Hey everyone,
Araminta here.
In May of this year, one of our best performing payment clients saw a 30% drop in their leads 😱.
This was right after the May Google Core Update, and this was a company that we had helped get from 0 inbound leads to 30+ per month. But after this update, they were now back to under 20 per month (a lead is worth $100k+ for them).
We knew our strategy and the content itself was strong, but we also knew we had to investigate and figure out what had and happened and how to fix this.
We immediately got into action. Here’s the steps we took to fix this:
1. Diagnose the problem: what happened to our articles?
The first step was understanding what happened. Since we had access to the client’s HubSpot, we already knew which pages were helping bring in the most leads and we knew which keyphrases they were ranking for. In other words, we knew which were “high impact”.
It didn’t take us long to find out what had happened: those high impact pages were no longer ranking as high as they used to, for seemingly no reason.
For example, one of the highest performing articles was a “[competitor] alternatives” piece. Once that stopped ranking on the first page, the number of leads decreased (on a side note, it’s cool when you can attribute leads and customers so cleanly to one specific piece of content).
You may ask: do people really still Google? How could this have such a big impact on overall leads?
This company gets about 50% of their leads through Google, and 50% through AI search, but they’re intrinsically linked. Ranking high on Google impacts your LLM visibility; LLMs scrape Google and ranking higher usually means you’re more likely to get cited.
That’s why ranking low on Google impacted both organic and LLM leads, and why we saw such a decrease.
So we made a list of the articles that had lost rankings and prioritized them by the ones that had brought in the most amount of leads in the past. For example, we noticed that competitors posts – some of the highest converting pieces – had been hit the hardest.
We also made a list of GPT articles that were cited the most. Interestingly, several of the GPT articles we had created with the main purpose of increasing LLM visibility had actually started ranking for solid keywords, bringing in clicks and leads. We noticed some of these had also been performing worse organically after the update.
2. Optimize and improve: making the articles more useful to both Google and readers
Once we had a list of articles, our strategist, Elliot McGuire, set to work on improving them. This meant:
- Changing up the format of the articles more often. Good content often follows a familiar pattern, but there are plenty of ways to mix up the formatting: adding tables, diagrams, using more unique examples, and so on.
- Adding more useful / informational sections so the articles are not too salesy. It’s important to talk about your product, but it shouldn’t make up the bulk of your blogs (this is a mistake we made with some of the earlier GPT articles).
- Adding more internal links and backlinks: both of which are crucial for the overall SEO health of any website, and should you give the best chance to rank.
- Changing up the authors. Publishing a large volume of articles under the same author could make the content feel templated, or mass-produced. Mixing up the authors from time to time is a good way to make the articles feel more authentic: both to readers and to Google.
- Adding quotes from Subject Matter Experts. This is important for adding originality to the content: this is something that an AI tool can’t replicate.
- Mixing up the dates. We tend to create GPT articles in blocks of 5 or 10. To avoid looking mass-produced (which they are not, but Google can take that as an incorrect signal), it’s worth making sure you’re not publishing too many blogs on the same date. While we don’t have concrete evidence this works, it’s such a simple precaution to take, it’s worth doing in general.
At the time in May, marketers in the industry were saying that this recent Google update was targeting websites that were publishing a lot of “listicles” – the types that start with “Top payment providers in healthcare”, mention themselves as number 1 and briefly mention competitors.
But we knew this was not true, mainly because we’ve been writing listicles for years and all our clients were still doing well even though they had published listicles. For many clients, they are our best performing content.
It’s not that listicles are inherently bad, it’s low quality content that is bad. You can write a listicle that genuinely helps the reader. What Google penalizes is if you publish hundreds of listicles that are very thin on content, too formulaic, and don’t actually help the reader.
(You can read more about this here: Are self-promotional listicles ruining your SEO?)
While we knew our comparison articles were of good quality, we wanted to be safe and update them anyway. Content is this client’s major source of pipeline so it’s crucial the content keeps performing.
To update them:
- We made them less salesy. It’s crucial you give a good account of your product to your reader, but in some of the GPT articles, we realised that we spent too long talking about the client. To fix this, we added some more genuinely useful content, such as helpful sections like “How to pick the right provider’, or ‘pros and cons of each solution’.
- We also made edits to minimize bias. For instance, by not ranking our client as ‘no.1’ in all the ‘best of’ listicles. It’s a small change, but one that some SEOs have argued can make a big difference in Google’s perception of the content. We also ensured the comparison was balanced by pointing out our competitors positives. For instance, if we were comparing to Stripe, we’d mention how their documentation is exceptional and that they are a great starting point for many businesses. This shows confidence and builds trust with the reader and in theory, should also help Google see us as a reliable, useful source of information.
- We made sure every rewrite was done by a human writer, with no AI involved. While Google says they don’t penalize AI content specifically, AI content just always feels…AI. By design it will always be programmatic, and each article inevitably sounds like the other. In the interest of making sure our content was as good as possible, and to avoid any hints of ‘mass published’ spammy content, we made sure to have a human handle the writing here.
3. Track and monitor closely
It took about 4 weeks to complete all these edits on the highest priority articles, and we’re happy to say that by week 2 of June we could already see an improvement.
The rankings of the most important keyphrases were back up, and the leads went back up to the same levels they were at before – in fact hitting a new high in July.
Even though traffic is not the main metric we track to evaluate success, it was also good to see traffic grow back up to new highs.
You can see a small dip in April / May, and then how traffic grew again afterwards.
Blog traffic in particular hit a new high after we made some of these content improvements. Core update or not: it shows the value of going back and improving your existing blogs!
Google Core Updates are real and you want to be prepared – however if you have good content any performance dips can be fixed
The dip in leads and rankings really made us aware of how careful you have to be with content.
For this client, it was a matter of quickly diagnosing the impacted articles and making some tweaks to already strong content. It took time and effort to get the performance back up, but if we had created 100+ pages of AI content for this client, it’s likely they would have lost all their rankings and we would be back at square one.
This is why you need to be careful with how much AI you use with creating content, and why the biggest priority is that the quality is high and that you make the content as original as possible.
And if that’s your situation and you see your traffic, rankings and leads drop after an update, don’t panic: it’s fixable.
More content does not always equal more leads, sometimes it's going back to what you already have and making it sol
Mint Studios Recommended Reads
- Self-Promotional Listicles Aren't the Problem. Bad Content Is: Great article from Grow & Convert on this topic.
- The 2,000 AI Blogs Experiment: Does Google Penalize AI Content?: Interesting experiment run by Otterly
Thanks for reading,
Araminta & the Mint team 🎉
We help financial services and fintech companies turn content into measurable pipeline. Learn more about what we do.
