Most businesses start with interest targeting and wonder why their ads don’t convert. Interest targeting is the weakest form of social ad targeting available — and it’s what the platforms push you toward by default because it spends budget fast.
There’s a clear hierarchy of targeting quality on social platforms. The closer your audience is to people who’ve already taken action with your business, the better your ads perform. Here’s how the hierarchy works and where to focus your budget.
The Targeting Hierarchy — Best to Weakest
Understanding this order changes how you think about ad spend.
1. Custom Audiences (highest quality) 2. Lookalike Audiences (high quality) 3. Retargeting audiences (high quality) 4. Behavioral targeting (medium quality) 5. Interest targeting (low quality) 6. Broad demographic targeting (lowest quality)
Most ad campaigns run by inexperienced marketers spend the majority of budget at levels 5 and 6. Most campaigns run by experienced marketers spend the majority at levels 1–3. The performance gap between these approaches is not marginal — it’s typically 3–5x in cost per result.
Custom Audiences: The Highest-Intent Targeting Available
A custom audience is built from data you already have: your customer email list, website visitors (via the Meta Pixel or LinkedIn Insight Tag), app users, or people who’ve already engaged with your social content.
The key insight: these people have already expressed interest in your business. Showing them an ad costs the same as showing a stranger an ad — but the conversion rate is dramatically higher.
Customer list audiences: Upload your existing customer or lead email list to Meta or LinkedIn. The platform matches those emails to accounts. Typical match rates are 40–60% on Meta, lower on LinkedIn. Use this to run exclusion audiences (stop paying to reach people who’ve already converted) or to cross-sell existing customers.
Website visitor audiences: Install the Meta Pixel or LinkedIn Insight Tag on your site. This creates audiences based on page views — all visitors, specific page visitors (like your pricing page), or visitors who triggered a specific event (checkout started, form submitted). A visitor who’s been to your pricing page is a fundamentally different prospect than someone who’s never heard of you.
Engagement audiences: People who’ve watched your videos, clicked your ads, or messaged your page. These audiences tend to be large and responsive.
Lookalike Audiences: Scaling What Works
A lookalike audience takes a custom audience (your best customers, your website converters, your email list) and finds new people who share demographic and behavioral characteristics with that group.
The quality of a lookalike audience is entirely determined by the quality of the source audience. A lookalike built from 50 recent buyers will outperform one built from 10,000 random website visitors. Garbage in, garbage out.
Lookalike sizing: Platforms let you choose how closely the lookalike should match the source. 1% lookalikes are the tightest match and typically perform best for conversion campaigns. 5–10% lookalikes are larger but less precise — better for brand awareness at scale.
On Meta, lookalike audiences require a minimum source audience of 100 people (matched, not uploaded). For best results, use at least 1,000 matched people. LinkedIn lookalikes require a minimum matched audience of 300.
Retargeting: Reaching People Who Already Know You
Retargeting is showing ads specifically to people who’ve already interacted with your business — visited your website, watched a video, clicked an ad — but haven’t converted. This is separate from custom audiences above but operates on the same logic: prior interaction predicts future action.
Retargeting campaigns consistently produce the lowest cost per conversion of any social ad campaign type because the audience already has context. A prospect who visited your services page yesterday and sees your ad today is in a very different mindset than a cold prospect.
Retargeting window: Most platforms default to 30-day windows. For considered purchases (high-ticket services, B2B decisions), extend to 90–180 days. For impulse or low-consideration purchases, 7–14 days may be sufficient.
Retargeting frequency: Cap impressions at 3–5 per user per week. Above that, you’re spending money on people who’ve consciously decided not to convert yet.
Behavioral Targeting: Platform-Specific Data
Behavioral targeting uses the platform’s own data about what users do — not just what they say they’re interested in. Meta’s behavioral categories include things like “frequent travelers,” “engaged shoppers,” or “small business owners” — these are inferred from actual behavior on and off Facebook, not self-reported interests.
Behavioral targeting is meaningfully better than interest targeting because it’s based on actions, not categories people clicked. An “engaged shopper” (someone who regularly clicks on and purchases from ads) is a better audience for an e-commerce ad than someone who says they’re interested in shopping.
On LinkedIn, behavioral data includes job title changes, skills endorsements, and content engagement within the platform — which makes it unusually precise for B2B targeting.
Interest Targeting: Why It’s Weak (And When to Use It)
Interest targeting targets people who’ve indicated interest in a topic category — “small business,” “photography,” “fitness.” The problem is the category size. “Small business” on Meta targets tens of millions of people with wildly varying levels of relevance. The signal-to-noise ratio is poor.
Interest targeting works in two limited scenarios:
- Cold audience top-of-funnel when you genuinely have no custom or lookalike audiences to build from (a new business with no customers and no website traffic)
- Combined with behavioral and demographic filters to narrow a broad interest category down to something approaching precision
Use interest targeting as a placeholder while you build custom audiences — not as a long-term strategy.
Platform Differences That Matter
Meta (Facebook/Instagram): Largest audience data set of any social platform. Custom and lookalike audiences perform well across nearly all business types. Pixel-based website retargeting is reliable. Best for B2C, e-commerce, and local businesses.
LinkedIn: More expensive per click ($5–$15 CPCs vs $1–$3 on Meta), but uniquely precise for B2B targeting by job title, company size, industry, and seniority. Lookalike audiences on LinkedIn are smaller but more professionally relevant. Justified for high-ticket B2B services where the CPL can absorb the cost.
TikTok: Custom audiences and lookalikes work similarly to Meta. Behavioral data is strong for entertainment and consumer products. B2B targeting is weak — the user base skews younger and the professional signal is thin.
Pinterest: Interest and keyword targeting (yes, keyword targeting — Pinterest functions as a visual search engine) work well for product discovery. Custom audiences from email lists work, but smaller match rates than Meta.
How to Stack Targeting Layers
Single-layer targeting (just interests, just job titles) is almost always weaker than stacked targeting. Combining layers narrows your audience but increases relevance. An example for a B2B service business:
- Layer 1: Job title = Marketing Manager or Business Owner
- Layer 2: Company size = 10–200 employees
- Layer 3: Geographic = [Your service area]
- Layer 4: Exclude existing customers (customer list exclusion)
That combination produces a smaller audience than any individual layer — but the people in it are far more likely to be relevant prospects.
Don’t stack so aggressively that your audience becomes too small for the algorithm to optimize. On Meta, audiences under 50,000 people give the algorithm limited room to find converters. On LinkedIn, even smaller audiences work because the cost per action tolerance is higher.
Budget Allocation by Targeting Quality
A practical framework for distributing ad budget:
- 50–60% to retargeting and custom audiences (highest intent)
- 25–30% to lookalike audiences (scaling what works)
- 10–20% to cold behavioral and interest audiences (prospecting)
If you’re spending heavily on cold audiences with no custom audience campaigns running, you’re likely paying for reach that isn’t converting.
If you want an honest read on how your current social ad targeting is set up before making changes, run a quick check with Honest — our free audit tool covers social media targeting configuration alongside other performance factors.
For businesses ready to hand off social ad management entirely, see our social media management packages or review our fixed-price packages to understand what’s included.
FAQ
How large does my email list need to be to create a useful custom audience? On Meta, you need at least 1,000 matched contacts for a custom audience to perform well; 5,000+ is better. Match rates run 40–60%, so upload at least 2,000 emails to reliably get 1,000 matched. On LinkedIn, 300 matched members is the minimum; 1,000+ gives better results.
Is it worth running social ads without the Meta Pixel installed? No. Without the Pixel, you can’t retarget website visitors, build lookalike audiences from converters, or track which ads are actually producing results on your site. Install the Pixel before spending a dollar on Meta ads.
What’s the difference between a 1% and a 5% lookalike audience? A 1% lookalike contains the people most similar to your source audience in the target geography. A 5% lookalike is five times larger but includes people with less in common with your source. For conversion campaigns, start at 1–2%. For awareness at scale, expand to 5–10%.
How long does it take to get meaningful data from a new ad campaign? Meta’s algorithm typically needs 50 conversion events in a 7-day window to exit the learning phase. LinkedIn needs similar volume. This means a campaign generating fewer than 50 conversions per week won’t optimize effectively. Either increase budget or use a higher-funnel conversion event (link clicks, landing page views) until conversion volume grows.
Should I use detailed targeting expansion on Meta? Meta’s “Advantage Detailed Targeting” (formerly “Expansion”) allows the platform to show your ads to people outside your specified targeting if it predicts better results. For cold audience prospecting, it can improve efficiency. For retargeting campaigns, turn it off — you don’t want to expand outside your known audience.
What’s a good cost per click to aim for on social ads? This varies significantly by industry and platform. Meta averages $1–$3 CPC across industries; LinkedIn averages $5–$15. What matters more than CPC is cost per qualified lead or cost per acquisition. A $10 CPC that produces a $50 CPL is better than a $1 CPC that produces a $200 CPL.