Don't believe the SEO Threadboi hype...Most of our websites are getting less traffic from Google, NOT more. The SEO bros showing their KPI (Key Performance Indicator) of traffic growth and everything is going up and to the right, might work for them, but most of us feel like "what am I missing, everyone is crushing it?"
Well I'm here to say the unsaid reality for many of us...my traffic from search is down 41% from its peak in early 2020, but the sky isn't falling.
Let's be real...useful marketing data to inform our KPIs are disappearing, we all have to look at ourselves and ask… have I set up my C-Level to judge the efficacy of my SEO Campaigns on a set of KPIs that no longer serve me in a world of disappearing SEO data, dark social, cookie deletion, new SERP layouts and AI "search" in 2024?
All is not lost! People are still buying stuff, they are just harder to track than ever.
The great marketing data disappearing act
What started with "not provided" and is now Performance Max (make sure you know the downsides of PMAX) has ballooned across all platforms. We are getting less marketing data than ever, but our KPIs don't reflect that reality.
Then stop and think about the zero click marketing trend where we’re basically penalized for driving traffic to our websites by social platforms (I was able to 4x my social traffic, but recognize I'm an outlier). Notice the rise of “link in comments” on LinkedIn?
In a zero click marketing world, you need to not only look at your web analytics metrics (did they convert), the SEO platform KPIs (rankings, links, traffic, etc) but then also all the in platform data (Impressions, Engagement, etc)...all of which have different definitions on different platforms - try to make sense of all that with all the data, now imagine without.
How does this study from Gartner claiming a 25% drop in traffic from search engines change how you report?
Do you just say "Gartner are idiots, they always get things wrong!" or do you seek to validate this with your own data? Hint: Your CEO doesn't think Gartner are idiots.
That is what I intend to do in this post (this focus on KPIs for SEO in a ChatGPT world started the end of last year in this post) help us move from the traditional KPIs that have served us well to the new KPIs in 2024 and beyond that may help you have real consulting conversations about SEO budgets and SEO investments in a world where all we got is a few generic "studies".
So far in 2024 AI Search is a ZERO data environment
How can you develop KPIs for search when there is no data on the new technology displacing some people's search behaviors, Gen AI?
Sure you could make traffic from AI engines a KPI for "SEO" sure, but have you looked at how little traffic they send? The C-level isn't going to want to hear about your 10 visitors from Gemini.
Here is my Perplexity traffic trendline, last month Google sent me 30,000 visitors when Perplexity and ChatGPT combined sent me 200.
AI platforms, where some of your customers might be getting answers to problems give you ZERO data too, Copilot, Gemini, Perplexity, ChatGPT, Claude, Meta AI = YOU GET NONE! I get that the traffic might be low, but we don't even get the questions people typed in...reminds me of not provided in SEO all over again and how we had to change KPIs. Now its that time again.
Well guess what that leaves us with? A rats nest of diminishing data for a group of marketers supposed to be data driven. Goals still need to be hit.
However, I believe that when data from one silo gets lost, we should look for other data sources and silos to help offset how much we have to rely on gut (or in platform reporting).
In 2024 & 2025 you are going to need to convince the C-level that the SEO KPIs you've trained them to judge you on, are diminishing in value.
The Past & Present State of SEO KPIs (1999-2022)
Can it be that it was all so simple then, I recommend coming up with new KPIs, as there's a lot of problems with these. The SEO KPIs that worked in a pre-AI and data rich world, need to be augmented with some new SEO KPIs I'll talk about later.
SEO KPI: Indexed pages
If I get more pages in the index, I’ll get more visibility and with more visibility, I get traffic -you could get more pages in the index, which is a good thing but again how many more pages would you need to get in the index to offset the losses you are likely experiencing in your declining SEO traffic?
FYI I think organic social is the way to offset traffic losses on SEO and I got the data to prove it, but that is for another post.
SEO KPI: Traffic
More pages = more visibility for your website = more visibility = more clicks, more clicks = more leads and sales!
You would look in Google Search Console or Google Analytics pull a report that showed traffic to your landing pages and boom, if the chart is going up and to the right you were doing your job.
Rankings as a KPI
You used to could rely on rankings as the main KPI, because you could assume a certain click through rate on those rankings, but then Google SERP layouts changed. Did our expected CTRs change with them? If not we've set up our c-level to assume certain CTRs on a ranking and that just isn't a quality metric anymore.
I do not know 1 SEO who changes CTR expectations on a keyword level, and tracks 10,000+ keywords, it is quite hard, and lets be honest there are so many factors that impact the layout that it makes so sense to try.
Just consider a few changes to the layout that could result in you hitting a Ranking KPI and miss a Revenue or ROI KPI:
- 4 ads vs 3
- Mobile vs desktop
- Answer boxes
- Rich snippets
- Video results in the SERP
- All of these layout changes significantly impact the expected CTR.
If you want to use something to benchmark your CTRs, I think this is the best tool out there right now, but even this tool isn't keeping up.
Links as a KPI
Links…don’t even get me started. I've always said, the distance between your KPIs and what your boss shows their boss, determines how much visibility you get in an organization...I have yet to hear a CEO ever talk about links in a quarterly earnings call.
New SEO KPI's should be offset with business metrics every time.
SEO Traffic vs Revenue
Trend out your SEO traffic against your revenue, pre and post ChatGPT launching. As you can see below our revenue is flat over the last 2 years but our sessions from search are trending way down, 41% to be exact.
Why the old KPI fails:
In this new world traffic going up doesn’t mean revenue is going up, and yes while it is quite difficult to control revenue for many B2B’s it’s what your CEO looks at so the closer you can get the better.
What you need for the new KPI:
Access to revenue, usually in your salesforce or something.
Why this KPI is better:
Context, context, context. If my SEO traffic is dropping but it isn’t impacting the business, then maybe I was overweighting how these two things worked together.
How to use this new KPI in the C-suite:
This allows you to avoid the "traffic is down, the world is on fire" statement from your boss (assuming your search traffic is down and revenue is flat). This is an opportunity to be a digital consultant, not just an SEO consultant here to preserve SEO budgets / power ... Go hunt down what's growing and be open to helping your company move budget there (it just might be AI, but it may not be). Either way you get to tell everyone to calm down, the traditional KPI of SEO is down, isn't impacting the overall business KPI of revenue, so you got time.
If SEO traffic is down and revenue and leads are down, you might be in for a different approach.
Here I am using my own data to show you that yes, it looks like my main KPI of SEO traffic is down, but our sales don’t seem to be impacted. My hypothesis is that people are still getting to our content somehow or finding our brand somehow to sustain sales / revenue.
I’ll show you later how I back into if this is likely AI or not.
It doesn’t matter what division you are in, SEO, social, SEM, email, if your job is to help the business, you need to start to overlay your typical metrics against the company’s revenue (CEO’s get hired and fired on revenue, not rankings so report things in their language).
Track the divergence or congruence between your KPIs and revenue, leads, or some other higher order metric.
Leads the same, traffic down = Your customers might be going to AI search more!
Let's move up the funnel, revenue might be too far removed…so I looked at leads, flat over 2 years, so we’re not seeing a drop in leads or revenue in spite of a big drop in SEO traffic.
SEO traffic vs leads / lead quality
Lets use my data for my own company, Seer Interactive:
If you compare my leads Jan - June 2022 vs Jan - June 2024, you’ll broadly see that we have the same lead numbers, again our SEO traffic is down 40% yet our leads are flat. What does that mean from a KPI perspective? That I should rely a lot less on the math that said...if I can get rankings to x, and traffic to y, I should get Z amount of leads.
This might not be for everyone, but I feel like building a brand makes your more resilient to SEO swings, as your brand will still show up in questions people have whether they are on Google or Gemini.
Traffic from SEO can decrease, but my lead flow doesn’t.
Now maybe SEO is a leading indicator and I’ll have a massive hit in leads later, but we’re in October and this Jan-Jun 2024 seems far enough away that we would have seen the impact.
What you need for the new KPI:
Access to leads, usually in your HubSpot, Salesforce, Pipedrive, wherever your CEO goes to understand how sales are, you need to be there. If you are ecommerce you don’t get “leads” so for you check some of the KPIs below.
How to use this new KPI in the C-suite:
It looks like yes my main KPI of SEO traffic is down, but our leads from the website don’t seem to be impacted. I’ll look for offsetting metrics to see where people might be coming from, so we can adjust our time and resource investments, but overall in spite of our traffic being down 40% we are not seeing that impact the core leads. Let me set up a new option on our “how did you hear about us” that includes ChatGPT/AI.
SEO Brand Metrics vs revenue
What would I look at next to build out a more future looking / forward looking set of SEO KPIs to offset the old school traditional ones?
For me it was brand (direct) & homepage traffic (from search) against SEO traffic. I noticed when analyzing both direct and homepage traffic as a KPI that they much more closely aligned with my leads and revenue than SEO traffic.
The idea is simple, if people do a ton of research in AI tools, then they are often left “Googlling” your brand (“Seer Interactive”) or a Brand + product name (“Peak Design Tripod”). You need to determine whether in GSC (where you get keywords and page) or GA/Adobe (where you get pages only) how you want to deal with the fact that some pages might rank for brand and non brand.
Take a look at my site (B2B), neither of these more "brand leaning" SEO metrics, are trending down, they are...flat. Just like leads and revenue. Not down 41%.
What you need for the new KPI:
Most of us have branded search trends, or homepage traffic trends in Google Analytics, Adobe, or whatever you use, you just should make sure you tighten them up, look at them mobile vs desktop as well, so when AI features roll out on device you can check what happens on the day of an update from Apple or Android and how it impacts your traffic.
How to use this new KPI in the C-suite:
Show don’t tell on this one!
Walk your leaders through the layouts of results imploring them to allow you to also track KPIs that reflect the future state of SEO/GEO/AIO or whatever optimization we're calling it.
Use several Gen AI "engines" - Google AI Overviews to Gemini, Perplexity, and ChatGPT.
The layout impacts the traffic, remind the c-suite that maybe when people do research online, they eventually narrow it down to a company and google it, try to use their own behavior in ChatGPT to remind them of how they search, make it real.
Yes our SEO traffic is down 41%, do you use ChatGPT to narrow down choices?
C-suite: Yes I do.
Then once you narrow it down what do you do?
C-suite: I Google the company, model number, product name etc and buy it.
Great, so back in the day you might have hit their site during your “research phase”, now you might hit their site once you have made up your mind, lets look at a few examples…
Get them on your side:
Play the "find the link" game with your execs
Let’s say you ask ChatGPT a question like “Give me a B-corp certified SEO company”, cause you are seeking a company aligned with our values… notice how how every layout shows you a link, that is easy to click on?
We believe that people will do, what you did, Google the company. Oh and here is a research study from Bing that shows their research on when AI search was a better user experience and when traditional search was a better experience:
Here is Gemini
ChatGPT 4o
Gemini 1.5 Pro with Search
Perplexity
Caveats to using Direct & Branded traffic as a new SEO KPI.
Job searchers: When your company posts a job, people will be googling your company more…you may have to find a way to control for this traffic, like if the person Google’s us and goes to any pages related to careers or culture remove them from this report.
Your own people: People Google their own companies, or come in direct all the time. Be careful for that too.
-png.png?width=1920&height=1080&name=Large%20Quote%201%20(1)-png.png)
Think bigger, to get people to search for your brand means an SEO has to be thinking, how do I create brand value, when I mostly get credit for non brand traffic improvements?
At Seer we've been at this intersection of PR/Brand and SEO for over 10 years check the video below as proof, we're ready for a world where SEO needs to collaborate with brand and product to show up in Gen AI "engines".
SEOs looking to get those "AI rankings" better learn to partner with your product people, brand people, advertisers, PR agencies, etc. How can you pivot your SEO skills to assist those people get more from that they are doing, so you can show up in more branded search?
New KPI: My site visitors on ChatGPT vs global average?
The tool to find this out is sparktoro. Make sure you understand some of how the data is aggregated so you don’t make assumptions that you shouldn’t make, like I keep doing. Here are two resources to help you understand…how many profiles do they track & a bit on accuracy and bias in their data.
In this report I’m analyzing stack overflow - who denied early on that ChatGPT was impacting their traffic, telling all of us that the analyses were overblown, read for yourself.
If you wanted to validate this statement, you could just go to sparktoro, put in stackoverflow.com and see if the people who visit stackoverflow are likely to also visit ChatGPT more or less than the global average.
Here is the data for Seer Interactive’s site, openai.com is 6th for people who visit our site vs the global rank of 20 in Sparktoro. Meaning my audience is more likely than the average to be on ChatGPT. Therefore I can adjust my thinking accordingly about how important it is for me to invest time and effort understanding how I rank in ChatGPT and other GenAI tools.
How to use this new KPI in the C-suite:
Executives are very comfortable with incomplete data and having to make assumptions, that is why they are in the seat that they are in. What isn’t OK is showing up and not giving your best guess, so when you get asked “are our customers on ChatGPT” I think you can say… Our customers are more likely to be on ChatGPT vs the general population according to Sparktoro.
I just took an “old school” company and compared them to a more digitally native company, and sure enough Sparktoro showed that people who visit the more digitally native company’s site were more likely to use ChatGPT (as they are likely more digitally savvy) than the old school company. Hmmmmm. (Maybe I am looking at this wrong and Rand Fishkin will come in and lay the appropriate smacks down).
KPI - Wikipedia Traffic vs Your SEO traffic
This one is a bit "out there".
Build your own Wikipedia traffic tracker for high ranking Wikipedia pages. It is getting easier than ever for non coders to build stuff, check out Replit's AI Coder / helper.
This is the 201 version for you innovators and status quo crushers out there, it could also be rife with bad ideas and unthought of issues, but we learn by putting our shit out in the world and letting people critique right?
%20(2)-png.png?width=500&height=200&name=Blue%20Modern%20Social%20Media%20Strategy%20Presentation%20(Mobile%20Banner)%20(2)-png.png)
If most of AI overviews on Google are showing up for informational searches according to this study, then I would use my ranking tool to find Wikipedia pages that rank highly and see if my drops on those keywords where I’m visible and Wikipedia are visible are dropping at the same rate. Better data.
For me, I used AI to help me code up a prototype of traffic to the Wikipedia page for stackoverflow. An SEO who is unaware of the fact that Wikipedia's traffic is publicly available or doesn’t lean into “making things” with AI would have to rely more on gut or best practice.
In a world where less and less reliable data is available to me, I’m going to scrap and claw to get some data that helps me intelligently guide my ideas and consulting efforts for my clients.
You could compare the Wikipedia page traffic for high ranking wiki pages against your own GSC data (or analytics data at a page level) to help refine your gut…if you rank below Wikipedia are you seeing similar drops in traffic in the AI world?
Or are they going in different directions, which would further inform your gut instincts - welcome to SEO KPIs in 2024. You might have to look for new data, internally, externally, etc to show your true value, and understand real risks to your existing KPIs so you can guide your company accordingly.
Ummmm, stack overflow I graphed out your Wikipedia page traffic and when you were saying “ChatGPT is overblown and isn’t hitting our traffic by 50%, it is more like 5”, I let Claude + Colab code me up a solution that shows…nah son, you lyin’. Now imagine if their SEO team didn't come out with that head in the sand response and instead built something to show their execs that actually Wikipedia is dropping at the same pace (or more aggressively).
How to use this new KPI in the C-suite:
If you are making predictions with incomplete data (as you will have to in 2024, 2025 and beyond), your ability to walk through your methodology and the data you used will be helpful. Imagine seeing a 50% drop in Wikipedia traffic for a keyword or theme you also rank for. You might adjust your predictions a bit better with this data than without it, or using Gartner's prediction.
My SEO traffic is dropping, does that mean that SEO is less valuable?
I'm early and playing with this concept. So I'm not prepared to lay out all of my ideas here. But I do believe a lot of people will see numbers like mine (41% drop) and under invest in content, it seems like a logical conclusion.
But I think there's a content type out there that creates a lot more value for your brand (this is why tracking brand search and homepage traffic is so important).
It is that opinion driven, atypical content that an LLM would never give you, like when I de-optimized my high ranking page.
-png.png?width=1280&height=720&name=Old%20SEO%20KPIs%20(4)-png.png)
Maybe in the old SEO world you would produce 50 pages in 200 hours worth of work. I think in the new world you might deploy that same 200 hours and only come out with 10 pages.
But those 10 could have insane amounts of lift for the KPIs that matter, and also positioning your brand in a way that makes you more visible because you've answered a question in the unique way.
I'm realizing that posts that take off with humans on social, can drive a lot of traffic that 1) offsets the sting of SEO traffic and 2) Drives newsletter signups with a LOT less traffic.
I've seen a post with 36,000 visitors from search get 2 signups and a post with 400 visitors from social get 18 signups.
A generic answer that comes from a Wikipedia style post on "what is a CRM" isn't better than an LLM in most instances, but the more opinion driven, atyipcal idea driven content, gets humans clicking.
Major shout out to John Henry (an alumni who founded this agency), who reminded me that this was possible you can look at the types of content that are able to pull visitors from LLMs, you can learn what types of content, and questions, lead to people still visiting your site (thus getting traffic) - this might be a clue for you on where to invest your time.
KPI - Newsletter signups vs Search traffic
I got one more KPI for you to consider...All traffic isn’t good traffic, my search traffic is down 41% right? That is a typical search KPI, but my newsletter signups from the website is up 65%. This is another example of a "am I losing the traffic that wasn't that helpful anyway?" type KPI.
For me, traffic is great, but traffic that is showing people who might be interested in hearing more from me is better.
How to use this new KPI in the C-suite:
Use a metric that isn't traffic to qualify if that data loss is as big a deal as your leaders might think.
You just might be losing traffic on pages that really never had a business impact anyway, wouldn't it be great to be able to defend your value to an executive by saying yeah the traffic went down to a bunch of informational pages that weren't really driving outcomes for us anyway (I do wonder how that will impact SEM, when they could build audiences based on site visitors).
Tell your leaders, while I don't have all the data I wish I had, I have figured out which pages retain the highest value and I've doubled down on pages like those.
What do do if your KPIs are overweighted to SEO in a 2020 world?
Don't fret! You're an early mover if you've read this entire post.
Be honest, traffic is likely to drop, no matter what you do. People are not searching less.
The first thing I would do is show that Gartner study to my leaders, showing them that I'm not afraid to say this might be the new reality for us.
But have an heir of "but we're going to find a way to help you make decisions."
Tell them that you're not okay with generic studies and that you intend to not have this company rely on those generic studies to determine your future impact. You going to have to tell them that the KPIs they've been judging you on will not work in the future, and over time they will atrophy in their ability to show them how you're performing.
Build trust, show the Gartner trend vs yours, are they right?
Now that you've shown them the report, from a source that many of them trust, show them the actual data. In my business and analytics leader would come to me and say 25% is an underestimation of the impact, let me show you how baby impact is, and what I intend to do to backfill those goals until we can have a larger conversation around new KPIs for our business.
Get in front of this issue, go cross divisional, help offset the losses in visibility anywhere you can.
Now you should be open for a larger conversation around how to backfill those numbers in the short term, cuz that's sea level executive has numbers they got to hit and you need to be a part of helping that to happen.
The earlier you can say to them, hey I'm only going to be able to contribute 70% of the goal if we stay at this pace, the better off you are. But then you've got to also come with five ideas on how to backfill that hole.
If you've been doing any research on how to "rank" in places like AI overviews, ChatGPT, or any other Generative AI "search" then this is a great time to show them your research, which queries are most disruptible, and which ones are most resilient, find out what is most common in the ones that have high resiliency and pivot your strategy to put a disproportionate amount of effort on those.
Start reporting on the SEO KPI that replaces ranking: Track your brand and which questions you show up for.
Our favorite thing to do? Take 1,000 or 10,000 questions (pull from PAAs, chats, customer calls, anything), and use AI to put the questions into themes and then cross reference which brands show up for which themes, here is a DIY version.
We were doing some research in the automotive space, and we saw for one brand that affordability was where their brand showed weakness in AI answers across ChatGPT and Gemini, when presenting that to the team, we need to be ready with ideas on how we can use this data to support PR, Marketing, Content, and other areas to get us showing up more, if that is important to the client.
The days of writing a few pages for visibility are going to be over, its going to need to be a team effort, where the LLMs are trained on seeing your brand as "affordable" or whatever brand attribute you care about.
Sign up to have us check questions for your brand and see the gaps between the attributes you want, and the attributes you have.
