Hey Mozcon Peeps, yesterday I was skipping slides like crazy, it lead to a disjointed presentation, sorry about that.
To fix that I am getting part I of laying out what I was thinking that lead me to some of the things I presented, it gets you into my mind a bit on the dots I connected to create that presentation.
Sorry for rushing through my slides, I owe yall a much better laid out view into what I am seeing that lead to that presentation…so here is my attempt to do that. I will post this t our site later today, but I feel so bad that my flow was off, tha I’ll let ya’ll watch me as I put this together, and let you all comment to me so I can respond. Sorry about the delivery.
I opened showing my traffic from search down 41% in 18 months, but my revenue is flat / slightly up. This was the catalyst for this presentation, how can I lose that much organic traffic and visibility and still grow a little bit. I also wanted to show this because this is the reality for most of us in the room, yet most of us as speakers don’t acknowledge this. I didn't get hit by any penalties, lets get real…people are getting to our sites less….
Wil, I want to do the stuff you showed, but my KPIs, boss, etc.
I was lucky AF to run into Katie Kirchner, whom I thoroughly enjoyed working on SEO with after my session. I gave her my cliffs notes on how to get started on this path, and realized that I cut that out of my side deck. So, here is how you try to get started.
New school vs old school SEO
1st rule of getting your boss / org / client to level up your SEO to lean into anything that is not what they deem as “traditional seo” - Don’t make your desire to try something new and untested their problem. Don’t ask them for time to work on “community style” SEO.
The “earning your trust” approach
This question is a major unlock for me when I run into friction with my “new school” SEO approach…What do I have to do, or metrics I have to hit to help us hit our current goals, once I hit them will I have earned your trust to invest 5%, 10%, etc of my time into this “new” approach for the next 3 months to see if there are legs there. If not, we tested something new that could have helped us break out, and we didn’t sacrifice our goal attainment in the process”
The key is “what do I have to do to earn your trust” - you are putting the responsibility on YOU to do the work to earn their trust and hit their metrics first. (This works great for raises too)...What metrics do I need to hit, to make a 15% raise a NO brainer for you to provide me crushes the other folks who say “I deserve X, let me show you why”.
The “forgiveness” approach
Find an old post that you feel was over “SEO’ed” as a test. I recommend finding one that ranks still but for an old topic, or old product - remove 80% of the SEO content and push the keyword laden parts in 1 maybe 2 paragraphs to the bottom of the post, and instead start taking an SME and asking them questions to update it (more on that below).
What are we doing for “SEO purposes” that work against community building?
This was the opening example of TNT Amusements, where the same people who have made over 2000 pinball videos on YouTube were asked by their SEO to “talk about how fun pinball games are” for “seo purposes” Here is the page where the instructions were left.
What is the takeaway?
Where are we as SEO’s following a checklist of things to do for SEO and ignoring people in our own companies that are super passionate about a topic / what we do?
Checklist / “Me too” / Semantic SEO is leading us all to creating unremarkable content, just slightly better than the next one. Dare I say studying what the other people who rank above you are doing is step one in creating unremarkable content. This is the sea of sameness issue I discuss here where we are all following the same playbook.
Ryan law was 5 years ahead of me on this :) Check out his post on copycatting in content development.
The Ahrefs team is also putting out an interesting POV on “content scores” for SEO tools & how they are falling short. The post look at how a higher content score doesn't lead to higher rankings, yet so many tools are giving us scores, leading us to more unremarkable content, which really doesn't help with rankings all that much, but definitely keeps us from having spicy points of view in the marketplace.
SEO Reward systems are the problem
How do you get your bonus? Is it rankings? Traffic? Revenue? I bet it’s not brand building. Great content that is community driven vs commodity driven. That is why you need to advocate for change in your KPIs or otherwise you’ll be rewarded for building commodity content vs community content.
My mozcon presentations will at most get 1,000 people in the room, I have blog posts that were written once that will get that in a week. Imagine a blog post that took 10 hours that can get you 40,000 visits a year up against a presentation that can get you 1,000? What would you be rewarded for?
How can my revenue not drop when my traffic does?
Well, I believe it is you, the person reading this post, you are in our community, whether you agree with us or not, you likely recognize Seer as a company trying to do good, help our industry, etc etc. That community building, builds resilience, community will lift your business up when your traffic is down. That is how I believe our organic traffic can drop 41% but our business is fine. Guess what was growing? Social traffic, you know the place where humans share with other humans, there are no algorithms to trick humans into linking or sharing, you have to create quality stuff.
The SEO conundrum, today vs tomorrow.
The hardest thing about SEO is the balance, don’t do what works today and you are snake oil. Do too much of it, and you are penalized. This is an area that I am super proud to have navigated Seer through over the 22 years we’ve existed. I feel like I can see a tactic that works and determine its “abusability” innately.
Every time there was a major algorithm update, we didn’t get worried because we play both sides of the fence for our clients, we’re hedged. It’s the “mutual fund” approach I talked about. You have to re-balance your portfolio on a regular basis.
Low quality links, you better believe we got them back in the day, but as they were working I would double down on what is a more human / high value approach, so when updates happen we don’t get hit. Algorithm updates have been good for our business over the years because we’ve kind of branded ourselves as having a foot in both.
That is why I am not saying, everything community driven, what I am saying is “invest a small part of your portfolio in working on human signals”. You should do that for 2 reasons. 1 - lets just say gartner is right and 25-50% of traffic is gone or this new report is right. Well flexing your community muscles, lets you position yourself as a bridge that knows how to optimize for SEO and humans, so no matter which way they go you have an idea on how to optimize both.
The future: Internal Influencers will build your content (for SEO) & for community
Long term I see a world where, there are internal influencers at companies.
Think about it, I am an influencer (yuck) at Seer. I should want to grow as many other influencers as possible, and now that we have more marketing muscle coming at Seer, I want to try to find ways to elevate the profiles of a few people.
This will get me links and content that ranks, the key is you need people with spicy opinions that want to actually “be better” at this social influence thing.
Companies are going to fuck this up because of the “what if they leave” after we invested all this time and money? I’ve been trying to find ways to support my alumni after they leave for over 10 years, heck we did a video interviewing people who gave us notice that they were quitting.
You company LinkedIn / social page is where good content goes to die.
4 months ago when I took over our marketing (I mean I was doing all the canva thumbnails and everything, every day), I learned a ton. Lets get after it.
I was shocked that Seer and I both have the same amount of followers on LinkedIn, yet my personal account was crushing on every metric. I spent weeks trying to improve the company account, I was holding back “bangers” from my personal account just to try to improve the engagement on our company account.
After a few weeks I realized that for me company accounts only work for culture and jobs (I was able to get some traction on Twitter but not much).
Here is an example of me testing this:
Here on threads where I have very few followers, I’ll post something and watch HUMANS take to it, this was one of my best performing posts on threads. I like laddering my socials, by testing out on places were I have a smaller audience then if I see great engagement, I know humans liked it, so I ladder it up and up. I took that post from threads (800 followers) and posted to LinkedIn on the company account (24,000 followers) - it got no visibility until I reposted it.
I’ve tested this 20x, and now have some to the conclusion that company accounts are the wrong place to get visibility. This is why companies should invest time and money into building more internal influencers.
Consider using tools, I’m using sales navigator, to try to stay on top of people in your company who are posting things, and see what they are posting that is breaking into the upper bounds of their typical performance.
Generative AI is going to have the same problems as search
What I LOVE about generative AI is that it removes all the friction that websites put in the way. Cookie consent, your newsletter popups, bad answers, etc etc, it’s all a bit much. But they are super prone to sea of sameness, with their predictions and what not, especially when you are doing one shot queries.
Are people going to get answers from these places…yup, is it going to ding my traffic…yup. But if I build content that engages community, the community will be sharing it.
I built 3 blog posts that got 0 top 10 rankings…it was the best thing I’ve done.
If you were in my presentation, you saw me show this…what you also saw was that humans loved those posts (relative to my other posts) even though Google didn’t. Now the kicker for my new marketing team is to take that content we now know people like (because I can see the social spikes), and see what “keywords” could it rank for if we spent any time optimizing. I almost wonder…should we be optimizing content AFTER the humans have “voted”, because now you know its sending Google signals that humans love it, so now maybe we optimize it or boost it. FYI I boosted my best post and it failed miserable, so I got more to work on on the paid boosting side.
How do I know if I am winning for humans?
Take a look at your UTMs, did any of your content get mentioned in industry pubs, also hs_email is the default UTM for hubspot meaning your content passed a human check, and was quality enough that someone decided to pass it on to their trusted readership. That is a great signal.
There are also ways to find your slack groups where you are getting traffic / mentions, you can see telegram, whatsapp (log files), and other places where humans share with other humans.
Retrieval augmented generation, if this is the future, then AI search will need your skills
This is an area I think every forward looking SEO should be keeping an eye on. If OpenAI & Gemini start triggering web searches for queries, this is going to be a big thing. The depth at which AI can scan things well past the top 100. Imagine when you add just 1 or two adjectives to a search query that generative AI can scan the top 100 docs and can augment the answer by retrieving 100’s of web documents.
I wonder in this world if you basically attempt to get as many ranking keywords in the top 50 or so. Once you look at your AI search traffic by page, maybe you take your keywords ranking 50+ and try to optimize those up a bit? I have no idea but I am thinking of what I would do next if in an AI search world, top 10 is less important than top 20, 50, or 100? I would start really looking for my paid queries to see words around my core keyword, and then not only try to rank top 10, but also find some tangential words. Imagine a world where for each blog post you put in:
Various roles (1 year in the industry, 5, 10, etc)
Titles (VP, CMO, Analyst, etc)
Industries (ecommerce, SaaS, etc)
I believe in this future state people will start to identify themselves as “I’m a CMO of XYZ brand, and I’m working on my SEO campaigns, how might ABC tactic impact me?” I guess trying to build content that takes into account role, title, industry could be helpful to you expanding your reach.
Are your customers using AI search?
Gemini, Perplexity, Etc aren't giving you any data are they? I don't think blank stares work anymore, we have to have a point of view and an approach. In this part, I started thinking...if people start using chat search, I think the way they search Google will start to change, and the change I hypothesize will happen is people will start more queries with the word "I" - "I am a _______" "I would like to get help with _______"
So take your paid data and trend it out over time, we've seen a client getting over 100% growth in the last 9 months, maybe their customers are starting to use chat search more.
I dunno if you ca do this, but what would happen if you took your content and built a custom GPT for people to interact with, and take that page and add a big link to your custom GPT, and see if people prefer that experience over your longer blog post, like we did here for our screaming frog guide.
Are the answers on AI search about my brand right?
Searching multiple LLMs sucks, opening all of those tabs, just to get 5 different answers. Do them all in one tab with Poe.com - in this video you can see how I am using their multibot feature as a "dogpile for LLMs". Then you can use the LLM to compare all the answers to each other to see where they are hallucinating, and then because it is a persistent chat just keep running it. Here is the video.
You can also use ChatGPT for Sheets to take PAA's and see what answers ChatGPT is going to give you right in sheets. Recently I added entities in GPT for sheets, which then with an apps script can help me see what variation I'm seeing in ChatGPT's answers to my query "Tell me about Seer Interactive".
How can I better interview SME's?
I mentioned a series of tools. Vapi.ai (Follow Christian Here) was the one that uses AI to ask someone questions, letting you interview SMEs at scale, I really like this tool because in the system prompt you can give it directions to help draw out of an SME their thoughts, you can make the prompt get the user to think in terms of a CMO and a practitioner, getting you that content that ca be broader for multiple roles.
You are an editor who works for an agency, your job is to take the text below and ask the caller how have things changed recently that would make this old most more up to date. When they give you questions, drill in deeper. They are typically practitioners, so you should try to encourage them to think through how these updates would be described to a CMO or a marketing leader as well as a practitioner. You should ask them to think through their ideas in steps, trying to push any long winded answers into how would you tackle this in 5 steps, you should also guide them to make predictions and "hot takes" about how this might change in the future. Remember to read the entire post below to gain all the knowledge you can about how we approached this blog post the first time, then prompt me with specific question that helps me to update the post and make it relevant today.
Ask one question at a time, and specifically read back to me a site section from below, then ask me my opinion.
Pi.ai might eventually go away but for now, use it, it is an AI that will pull your thoughts out of you, vs just using AI to write your content. To build community you need to have ideas and opinions, not just facts in your posts, eventually facts posts will likely be gobbled up by AI overviews.
C4 data set - who trained the early training models?
I am hoping someday we can get more visibility into what sites train the models. For instance Moz.com was the 400th highest ranked site, search engine land was like 3000th, so that means moz has more influence in the model. I Hope some day - that we get more updates, here is the post where you can see if your site was used to train the early models.
