Meta Ads in Depth

The furniture brand that turned £3,000 into £28,000

A small furniture maker in Bristol had been trying Meta Ads for six months with mixed results. She'd spent roughly £3,000 across various campaigns with limited return.

In January 2023, she made one structural change to how she ran her campaigns.

Instead of targeting broadly by interest ("home décor," "interior design") for all ads, she built a three-audience structure:

Cold audience: A lookalike audience of her 200 existing customers (uploaded email list). She ran video ads showing the furniture being made — the wood, the craft, the process. No price. No shop link. Just the story.

Warm audience: Website visitors from the last 30 days who hadn't purchased. She ran carousel ads showing her bestsellers with prices. "You browsed this." Direct and specific.

Hot audience: Add-to-cart abandoners (tracked via Meta Pixel). She ran a single image ad: "Your [specific product name] is still available. Free delivery this week." Urgency, but genuine.

Same budget. Three-audience architecture. ROAS went from 2.3× to 9.4× in 60 days.

(Illustrative scenario based on patterns common in direct-to-consumer e-commerce advertising. Specific figures are representative of real-world outcomes — not a verified account of a specific named company.)

Meta's campaign structure

Meta Ads uses the same three-level hierarchy as Google, but the logic differs:

Choosing the right objective:

The campaign objective tells Meta's algorithm which type of person to show your ad to — someone likely to click, someone likely to fill a form, someone likely to purchase. Choosing the wrong objective is one of the most expensive mistakes in Meta advertising.

The critical rule: For any campaign with a commercial goal (getting customers, leads, or sales), use Leads or Sales — never Traffic. Traffic campaigns send clicks; they don't optimise for the type of person likely to convert. The CPC is slightly lower but the CPA will be much higher.

The Meta Pixel: the foundation of performance

The Meta Pixel is a snippet of JavaScript code installed on your website that tracks what visitors do after clicking your ad — and sends that data back to Meta.

What the Pixel enables:

• Conversion tracking: Was this purchase or lead form from an ad? Which ad?
• Retargeting: Show ads specifically to people who visited your site, viewed a product, or added to cart
• Lookalike audiences: Meta finds people who statistically resemble your converters
• Optimisation: Meta's algorithm learns which types of people convert, then finds more like them

Without the Pixel: You're flying blind. You see clicks but not conversions. You can't retarget. You can't build lookalikes from your best customers. You can't use the Sales objective effectively.

Installing the Pixel: One-time setup through Meta Business Manager → Events Manager. Platforms like Shopify have one-click integrations. WordPress requires a plugin or code installation. Always verify the Pixel is firing correctly using Meta's Pixel Helper Chrome extension.

Standard events to track:

• PageView — someone visited your site
• ViewContent — someone viewed a specific product page
• AddToCart — someone added a product to their cart
• InitiateCheckout — someone started checkout
• Purchase — someone completed a purchase (with value)
• Lead — someone completed a lead form
• CompleteRegistration — someone signed up or registered

Each event builds an audience segment for retargeting and gives Meta's algorithm data to optimise against.

Audience architecture: the three-layer system

The furniture maker's success came from understanding that Meta audiences are not equal — and not interchangeable.

Cold audiences (no prior awareness of your brand):

• Interest targeting: People who follow competitor pages, have relevant interests, or have demonstrated relevant behaviours on Facebook/Instagram
• Lookalike audiences: Meta finds people statistically similar to a seed audience (your customer email list, Pixel purchasers, video viewers)
• Broad targeting: No specific targeting — Meta's algorithm finds the right people from scratch, using conversion event data as the signal

Warm audiences (have interacted with your brand):

• Website visitors (all visitors; specific page visitors; within time windows like "visited in last 30 days")
• Video viewers (people who watched 25%, 50%, or 75%+ of your videos)
• Instagram/Facebook engagers (liked, commented, messaged, or visited your profile)
• Email list (uploaded as a custom audience)

Hot audiences (have taken a strong action):

• Add-to-cart abandoners
• Checkout initiators who didn't purchase
• Lead form openers who didn't submit
• People who visited the pricing page 2+ times

The audience exclusion principle: Always exclude recent purchasers from non-purchaser ads using custom audience exclusions (e.g., exclude your "Purchasers — last 30 days" custom audience). Note: Meta removed the ability to exclude by detailed targeting categories (interests, demographics) beginning in late 2021 and expanding significantly through 2022–2024 (check Meta's current documentation for the latest restrictions), but exclusions using custom audiences and pixel-based lists remain fully available. Overlapping audiences waste budget and send conflicting signals to Meta's algorithm.

Meta ad creative: what actually works

Meta advertising is a creative game more than a targeting game. The algorithm is good at finding the right people — your job is to give it creative that makes those people act.

Creative formats and when to use them:

Single image: Fast to produce; strong for simple offers; works well for retargeting (show the specific product they viewed). Best with a clean background, minimal text, and strong visual contrast.

Carousel (multi-image): Ideal for showcasing multiple products, showing a before/after, or telling a sequential story. Each card can have its own headline and link. Works well for e-commerce product discovery.

Video: Highest potential engagement; required for building video viewer audiences (for retargeting). First 3 seconds must arrest attention — thumb-stopping hook is everything. Facebook/Instagram video with captions (most video is watched without sound).

Stories/Reels ads: Fullscreen vertical format. Must feel native to the platform (looks like organic content, not a polished ad). Tap-forward rate is the primary engagement metric — fast pace, strong hook in frame one.

The creative testing framework:

Don't guess what creative works. Test systematically:

• Start with 3–5 creative variations in the same ad set
• Allow each at least 3–5 days of data before evaluating
• Pause the worst performers; iterate on the winners
• Test one variable at a time when iterating (hook vs. hook, not hook + offer + format simultaneously)

Using AI for Meta ad creative: "Write 5 Facebook/Instagram ad concepts for [product/service] targeting [audience]. For each: describe the visual (what's in the image or video), the hook text (first line of primary text, designed to stop the scroll), the primary text (under 125 characters, one clear benefit), the headline (under 40 characters, one clear CTA direction), and the CTA button text. Make 2 of the 5 feel like testimonial/social proof ads, 2 feel like product demo ads, and 1 feel like a story-based ad."

The Meta attribution window

Meta attributes conversions to ads based on a configurable attribution window — the time period after an ad click or view in which a conversion is counted as "from" the ad.

Default attribution window: 7-day click, 1-day view This means: if someone clicked your ad and purchased within 7 days, the ad gets credit. If they saw the ad (without clicking) and purchased within 1 day, the ad gets credit. For video ads, a third window applies: 1-day engaged view — someone who watched at least 10 seconds of your video ad and converted within a day.

Why attribution matters: Different attribution windows show different ROAS figures. A campaign with 7-day attribution may look profitable, but a 1-day-click-only window (better for measuring direct response) may show a much lower number. Understanding the attribution window you're looking at is essential for interpreting performance data accurately.

There Are No Dumb Questions

"How much should I spend per ad set to get meaningful data?"

A useful rule of thumb: allow your ad set to accumulate around 50 conversion events before evaluating performance. If your target CPA is £20, that's roughly £1,000 in spend — but the event count matters more than the spend threshold. For smaller budgets: at minimum, wait for 5–10 conversion events (or 5–10× target CPA in spend) before any significant changes. With fewer events, the algorithm hasn't had enough data to learn, and performance fluctuates too much to draw conclusions.

"Should I use Advantage+ (Meta's AI-optimised campaigns) or manual targeting?"

Advantage+ campaigns have outperformed manual targeting for many advertisers with sufficient conversion data — and they're now available across objectives, not just Shopping. For businesses with 50+ monthly conversions tracked by the Pixel, Advantage+ is worth testing against manual setups. For beginners or businesses with insufficient conversion data: manual audience targeting with a structured cold/warm/hot architecture gives you more control and learning. Don't use AI optimisation before the algorithm has data to learn from.

Back to the Bristol furniture maker

After six months of mixed results, the structural change wasn't the budget or the platform — it was the message matched to the audience. Cold visitors saw the craft: the wood being worked, the process, the person making it. Warm visitors saw the products they'd already browsed, with prices. Cart abandoners got urgency about the specific item they'd left behind. The £3,000 that had produced scattered results became £28,000 in bookings at 9.4× ROAS — because the ad worked at each stage of the relationship. People didn't just buy furniture. They bought into the maker behind it, which the creative made possible.

Key takeaways

• Campaign objective is the most important campaign setting. Use Sales or Leads for commercial goals — Traffic campaigns optimise for clicks, not conversions.
• The Meta Pixel is non-negotiable. Without it, you can't track conversions, build retargeting audiences, or create lookalikes. Install it before running any conversion campaign.
• Three-audience architecture (cold/warm/hot) outperforms single-audience campaigns. Each audience stage needs different creative, different messaging, and different expected CPA.
• Creative is the primary performance lever on Meta. The algorithm finds the audience — your creative determines what that audience does. Test systematically, kill losers fast.
• Exclude audiences to prevent overlap. Purchasers excluded from non-purchaser campaigns; warm audiences excluded from cold prospecting. Overlap wastes budget.