Week 4 of AEO: How It's Actually Going

We’re back for an AEO check-in. (New here? Catch up on the series here.) Last time we shared our action plan: complete HubSpot’s AEO training, spend $500 on ads, publish content weekly, update our directory listings, work on some case studies, and explore Reddit. Some of it went better than the rest, ha.

HubSpot AEO training: done, and useful

Four lessons stuck with us.

• Write for questions, not keywords. AI search responds to conversational questions. We need to focus on the questions that our customers are actually curious about. Some examples could be: “how do I manage and qualify inbound leads on my website?” or “how can I add support 24/7 with limited staff?”
• Make answers easy to extract. AI wants a question, then a direct answer, then detail and evidence. Many of our pages bury the answer under marketing copy. HubSpot says most existing content just needs better structure, not a rewrite. (Hooray!)
• Prove you know your stuff; create content around it. Google’s E-E-A-T framework (experience, expertise, authoritativeness, trustworthiness) matters. “The Ultimate Guide to AI Chatbots” is (unfortunately) forgettable. “How Industrial Suppliers Should Handle $100K+ Website Leads” sounds like it came from someone who’s been there. For us, this means honing in on our longstanding history and business (18+ years), customer examples, lead conversion data, etc.
• Measure AEO on its own terms. Rankings don’t tell you much about AI visibility and standardized metrics are still early. The fix is a monthly check: run the questions our buyers ask through ChatGPT, Perplexity, Gemini, and Google AI, and track whether Olark shows up.

Ads: we spent (even more) money

We said we’d pay $500 and see what happens. We spent a little more: $839.81 to be exact. Here are the results: from September 10–23, our clicks campaign earned 23,230 impressions, 134 clicks (a 0.58% CTR), and one conversion, for $839.81. That’s about $6.27 per click and roughly $840 for a single conversion.

This obviously didn’t change how AI tools talk about us. You need to have an understanding of performance marketing and a monthly budget to make the most of AI ad spend.

Case studies: yay, we have some champions!

Is there anything better than a customer success story? Especially in the new world of AI, it can feel like so much noise with little proof of results. We are happy to report that multiple customers are willing to publish case studies with us. In fact, a few are even interested in hosting a Q&A with our CEO to share their Olark AI results with those who are curious about AI. (If you’d be interested in joining that, let us know!)

This part of the action plan feels really good. Once the case studies are finished, we’ll get ’em on our website.

Directory listings: a small archaeological dig

A lot of our “Olark Live Chat” listings out in the world were badly (and sometimes hilariously) outdated. The logos were years old, the descriptions of what we do, all of it archaic. The highlight was hunting down our DUNS number, which turned out to be two DUNS numbers. (If the internet can’t agree on how many of us exist, it’s not going to agree on what we do!)

Our CEO, Ben, spent time using AI to build a list of everywhere Olark is mentioned. We got hundreds of results: from job listing sites to community forums to podcasts and social media. It will take weeks to update it all. But this proved incredibly useful. Highly recommend you do this for your company.

Weekly content: we didn’t do it

Oops? One theory was that consistency is what we’re missing, and we still didn’t publish weekly. The silver lining from the HubSpot training is that restructuring the pages we already have might count for more than publishing a bunch of new ones. But still: the learning here is that we need a better internal system in place so it becomes a priority. This is difficult for a small team, but seems necessary.

Reddit: nobody agrees on what AI chat is for

From what we’ve read, Reddit is a powerful tool for AEO. We dug into it a bit. What’s awkward here is that you can’t just go onto Reddit and market your product. Apparently that’s a no-no. You need to develop a strong reputation across the forums to be trusted. So the best action is to make like Mr. Rogers and be a friendly, helpful neighbor. We browsed around to see what we could find and where we could be helpful.

What was interesting is that Reddit commentary mirrors what we hear in sales calls. There is no single use case. We are all grappling with what AI can do for our businesses. Some need AI wired into their CRM, and some want a frontline lead catcher. A consistent theme is justifying the budget. Also, at the end of the day, how do you keep a human in the loop?!

The same names keep coming up: Intercom, HubSpot, Zendesk, and Drift. The gap we noticed is in the middle. There are businesses that don’t need a full help desk but do want a solid lead-gen tool, and nobody seems to own that conversation. Maybe that’s where Olark fits. How we show up there without being spammy is still TBD. Reddit is definitely a long play, and you don’t fully control the journey.

Where that leaves us…

We feel better trained, but we also need a better schedule, more cleanup, and more focus on content. That work takes consistency and time.

Now that we’ve completed 4 weeks of AEO, we’d love to hear from you. Did you like this series and would you like it to keep going? Please let us know if you found this content helpful, and if there is something specific you’d like us to dig into. We’d be happy to! We’re all in this together.

Team Olark

P.S. If you’re curious about what’s new with us, you can now build a free AI chatbot on the Olark site. Let us know what you think. We love feedback!