerations when it comes to making a large number of product pages sufficiently distinct and of reasonable quality:

1) Time
Even 30m per SKU is two months worth of 8 hour days.

2) SKU overlap / similarity
Obviously I can't share details, you'll just have to trust me when I say that:
- the differences between multiple SKUs, even if explained in detail, are very slight and could be reasonably addressed in a sentence; and
- all of these differences are really the same difference.
So, more than the time to manually write the text, it's the planning involved to try preserve distinctness and not failing miserably.
Even with a huge amount of planning and effort we are STILL going to end up with hundreds and hundreds of near duplicate pages.

3) There's no point anyway
This is really the killer one. No-one searches for the SKUs - either by name, or by their usage at the individual SKU level.

However there ARE many different ways that people search for GROUPS of products. So we've decided that our category pages are the ones to lavish attention on. We CAN make these distinct, and we can provide quality information on and around them.

I don't want to use 'noindex' for the product pages, so I see my options as follows:

1) Let them be indexed and accept 1,000s of pages of very thin content, albeit with strong user signals (it's a busy site with a good email subscriber base).

2) Set a canonical tag on each product's page to its category
Products do sometimes span categories however.

3) Set a canonical tag on all product pages to a master product page e.g. example.com/product
There is essentially one product page.

I'm pretty much decided against (1) but I'm not sure what (if any) pros and cons there are between (2) and (3).

Has anyone done anything like this before? Even if you have not, any and all advice very welcome.

[JesterMagic]

I am no expert in eCommerce sites but do have a few thoughts.

Question: It sounds like the old site performed fairly well. How was it setup in terms of the questions you are asking now?

When changing your URL structure you will want to make sure to redirect the pages that have good backlinks to their new pages. You don't want to loose those ranking signals.

I know you mentioned you have categories but products can span multiple categories... Can very similar products be grouped together? For example Widget A comes in many different colors and sizes and each combination has a different SKU. Each page for these SKUs has the same description but the color and size fields change. Can you pick the most popular SKU in this group and make it the default canonical for all the rest of the SKUs? Then you can do the same for Widget B, etc...

Or does this not work because Widget A and Widget B themselves are very similar?

[FranticFish]

The site currently does (1)

It's been in a slow death situation for years (hit by Panda v1, released, looks to have been hit by Hummingbird, hit by mobile-first indexing). In the last twelve months organic sales have ceased and It currently has sales from direct traffic and email only, although it is still just inside top ten for two trophy terms). It has a strong legacy user base gong back to the 90s but is not winning any new customers.

Yep we will be doing page to page 301s regardless.

Can very similar products be grouped together?

Yes, in some case, but this would still still only reduce page numbers by perhaps 10%.

[n0tSEO]

In the last twelve months organic sales have ceased

Have you looked into this? If so, what have you found? Because something must have happened for organic sales to cease completely, like a penalty or an algo upgrade that suddenly made product pages less relevant.

You say that searchers seem to be interested in groups of products rather than individual products. Can you give us a little more information about this?

[FranticFish]

The sales Analytics setup is not perfect. The shopping cart is on another domain and cross-domain configuration is not in place, so all traffic shows as Direct. However eCommerce is enabled in Analytics and that shows:
- referral sales fell by 75% in July 2018 and then slowly tricked down to next to nothing by Summer 2019
- organic sales saw a spike in March/April 2019 (their best year for some time) and then dropped by 90% in May 2019 and then to nothing in June 2019
They still make one or two sales from these channels but we're talking 10-15 in the last twelve months (and they're high volume low price low margin).

I can find no smoking gun for either of these.The 'Medic' update took place around the time their referral sales traffic died. They do have an old link profile (90s) and perhaps their pipeline lost authority.Apparently there was a core update in June 2019, but it doesn't really match. With all the other generally confirmed updates (Panda most notably) I can either find clear correlation or no correlation - but not a sketchy correlation..

I do not have budget for a detailed link audit (part of our proposal but not approved), just quick bench-marking via a well-known provider that allows limited free access to some of their metrics. However I'm reasonably confident that the site's link profile is not regarded as toxic. They were not hit by Penguin and still have two page 1 rankings for decent terms. Their competitors all have (i) more authority score, (ii) more linking domains, and (iii) more fresh links. So I'm guessing that their profile cannot be too shabby and that the numbers are not too representative of their actual 'clout' in Google (indeed, looking at just the numbers they stand out like a sore thumb compared to their competitors).

However the key point is that we're not tied to what they've done before. We're basically looking to start again with the authority left in the domain and to take whatever content hasn't been scraped, repurpose it, write a load more. We're rejigging the categories and adding a knowledge base which will be interlinked with them.

Re: groups of products, it's hard without going into specifics. All I can say is that for most products we'd be hard pressed to write more than a short paragraph, and half the customers would probably disagree with it too! Then, once we got to 100 products tops we'd run out of different ways to say the same thing. There are about 30-40 possible categories we could use, and we could do a good job on these.

The more I think about it the more I'm inclined to use canonical to the first category page the product belongs to (less than 5% of inventory sits across two categories). Would welcome thoughts on this from anyone but especially from anyone that's used canonical 'creatively' in a catalogue situation.

[JesterMagic]

When canonicals are not present on the actual site Google try's to determine your own canonical, These are often wrong. Just curious when you search for different products what pages of your site does Google return? Maybe there is an indication their what you need to do?

In some cases Google knows what they are doing and maybe they handle ecommerce sites a little differently since they know a lot of sites out there may be in a similar situation as yours.

One thing I would do is use structured data for each of your product pages. You can include SKU's and other identifying attributes and this in itself would be another signal that each of these pages is for a separate product. I would then include a canonical for each product page even if they are only slightly different. In reality this is technically what you should be really doing and the only reason you are not is because you believe (and maybe rightly so) that Google will not figure it out.

If after setting this up and it doesn't work after a few months you can then reevaluate and your only option at this point IMO is keep the structured data but then figure out the best way to group your products into groups for the canonical URL. Just make sure it is possible for a user who lands on the canonical URL that they can reach all of the other product pages that use the same canonical URL in one click.

You mention your site currently uses #1 but didn't indicate that it uses canonical URLs or structured data. My solution is similar to 1 but with the extra signals hopefully it will improve your rankings.

[FranticFish]

Thanks for reply and sorry for slow response.

Most of the product pages are not indexed currently, and there is no canonical in place or Schema. We will be using Schema markup via JSON.

I'm not really interested in carrying anything the old site does into the new one (except its external links). The site is dead in the water organically and has been for over a year.

Thanks again for your thoughts. It appears no one here has tried this :)