Unlock Every Facebook Targeting Option That Matters

Hello Readers 🥰

Welcome to today's edition, bringing the latest growth stories fresh to your inbox.

If your pal sent this to you, then subscribe to be the savviest marketer in the room😉

In Partnership with Tatari

Forward 2025: Where Top CEOs and Brands Unveil the Future of Advertising (Sept 4)

Customer acquisition is shifting fast – paid social is harder to navigate, TikTok faces uncertainty, and Gen Z is skipping Google for creators and AI. Forward 2025 is where ambitious brands learn what's next.

On September 4th at The Glasshouse in NYC, join this free half-day conference focused on the future of advertising.

Hear from Steve Huffman (Reddit CEO), Adam Foroughi (AppLovin CEO), and Allison Ellsworth (poppi Co‑Founder & CBO) – visionaries shaping how brands grow today. 

Plus, marketing leaders from Dave, Coterie, Saatva, TickPick, Ariat, and Gusto will share exclusive TV growth playbooks across three dynamic panels.

Cap off the day with a rooftop happy hour overlooking the Hudson, connecting with speakers and peers.

Forward by Tatari is built for ambitious brands to scale smarter in a shifting ad world, with TV as the next big growth channel.

Seats are limited. Reserve your spot and join industry leaders shaping the future of advertising.

Facebook Ads Targeting Made Simple

Facebook Ads is one of the most powerful platforms for reaching your ideal customer. With billions of users across Facebook and Instagram, the challenge isn’t finding people, it’s finding the right people. That’s where targeting comes in. By using Facebook’s wide range of targeting tools, you can refine your audience, improve ad efficiency, and maximize ROI.

Steps to Master Facebook Ads Targeting

1️⃣ Understand the Algorithm:
Facebook collects data based on user activity both on and off the platform from pages liked to websites visited. This helps it predict user interests and serve ads that match intent.

2️⃣ Use Demographic Targeting:
Filter audiences by age, gender, location, language, education, income, life events, parental status, relationship status, or job title. Start broad, then refine based on campaign performance to avoid overspending.

3️⃣ Leverage Interest Targeting:
Target people by what they care about: fitness, food, hobbies, travel, or entertainment. With countless subcategories available, you can go as broad or specific as your strategy requires.

4️⃣ Apply Behavioral Targeting:
Focus on how users act: device usage, purchase history, travel habits, or digital activity. For example, target engaged shoppers or frequent travelers if they align with your product.

5️⃣ Build Custom and Lookalike Audiences:
Upload customer lists or retarget website visitors with Custom Audiences. Then expand your reach with Lookalike Audiences, which find new people similar to your best customers.

6️⃣ Adapt to Privacy Changes:
With updates like Apple’s App Tracking Transparency, retargeting pools are smaller. Test broader targeting, refine creatives, and rely more on Lookalike Audiences to maintain reach.

The Takeaway
Successful Facebook campaigns rely on precise targeting combined with compelling creative. By understanding demographics, interests, behaviors, and leveraging custom data, you can ensure your ads reach people who are most likely to convert without wasting spend.

Bring AI Power Offline With Local Models

Want private, offline AI without per-request costs? You can run powerful open-source models locally using simple tools and a quick setup. Here’s how to get started fast with either a command-line or a ChatGPT-style interface.

Steps to Run Open-Source Models Locally:

1️⃣ Choose Your Setup Decide how you want to interact:

• Ollama for a lightweight command-line experience.
• LM Studio for a familiar chat UI that feels like standard assistants.

2️⃣ Ollama Method (CLI)

• Install Ollama for your OS.
• In Terminal, run: ollama run gpt-oss:20b.
• The first run downloads ~13GB, then the model works offline for private, local inference.
• Tip: Keep an eye on disk space and let the initial pull finish before starting longer chats.

3️⃣ LM Studio Method (GUI)

• Download and open LM Studio.
• Go to Discover and find GPT-OSS-20B.
• Click to download, then start chatting via the built-in interface, no command line required.
• You can manage multiple local models, switch contexts, and save chats.

4️⃣ Use a Web Playground (If Your Device Can’t Handle Local)

• Open the GPT-OSS Playground to try the same model family in your browser—no install needed.
• Great for lower-spec machines or quick experiments before committing to a local setup.

5️⃣ Tune Reasoning Levels for Speed vs. Depth

• Set low reasoning for quick, simple tasks (summaries, drafts).
• Use high reasoning for complex problem-solving, analysis, or multi-step planning.
• Adjust per task to balance latency and output quality.

The Takeaway
Running open-source models locally delivers privacy, reliability, and zero per-call costs without giving up strong performance. Pick Ollama for minimal setup, LM Studio for a polished chat experience, or the web playground if you’re testing on a lighter machine. Start simple, then iterate: add bigger models, refine prompts, and dial reasoning up only when you need it.

We'd love to hear your feedback on today's issue! Simply reply to this email and share your thoughts on how we can improve our content and format.

Have a great day, and we'll be back again with more such content 😍