- Custom audiences built from website visitors and past buyers outperform broad targeting for ecommerce.
- Lookalike audiences need at least 100 seed members; Meta recommends 1,000–5,000 for best match quality.
- A 1% lookalike built from your top customers is one of the strongest cold audiences in ecommerce.
- Custom audience uploads process within 24 hours and support unlimited total records.
- Layer age, gender, and device targeting on any audience to cut wasted spend fast.
- As of September 2, 2025, Meta restricts audiences that suggest health conditions or financial status.
Why Audience Targeting Matters for Ecommerce on Instagram
The right audience cuts wasted spend fast. Instagram runs on Meta's ad platform, which gives ecommerce advertisers some of the most detailed targeting options available. Per Meta's Marketing API documentation, there are two core approaches: specific targeting using your own customer data, and broad targeting using demographics and interests. For ecommerce, your own data wins every time.
The Two Core Audience Strategies
Broad Targeting: Demographics, Location, and Interests
Broad targeting uses publicly available signals. Age, gender, location, interests, behaviors, and life events. It works when you have no customer data yet, or when testing a brand-new market.
The risk is noise. A 25-34-year-old interested in "fitness" could represent ten different buyer profiles. Broad targeting takes longer to optimize and typically costs more per conversion than a warm audience.
Use it to fill the top of your funnel. Not to close sales.
Specific Targeting: Data-Driven Custom Audiences
Specific targeting uses data you already own. Website visitors. Past purchasers. Email subscribers. App users. These people already know your brand. They convert at higher rates and lower cost.
Meta's documentation confirms custom audiences can be built from email lists, phone numbers, CRM records, website behavior, mobile app activity, and engagement signals. This is the backbone of a strong ecommerce targeting strategy.
Custom Audiences: Your Best Performing Ecommerce Segments
Custom audiences are your closest thing to guaranteed relevance.
Website Visitors and Converters (Pixel-Based)
Install the Meta Pixel on your site. Then build audiences from everyone who visited a product page, added to cart, or completed a purchase. These are high-intent signals.
Retarget cart abandoners with a direct offer. Show past purchasers a complementary product. Exclude recent buyers from acquisition campaigns. Each segment earns a different message for a different moment in the buyer journey.
Customer List Audiences (Email/Phone/CRM)
Upload your customer email list or phone numbers. Meta matches records against user accounts. Per Meta's customer file custom audience documentation, uploads process within 24 hours and support up to 10,000 records per batch with no limit on total audience size.
Your best seed here is your top 20% of customers by lifetime value. Use that group to power lookalikes, and as a suppression list to avoid spending on people who already bought.
One important note: all uploaded data must be in hashed format. Meta's policy also restricts audiences that suggest specific health conditions or financial status. As of September 2, 2025, flagged audiences of this type cannot be used in new campaigns. Keep your list sourcing clean and compliant.
App Users and Engagement Signals
If you have a mobile app, build an audience from active users or in-app purchasers. Instagram engagement also qualifies. People who follow your account, liked a post, or watched a video are warm audiences you can retarget or exclude from cold campaigns.
These signals layer into your broader custom audience strategy and are easy to set up inside Ads Manager.
Lookalike Audiences: Scale What Works
Once custom audiences perform, lookalikes let you scale without guessing.
What Lookalike Audiences Are and How They Work
Meta finds people who share characteristics with your seed audience. Per Meta's Lookalike Audiences documentation, seed audiences require at least 100 members from a single country. Meta recommends 1,000 to 5,000 people for the best balance of specificity and match quality.
Lookalikes refresh every 3 days when active and take 1-6 hours to fully populate before delivery begins. The range runs from 1% to 10% of a selected country's population in 1% increments. A 1% lookalike prioritizes similarity. A 5% or 10% lookalike trades precision for reach.
Creating Lookalikes from Best Customers
Don't use your full customer list as a seed. Use your best customers. Filter by highest lifetime value or repeat purchase rate before uploading. A 1% lookalike built from your top 500 buyers will outperform one built from 50,000 mixed-quality records.
Start at 1% for maximum similarity. Expand to 2% or 3% once the 1% audience is fully tapped.
Lookalikes from Past Converters
Build a separate lookalike from your purchase converters. Meta tracks conversions through the Pixel. Any ad set with 100 or more conversions from the past 180 days qualifies as a seed source.
This lookalike finds people who behave like buyers, not just browsers. It is consistently one of the highest-performing cold audiences in ecommerce.
Advanced Targeting Combinations for Ecommerce
Age, Gender, and Location Targeting
Layer demographics on top of any audience. If your product sells primarily to women aged 25-44, narrow your targeting there. You are not cutting off buyers. You are focusing budget where it converts.
Location targeting works at the country, region, or city level. Exclude regions where you don't ship. That one step can meaningfully reduce wasted spend.
Interest and Behavior-Based Segments
Meta's advanced targeting includes purchase behaviors, device usage patterns, life events, and hundreds of interest categories. For cold audiences with no customer data, this is your starting point.
Be specific. "Running shoes" beats "sports." "Online shoppers who recently purchased sports products" beats both. Layer two or three tight interests rather than stacking ten broad ones.
Device and Platform Targeting
Instagram is mobile-first. Most ecommerce purchases on Instagram happen on mobile devices. Confirm your landing page and checkout are mobile-optimized before scaling spend.
Platform targeting also lets you limit delivery to Instagram only within a Meta campaign. This keeps your creative testing and reporting clean. Feed spend and Reels spend stay separate. Performance data stays actionable.
Building Your Audience Strategy with Coinis
The Coinis Campaign Launcher guides you through audience selection during campaign setup. Choose your audience type, configure targeting parameters, and structure your ad set without switching between tools. The Brand Profile ensures every segment sees consistent messaging that reflects your brand voice, whether you're targeting cold lookalikes or hot retargeting lists.
Or let Coinis do it.
From a product URL to a live Meta campaign. AI-generated creatives. On-brand copy. Direct publish to Facebook and Instagram. Real performance reporting. All in one platform.
Start free. Upgrade when you're ready.
15 AI tokens a month. No credit card.
Frequently Asked Questions
What is the minimum audience size for Instagram lookalike ads?
Meta requires at least 100 seed audience members from a single country to create a lookalike. For best match quality, Meta recommends a seed size of 1,000 to 5,000 people.
How long does a custom audience take to process on Meta?
Custom audience uploads process within 24 hours. Each batch supports up to 10,000 records, with no limit on total audience size.
Should I use broad or custom audiences for ecommerce on Instagram?
Start with custom audiences built from your website visitors, past buyers, or email list. These convert at higher rates and lower cost than broad targeting. Use broad targeting only for top-of-funnel testing or when you have no customer data yet.
Are there any audience types restricted on Meta in 2025?
Yes. As of September 2, 2025, Meta flags and restricts custom audiences and lookalike audiences that suggest specific health conditions (such as diabetes or arthritis) or financial status (such as credit score or high income). Flagged audiences cannot be used in new campaigns.