How to approach and monitor your SEO performance

This post is intended for an in-house eCommerce team with limited resources: a team of 2-5 who look after performance marketing.

We’ve worked with over 200 eCommerce brands. The common theme is prioritising paid channels over SEO.

To an extent, SEO is completely neglected, and the in-house team doesn’t know the contribution of brands vs. non-brand SEO.

Even when resources are limited, there’s no reason not to get the minimum set up for SEO so you can at least understand the channel and its contribution to your performance and, more importantly, isolate when something goes wrong. My god, when SEO goes wrong, you notice it’s contribution to the business, but at this point, it’s too late.

Let me tell you a story of how SEO can go bad. It was so bad that the brand shot itself in the foot, didn’t even notice and bled out for months, at which point the infection was too much for short-term recovery and the domain needed to be hospitalised for months (maybe went too far with that analogy).

When I first joined MADE.com, SEO had been in a slow and steady decline for a few months. This was partly the reason that they accelerated their hiring for my role. It was going to get pushed back to next year, but the slowdown in SEO prompted them to hire me as their first in-house SEO manager.

At this point, MADE had been investing aggressively in PR and brand building, so SEO had grown naturally. When that happens, no one cares about it as much; it just ticks over, and people at the top didn’t know you can actually scale SEO as a channel, just like the paid channels.

Long story short, after weeks of digging into issues and mapping the developer and website changes to limited performance data, I eventually pinpointed that it was the Sofa category that first got hit, which was causing a slow drag decline on the rest of the domain. 80% of this sofa category’s performance was driven by one URL (due to a lack of hierarchy and proper SEO strategy in place). I eventually isolated the issue to an A-B test the developers had run; they wanted to test the performance of your standard commercial listing page to a brand-led image-heavy landing page. As we know, A-B tests are fine for SEO when done correctly. This one was done in the wrong way. The A-B test was forced to redirect 50% of the traffic from the core sofa page to a new sofa URL. Eventually, Google had enough and received a page penalty, taking them from ranking 3rd for ‘Sofas’ to 40th+, a huge search term with over 100,000+ search volume and incredibly commercially led.

It’s the side of SEO that eCommerce brands neglect so much. While SEOs’ primary object is to grow the channel, a large amount of their value is to STOP issues going live that will impact performance. eCommerce brands, in particular, are dynamic, so there are so many changes happening each week on the website. There’s no doubt that some of these changes will detriment performance.

My job would have been so much easier if MADE had adopted some SEO monitoring. They could have identified the issue sooner, and I could have had additional historical data to use to get the answer sooner. Instead, I had to reverse engineer the impact using third-party tools and GSC/GA data.

The example above took over 8 months to recover when we eventually returned to rank 1. Oh, and did I mention that page alone was contributing £80k per month of free revenue on last-click attribution? Any business is going to feel that hit.

So, when MADE hired me, I thought about what I would have liked them to have in place so I could understand their SEO performance to date.

This is what inspired this post. The more eCommerce brands that have monitoring in place, the more they will understand their SEO performance, the relationship between the channel and other channels, and how to point out when something goes wrong sooner so there isn’t a mad panic to recover when we all know SEO takes time to recover.

Tertiary metrics are focused purely on outputs. These outputs can be growth-led, e.g., SEO-led developer tickets, blogs live, and backlinks built each month.

Tracking outputs ensures you can spot a slowdown in resources and investment in the channel, which will naturally cause a decline in performance.

To give an example, the brand team is investing heavily in online publications in Q1. Based on their metrics, they decide to pivot their budget to offline events instead of online publications. All of a sudden, in Q2, you are going to have a slowdown in backlinks to the domain. This is naturally going to impact your SEO growth over the next 2-3 quarters of the year. If you aren’t tracking your outputs, you would never know this.

The same can be said for development tickets and content. If your CMO or CEO asks why there’s been a drop in performance and you pinpoint it’s SEO, you’d like it 10x better if you could come back and say, “It’s because the PR strategy changed in March, and we’ve not had a dedicated SEO developer ticket for over 6 months,” which is the first step towards putting a business case forward to get more resources, either dedicated developer time or budget for an SEO agency to build links.

So, tracking these outputs:

• Tech tickets: set up a process with the dev team to get access to their historical changes to the website. They should have this already, but if you have access, you can monitor the changes to the front end of the website and the frequency and then cross-reference if there’s a performance hit. I’d also recommend joining these meetings or status updates from the developers, at a minimum, to build a relationship and avoid being that person who’s got access to their tracker and gives them extra work if something goes wrong.
• Content: Understand the content team’s resources and track the monthly SEO-led blogs (or any blogs) that go live. A simple URL in a Google Doc will be fine; you can then start looking up the performance of these URLs using GA or GSC data.
• Digital PR Backlinks: The easiest, cheapest way to identify backlinks quickly is still Google alerts. It is often overlooked how old it is now, but it is worth setting this up and tracking the backlinks when they go live, or the PR team will have their own reports they could share with you. If you have a budget, invest in aHrefs, one of the better tools for tracking backlinks at the end of the month. Google alerts are still better for more ‘real-time’ instant reporting.

You aim to get consistent output every month for all of the above. Tracking the outputs can be beneficial for cross-referencing when things go live versus when performance improvements or declines occur.

If you hire an SEO in-house or an SEO agency, you can track the above to get insights into general performance and outputs.

Tertiary metrics are focused purely on outputs. These outputs can be growth-led, e.g., SEO-led developer tickets, blogs live, and backlinks built each month.

Tracking outputs ensures you can spot a slowdown in resources and investment in the channel, which will naturally cause a decline in performance.

To give an example, the brand team is investing heavily in online publications in Q1. Based on their metrics, they decide to pivot their budget to offline events instead of online publications. All of a sudden, in Q2, you are going to have a slowdown in backlinks to the domain. This is naturally going to impact your SEO growth over the next 2-3 quarters of the year. If you aren’t tracking your outputs, you would never know this.

The same can be said for development tickets and content. If your CMO or CEO asks why there’s been a drop in performance and you pinpoint it’s SEO, you’d like it 10x better if you could come back and say, “It’s because the PR strategy changed in March, and we’ve not had a dedicated SEO developer ticket for over 6 months,” which is the first step towards putting a business case forward to get more resources, either dedicated developer time or budget for an SEO agency to build links.

So, tracking these outputs:

• Tech tickets: set up a process with the dev team to get access to their historical changes to the website. They should have this already, but if you have access, you can monitor the changes to the front end of the website and the frequency and then cross-reference if there’s a performance hit. I’d also recommend joining these meetings or status updates from the developers, at a minimum, to build a relationship and avoid being that person who’s got access to their tracker and gives them extra work if something goes wrong.
• Content: Understand the content team’s resources and track the monthly SEO-led blogs (or any blogs) that go live. A simple URL in a Google Doc will be fine; you can then start looking up the performance of these URLs using GA or GSC data.
• Digital PR Backlinks: The easiest, cheapest way to identify backlinks quickly is still Google alerts. It is often overlooked how old it is now, but it is worth setting this up and tracking the backlinks when they go live, or the PR team will have their own reports they could share with you. If you have a budget, invest in aHrefs, one of the better tools for tracking backlinks at the end of the month. Google alerts are still better for more ‘real-time’ instant reporting.

You aim to get consistent output every month for all of the above. Tracking the outputs can be beneficial for cross-referencing when things go live versus when performance improvements or declines occur.

If you hire an SEO in-house or an SEO agency, you can track the above to get insights into general performance and outputs.

A friend of mine who’s a CMO for an eCom brand. He had a highly experienced in-house SEO on his team and was paid over £80k. When performance declined, the SEO eventually went on a performance review and was removed from the company. We did an audit for them to identify the performance issues and found that there hadn’t been a change on the website for over 6 months. The SEO wasn’t actually doing anything, just joining meetings, and as an in-house SEO, it’s so easy to pull the wool over the eyes of broader marketers. However, by adopting a culture of tracking outputs, my friend would have noticed nothing was happening each month, hence the stagnation in performance.

The next step up in the metric ladder is focused on rankings and visibility, which the outputs aim to influence.

I recommend breaking up your keywords into buckets for eCommerce brands based on your URL structure OR categories.

You can do this as granular as you like/can do.

However, adopting a tagging system for keywords that contain ‘X’ keywords can help massively when pinpointing performance issues/wins across a huge website.

To give context, some of our fashion brands rank for over 100k keywords across hundreds of categories. To understand performance and gather insights, you need to break up websites of this size into mini-websites to make them more manageable.

2 core metrics here:

• Non-brand keywords: again, you can use aHrefs referenced above; they are well-priced. These keywords are only checked monthly. You can set up projects within ahrefs that track keywords daily. Due to eCommerce brands’ dynamic nature, I recommend setting up a dedicated ranking tool like SEO monitor AWR. SEO Monitor, in particular, has excellent bucketing or tagging methods and simple bucket-level performance. Each bucket can have a dedicated set of competitors to benchmark against and show where you’ve declined/increased for X category, and then X competitor is impacted. AWR is cheaper than SEO monitor; it depends on your budget.
• SEO visibility: If you use SEO Monitor, you will get a visibility score based on the keywords you input. To get full domain visibility, you can use Search Metrics (the OG but pricey) or Sistrix, which is cheaper. These are based on the huge amount of keywords in the database of these tools. Remember, with visibility to set expectations internally, it’s a great metric to understand trends and performance. However, it is a bit overly simplified, so you don’t want the internal boss’ to get too obsessed.
• New keyword ranking: These are great for tracking content specifically. If you are putting new blogs live each month, they are going to increase the quantity of keywords you are ranking for. So, it’s best to keep an eye on these new keywords each month. Again, Ahrefs can do this, or GSC gives some basic monthly reporting around “new pages with impressions for the first time.”

Finally, the most important metrics are the primary ones. These are non-brand revenue, traffic at a category level, and overall ROI.

To quickly understand your non-brand split, you can filter out brand terms in GSC to get a percentage split of brand non-brand. This is the simplest way to get the split; it’s not super robust, but it gives you insight and a metric to benchmark.

The dream split is close to 50/50, which shows you have a healthy brand to compete with non-brand terms. A split of 10:90 to a brand means your non-brand sucks, and you need SEO investment to grow it (unless you are a huge brand. In that case, this blog won’t be relevant for you). If it’s split the other way, 90:10 towards non-brand, you need someone protecting or monitoring your SEO; otherwise, there’s a high chance a mistake will have a huge impact on your bottom line.

You can also break up GA traffic by category and subcategory to understand the correlation between keyword rankings and traffic.

The main reason we split it out like this is the high seasonality eCommerce brands experience. Therefore, if you see a drop in traffic or revenue, you can use secondary metrics to determine if this is an actual performance issue or seasonality.

For example, if you rank number 1 for a term but traffic is declining, and you are still number 1, then it’s seasonality and demand, not performance. You can apply this logic across all of the keywords and categories.

Visibility is just a good way to aggregate the performance of multiple keywords in one overly simplified metric.

Remember rankings fluctuate. If you see a drop, you should wait at least 2-3 weeks to understand if it’s more permanent. The same goes for gains. Don’t celebrate too early in the world of SEO, but hopefully, after implementing what you’ve read in this blog, you will actually know when these increases/decreases are happening and not just react when it’s too late.