Skip to main content
Preview our new AI chat agent trained on your business live in one minute. Preview

Week 3 of AEO: Our Action Plan

Olark September 29, 2026

Week 3 of Olark's AEO journey: our AI-sourced traffic isn't great, so here's our plan. Training, a $500 ads experiment, weekly publishing, Reddit, original data, and rebranding our footprint.

A task list titled "AEO To Do:" with two unchecked items: Team training and Test ads
A task list titled "AEO To Do:" with two unchecked items: Team training and Test ads

Quick recap: last week we admitted our AI-sourced traffic is… not great (receipts included). This week: what are we actually going to do about it? We’re sharing our plan. Hopefully it helps you, too!

Training, because we need it

Turns out we already have access to AEO training through HubSpot, we just hadn’t used it. We’re going to actually go through it, as a team, instead of guessing our way through it.

Are ads worth it? Let’s find out for $500

We don’t know yet if paid can move the needle on AI visibility (not just regular search), so instead of debating it, we’re putting $500 toward a small experiment and seeing what it actually tells us.

Publish on a schedule, not a whim

Remember last week’s theory (that AI models reward consistent publishing over occasional brilliance?) We’re going to test it. Our growth team is committing to a weekly cadence, instead of our usual “post something great whenever inspiration strikes” approach. We even made a marketing calendar. Go us.

Reddit apparently really does matter

Here’s an interesting stat: Reddit is Perplexity’s single biggest citation source, making up almost half of what it pulls from. Google’s AI Overviews lean on it heavily too. Which means we need to put some energy into it.

So we’re going to actually try to show up on Reddit, answering real questions in subreddits where people already talk about AI agents. The catch: it only works if it’s genuine. Fake accounts and scripted self-promotion get you banned. This means real people on our team answering real questions, not planting threads. We’ll give it a shot.

Publish something AI literally can’t generate itself

The strongest visibility gains come from expert answers and proprietary data. Generic content is losing value as AI can already produce it on its own. For us this means we need to be sharing original stats (e.g., something from our own AI agent data, response times, conversion patterns, whatever’s defensible) that nobody else can cite because nobody else has it. How are we going to do this? Good ole case studies. We have a few customers we’re working with to write up case studies.

Rebrand our footprint, not just our site

Here’s an easy one to miss: a lot of the internet still thinks of us as “Olark Live Chat.” G2, directories, review sites, etc. If AI models are pulling from those listings to figure out what we do, and they all say “live chat” while we’re busy becoming Olark AI, we’re sending mixed signals about our own identity. So we’re going through everything and making sure it reflects who we are now, not who we were.

Next time: how it’s actually going.

See you then (plan in hand, cautiously optimistic),
Team Olark

Olark Essentials AEO