All episodes
VISIBLEwith Usama KhanSeason 1 · Episode 1

The AI impact on SEO is real, but your traffic graph is lying to you

The AI impact on SEO is real, but traffic drops tell the wrong story. What actually matters for B2B SaaS and how to adapt your strategy.

Listen — Apple Podcasts

Your organic traffic is probably down. If you run content for a B2B SaaS company, you already know this. AI overviews eat clicks. LLMs answer questions directly. And the execs on your team are starting to ask uncomfortable questions.

But traffic was never the right metric to begin with.

I sat down with Junaid Baig from Breaking B2B on the first episode of my podcast VISIBLE. Junaid manages SEO and content for B2B SaaS clients across the globe. And the pattern he keeps seeing across client accounts tells a different story than the one your traffic graph shows.

Traffic is falling. Conversions are not.

Junaid shared something that matched what I've been seeing with my own clients. Multiple B2B SaaS accounts saw consistent traffic drops over the past year. But their conversions went up. One of the law firms he previously worked with had their best conversion year since 2020, all while traffic declined.

This makes sense when you think about what kind of traffic AI is taking away. If someone searches "what is programmatic advertising," an LLM answers that instantly. No click needed. That visitor was never going to book a demo anyway.

The traffic you are losing is the traffic that never converted. The visitors who still click through from search are the ones comparing vendors, evaluating features, and looking for a solution. Those are the visitors that matter.

So the AI impact on SEO looks scary on a dashboard. But if your North Star metric is demos booked or MQLs generated, the picture might be very different.

Top of funnel took the hit. Bottom of funnel got stronger.

When I asked Junaid which content types got hit hardest, his answer was immediate: informational content. Definitions, "what is" posts, benefits lists, use case roundups. The typical top of funnel stuff that most SaaS blogs are full of.

This tracks with how LLMs handle queries. Someone asking "what is a CRM" gets a direct answer in ChatGPT. No citation, no recommendation. But someone asking "best CRM for a 20-person sales team that integrates with HubSpot" gets brand recommendations and source citations. That is a bottom of funnel query. And it still drives clicks.

That is why Junaid starts every new client engagement with commercial queries. What are ICPs searching when they are close to buying? What pain points push them to switch vendors? He only moves up the funnel after capturing existing demand first.

Customer research before keyword research

The most practical takeaway from our conversation had nothing to do with AI visibility tools or prompt tracking. It was about where keyword research should start.

Junaid's first request when he gets a new client: give me your customer data. Sales call transcripts, demo recordings, support tickets. Not search volume reports from Ahrefs or SEMrush.

AI has made this process faster. You can dump a batch of sales transcripts into Claude or Notebook LM and extract pain point themes, objections, competitor mentions, and the exact language buyers use. What used to take hours of manual listening now takes minutes.

This matters because the keywords that come out of customer research are the ones that match real buying behavior. They tend to be longer, more specific, and more commercial. They are also the queries that LLMs are more likely to cite sources for, because they require nuanced, product-specific answers that AI cannot fully generate on its own.

AI visibility tracking is still messy

I asked Junaid how his team tracks AI visibility. His honest answer: mostly Google Search Console.

The reason is simple. AI responses are dynamic. The same prompt gives different results based on who is asking, where they are, and when they ask. And the way real buyers prompt LLMs is nothing like the clean, short queries that most AI visibility tools track.

A real B2B buyer does not type "best project management tool." They type something like: "We are a 20-person team in logistics, currently using Monday.com, but it lacks resource allocation features. What should we switch to?" That level of detail produces a custom response every time. No tracking tool captures that.

So for now, the most reliable proxy is still GSC data filtered for AI referral traffic, combined with directional signals from AI visibility tools. Not perfect. But better than pretending prompt tracking gives you the full picture.

What would you find if you stopped watching the traffic graph?

If your traffic is down but your pipeline is flat or growing, the AI impact on SEO might be doing you a favor. It is filtering out the visitors who were never going to buy.

The question worth asking: are you still building content around the queries your buyers actually type when they are ready to evaluate, or are you chasing the top of funnel traffic that AI already owns?

If you want help building a BOFU-first content strategy that drives pipeline across Google and LLMs, book a strategy call and let's figure out what that looks like for your SaaS.

AI SEO

Want to discuss AI SEO for your B2B brand? Book a strategy call →