Happy Friday.
Friendly reminder that we fixed our Advertise With Us form. And we do mean friendly.
As in, it works now. If you tried in the last month or so and got nowhere, that wasn’t you. That was us. We’ve since had a word with ourselves.
So whenever you’re ready, the door is open.
ChatGPT ads are now open for testing
Ads are coming from inside the chatbot.
A new friend for your media mix: OpenAI launched a self-serve ads manager in beta in the US, which will allow you to set up your campaigns right away.
What’s included: Self-serve buying, measurement tools, bidding options, and third-party attribution. A strong infrastructure to back OpenAI’s claims about “the biggest new ad channel”.
Also, you’re not locked into OpenAI’s native UI. Your existing performance toolkit can come with you.
A few things to note:
- Self-serve access means lower barriers to testing.
- Third-party partners mean easier cross-channel measurement.
- Conversational placement puts your ad inside a decision.
Also, chatbot users behave differently than search engine users. They ask it for recommendations, comparisons, and buying decisions.
A Google search is a query, but a ChatGPT conversation is closer to a decision at all stages. Reach someone at that point and you’re playing a different game.
We’ll keep you posted when the first reports come out.
Snap’s reach is shrinking, TikTok’s comments are driving sales, and Reddit’s gaining financial clout
The internet never sleeps, and neither does your to-do list.
Snapchat’s growth numbers are doing a lot of PR work: The platform added 9M daily active users, bringing the number up to 483M.
Good news? Partly. Every single one came from low-revenue regions, while North America and the EU both lost users. Oof.
Snap’s Dynamic Product Ads still deliver, however, growing 30% YoY.
If your audience skews North American or European, your reachable pool is shrinking, which squeezes delivery, costs, and results. The topline looks healthy but your target markets may not.
Your TikTok comment section is closing sales: 77% of shoppers search for more product info after seeing affiliate content, per TikTok’s own data. And the comments seal the deal.
Comments keep the buying momentum going after the video stops. The video starts the conversation while the comment section finishes it.
Think about what conversations you want to start, prompt customer responses, and follow up with educational content.
Tutorials and “why it works” videos consistently drive second purchases.
Reddit is where people go to make up their minds about money: 70% of finance shoppers use the platform to validate major purchase decisions, pressure-testing options in expert communities before they commit.
Interactive formats drive nearly 2x engagement in financial services, and a Kantar study found Reddit outperforms other social platforms for long-term brand building.
If you’re in FinServ, that’s a combination worth paying attention to.
See how Meta, Google, and AWS are actually using AI in 2026 — for free
We’re co-hosting AI Skills Conf, a free virtual conference where speakers from Meta, Google, AWS, Scale AI, and Bolt show you what’s actually working in 2026.
A taste of the agenda:
- Build your AI Chief of Staff from scratch — memory, projects, and daily briefs
- The 2026 AI tool stack for founders and small-business owners, the panel we’re moderating
- AI ROI reality check: which use cases are delivering business value
Lineup: Paide Bailey from DeepMind, Frantz Lohier from AWS, Stuart Clark from Spotify, and more!
How to use Claude Skills to improve your PPC performance?
Using AI and building PPC workflows with AI are two very different jobs.
The core problem: One-off prompts don’t scale. Ask Claude to score your search terms today, and you might get letter grades. Ask again, and you’ll get a percentage. Not good.
But the team at Search Engine Land has a solution: Skills.
What’s a Skill? If you don’t know already, think of it as a saved playbook. A task-specific, procedurally precise set of instructions saved as a Markdown (.md) file.
It defines the steps, the logic, and the output format for a given task. Claude loads it automatically when you assign a relevant job.
Why this matters: The goal is consistency. The same task, run on different days, should produce the same format every time. That’s what makes it possible to plug AI into a larger workflow.
How to build one: Start a regular Claude chat, hand it an existing SOP, audit checklist, or process document, and ask it to convert that into a Skill. Claude uses a built-in Skill-building protocol to structure the output correctly. That’s easy.
One important guardrail: Don’t let competing Skills coexist. Two Skills that both run Google Ads audits will undermine the consistency you built in the first place.
Keep your Skill library clean.
Bottom line: The real upgrade comes when you pair a Skill with live data via Model Context Protocol.
Without it, a Skill is smart but static. With it, the AI pulls real-time account data and executes, not just recommends.
Finally, here are for Skills worth building first:
- Search term mining: Feed it your search query SOP and have it flag wasted spend and negative keyword opportunities on a defined schedule.
- Ad copy generation: Give it a landing page URL, target keywords, and your brand guidelines; it handles headline and description variations from there.
- Account auditing: Convert your existing audit checklist into a Skill and let it run the same checks consistently across every account.
- Budget reallocation: Encode your performance thresholds and lookback windows so the AI knows exactly when and how to shift budget between campaigns.
Read the full breakdown in the original post here. Happy building.
Sound like the sharpest marketer in the room… even if you only skim this over coffee
Financial headlines can feel overwhelming.
Get calm, clear insights—without the fear—from The Daily Upside. Some call it the closest thing to earning your MBA in Cambridge.
Join 1M subscribers who trust The Daily Upside.
What’s the best way to talk to your customers and not annoy them?
Despite the rise of TikTok or Slack, when it comes to brand communication, customers are acting like it’s the early 2000s.
They want you to stay in the inbox:
A massive 59% of consumers prefer email as their primary communication channel with brands.
Text/SMS comes in at 18%, but it’s a steep drop-off from email. Surprisingly, only 5% prefer communicating via social media, and app push notifications sit at 13%
This reinforces that email is the “commercial” channel. People expect offers there.
Social media is for entertainment, and texts are for friends (or delivery updates). Intruding on the wrong channel can have an effect opposite of building loyalty.
What you can do: Don’t neglect your newsletter. Send your highest-value loyalty content and offers via email and don’t bury them in an Instagram Story or similar.
AI EDUCATION: ChatGPT, Claude, Gemini, Midjourney… So many names, but what’s actually useful for you in your work? There’s a newsletter called The Deep View that exists to sift through all the noise and get you up to speed on what’s actionable with AI products, and it’s free. Join 512,000+ subscribers with one click and let AI empower you.*
TIKTOK: The platform’s AI Overview feature started describing Charli D’Amelio as “blueberries with toppings” and Shakira’s new video as “blue shapes moving across a screen,” so the platform pulled it back and limited it to product identification only. Doesn’t sound reliable, no?
THREADS: Two new tools landed on Threads: Automatic text splitting for pastes over 500 characters, and animated stickers in the post composer. Text splitting is live for everyone, while stickers are still being tested in some regions.
YOUTUBE: A mobile UI test is moving the Subscriptions feed from the bottom navigation to a swipeable top tab, sitting next to Home and possibly Movies and TV. It’s a small test on Android and iOS, with a wider rollout tied to early results. Watch your subscriber engagement numbers closely if this rolls out to you.
MICROSOFT: Custom columns in Microsoft Advertising now cover all conversion metrics, including primary and full conversions. Build your own blended CPAs, ROAS calculations, and ratios tied to what your business actually cares about. Reporting that reflects your goals beats reporting that reflects what the platform finds convenient to show you.
GOOGLE: Flooding your site with AI content gets you a short traffic bump before Google samples the new URLs and starts pulling back if engagement is low. The problem is never the AI; it’s skipping quality control. Editorial standards are a crucial part of the AI content process.
META: The Edits app now shows engagement rates, peak moments in clips, and how each Reel compares to your other uploads. That last feature is the most useful, giving you a direct benchmark against your own content rather than vague industry averages. Use it to spot what your audience rewards, then do more of that.
*This is a sponsored post.
You’re in a room with no windows and no doors.
All you have is a mirror and a table.
How do you get out?
You can find the answer here.
Interested in advertising in one of our newsletters?
Connect with over 100,000 of the world’s best marketers who read a newsletter by Stacked Marketer.
