I’ve had a rocky time in 2023 with Google updates on my main sites, and a few months back I wrote a newsletter detailing the main steps I was planning to take in order to recover them.
And in the last few weeks, good things have happened. Touch wood they continue.
In fact, here’s one of my main sites’ GSC:
And these are for keywords worth a lot of money, not just informational “what is a labrador”, but serious “best” keywords for high-ticket products.
So, let’s hope this holds.
Another site I run has had a fairly similar experience:
Again, these are extremely valuable keywords, not your average info content. So this increase is fantastic news.
So, here’s the main 3 things I did to recover my sites, so you can see if you think they would work for you, too.
1. Much more first-person work and more real product reviews
I had made the dumb decision to test a load of products, and then not project manage the other parts well and make sure the parts of “best” articles that these products featured in then also had the photos, and first-person words, and were re-written to reflect our real test.
So, we went back (and are still finishing the last few) and added over 140 mini reviews within these Top 10-style product round-ups within these posts.
Each probably takes 30 mins even if they’re only 200-250 words, because they’re designed to be super high-converting and segmented, but it’s worth the extra time.
We’re still rolling out the last 70 or so.
These are “When I tested it, I felt that the X was especially useful for doing Y, so based on Z, I highly recommend it.” Really precise, obscure yet helpful value judgements that make people think “Yes! This person understands what I’m looking for.” And then they buy.
2. Deleting the worst stuff (still more to do here)
I’ve learned a lot about clean index theory from Thomas Jepsen, and I agree with a lot of his theory.
I think that Helpful Content Updates are at least somewhat composed by a % of “unhelpful” content as determined by either their NLP or the MUM/BERT/Fred etc algorithms – but also from user metrics.
And if your oldest, out-of-date or just not really relevant content adds up, I think this can put you over the line for a helpful content penalty.
So, we rounded up all content ranking for less than 10 keywords, looked at the SERP and what the real search intent was, and if we missed the current intent and it wasn’t worth fixing it, we deleted it.
We also deleted anything we just subjectively felt was horrifically poor enough that we just wanted it gone. We can always republish on the topic later – it’s better to be drastic when your Google traffic is down 85%.
This accounted for roughly 10% of the index. We also noindex but did not delete some of the sponsored news content for new product announcements, so this further shrunk the index (I assume Google can tell what news is though, rather than punishing us for this? But can never be too careful).
3. Methodically improving content
So, originally we targeted the oldest content to improve, because it was most likely to be bad, and less “helpful” under the current 2023 meta of just having the answer in literally the first words of the article.
Even though these weren’t getting much traffic, and were unlikely to ever really drive significant traffic because either the SERP was now uber-competitive, or just had no real search volume or audience – we still improved them because we felt they were most likely to be unhelpful, and thus moving from a red to a green light for this piece would help.
While we have now mostly recovered, I’m not convinced this is the case.
I had a consultation with Dr Marie Haynes a few weeks ago, and she was incredibly helpful, and is a really great person who I am grateful to have spent an hour discussing ideas with.
I asked whether she recommended I continued going over these low-traffic, low-impact posts, and she recommended making sure the top 100 posts were improved and helpful as possible first.
I didn’t ask specifically, but I assume this feeds into a theory that there are weightings for content in the helpful content sensors, wherein your most important pages (closely associated with highest traffic?) have higher weightings in helpful content updates.
And therefore, you can have a higher impact improving the page experience on these already high-performing pages.
Which is great for me – it’s a great excuse to just do CRO on the top pages and try to convert more affiliate sales!
Nevertheless, it’s impossible to save for sure whether any of this moved the needle. And, I should have just gone harder with link building.
The smaller site that recovered did have a HARO campaign and we’re building link inserts at the moment, but I stayed away from that with the main site because it has such a strong link profile already – but I now think this was naive.
So, hopefully this is useful.
Next week I have some more cool stuff to discuss.