A dark post is a paid social media ad that appears in targeted users' feeds but is not published on the advertiser's public profile or page. On Meta it is called an unpublished page post; on LinkedIn it is Direct Sponsored Content. Dark posts let marketers test many creative variants at once without cluttering the brand's public timeline.
If you have ever run 8 headline variants on the same campaign without wanting your public page to look like a spam factory, you have already run dark posts. It is the default format for any serious paid social team.
What is a dark post?
A dark post is an ad that lives entirely inside the ad platform. It is created in Meta Ads Manager, LinkedIn Campaign Manager, or the X ad tool, targeted at a specific audience, and delivered into their feed — but the post itself is never pushed to your brand's public page. Organic followers do not see it in their timeline, and anyone who visits your page will not find it there either.
The name varies by platform:
- Meta (Facebook + Instagram) — "unpublished page post"
- LinkedIn — "Direct Sponsored Content"
- X / Twitter — "promoted-only tweet"
- TikTok — "Spark Ad" (published) vs "non-Spark Ad" (dark)
"Dark" refers to visibility on your own page, not to anything hidden from the platform or the user. Users see the ad in their feed and can like, comment, and share it — the post is fully public to the audience seeing it. It is only invisible to your organic followers who are not in the target segment.
Why dark posts matter for paid social
Every high-spend paid social team uses dark posts. Three reasons drive that:
- Test velocity. You can run 5 to 20 creative variants per ad set without your public page filling up with "which of these do you like better?" spam. Faster testing means faster learning.
- Audience-specific messaging. Different audiences see different copy, offers, and hooks. A CTO sees ROI language; a founder sees speed-to-hire language. Neither one sees the other's ad.
- Brand feed control. Your public page stays a curated brand channel — announcements, wins, culture — while all the messy testing happens in the ad account.
How dark posts actually work
The workflow is the same across every ad platform:
Creative: headline_v3 + image_v7 + CTA "Get demo"
# 2. Target a specific audience
Audience: SaaS founders · US · 25-45 · LinkedIn lookalike 1%
# 3. Ad delivers to feed — but never to your page
brand.com/page → nothing new posted
audience.feed → ad delivered
How to build one on Meta
Inside Meta Ads Manager, at the ad level, choose Create Ad instead of Use Existing Post. Attach the connected page, upload creative, write copy — and skip the "publish to page" step. The resulting post has a unique ID that can be reused across multiple ad sets.
How to build one on LinkedIn
In LinkedIn Campaign Manager, open the connected company page and choose Create Content then Direct Sponsored Content. LinkedIn stores the post at the page level but marks it as unpublished — it does not appear in the page's feed.
Dark post vs boosted post vs organic post
| Type | Lives on page? | Paid? | Best for |
|---|---|---|---|
| Dark post | No — invisible to page | Yes | Multi-variant testing, audience-specific creative |
| Boosted post | Yes — starts organic | Yes (added later) | Quick reach amplification for a proven post |
| Organic post | Yes | No | Brand voice, community, algorithm signals |
| Whitelisted creator ad | On creator's page | Yes | Running ads through an influencer's handle |
Real dark post examples
Two patterns cover 80% of how teams use them.
1. Multi-headline test at scale
A B2B SaaS team ran 8 headline variants on the same image, targeted at 4 job titles on LinkedIn (CTO, VP Eng, Head of Product, Founder). That is 32 dark posts total. After 5 days the "CTO + outcome-focused headline" combo showed a 3x higher CTR than the average. The team killed 28 variants and scaled the winner.
2. Offer test by neighborhood
A local fitness studio ran two offer variants on Instagram — "50% off first month" vs "free trial class" — targeted at 6 different neighborhoods within 5 miles. The 50% offer beat the free trial by 40% on cost-per-lead in every neighborhood. Neither offer ever appeared on the studio's public feed.
3. Retargeting with different creative per stage
audience: no site visit → "Meet theStacc — SEO on autopilot"
# Warm audience — case study proof
audience: pricing page visit → "How Fairview grew 3x in 90 days"
# Hot audience — direct offer
audience: demo page abandoners → "Book a demo →, get $500 credit"
Dark post vs boosted post — which to use
Both are paid. The difference is whether the post exists on your page.
Use a dark post when
- You are testing multiple creative variants
- Different audiences need different messaging
- You do not want ad copy on your public feed
- You are running retargeting sequences
- You want reusable post IDs across ad sets
Use a boosted post when
- An organic post is already performing well
- You want social proof (likes, comments) to compound
- You need a quick reach lift with minimal setup
- The audience is broad, not segmented
- The post is core brand content worth showing everyone
7 best practices for dark posts
- Test one variable at a time. Change headline OR image OR CTA — not all three. Otherwise you cannot attribute the win.
- Run 3 to 5 variants per ad set. More than 5 spreads budget too thin for the platform to find a winner.
- Give each variant 3 to 5 days. Kill the bottom 40% after that window and reallocate budget to the top 20%.
- Reuse winning post IDs. A single winning dark post can be dropped into 10 new ad sets. Social proof (likes, comments) compounds across placements.
- Match creative to audience temperature. Cold audiences need hooks; warm audiences need proof; hot audiences need offers.
- Track by post ID, not campaign name. A single winning creative can outlive 5 campaign iterations.
- Preview on desktop and mobile. A hook that works on mobile Instagram can die on desktop Facebook. Check both before spending.
Dark posts collect likes and comments just like any post — but you cannot see them on your page. Set up a comment monitoring workflow (Meta's Inbox, Sprout, or a Slack alert) so you do not leave angry customers or fresh leads sitting in dark-post comment threads for weeks.
Common dark post mistakes to avoid
- Ignoring comment moderation — comments still happen, and negative ones can tank your relevance score if unanswered.
- Testing too many variables at once — makes results unattributable and wastes budget.
- Killing variants too early — under 3 days is usually noise, not signal.
- Not reusing winning post IDs — losing all the social proof stacked on the winning creative.
- Broad targeting with dark posts — if the audience is huge, a boosted post plus algorithm optimization is cheaper.
- Forgetting UTM tags — dark posts do not carry organic post metadata. Tag every link.
Frequently asked questions
No. A dark post appears only in the feeds of users who match the ad targeting. It never shows up on your public page, profile, or timeline, so organic followers do not see it unless they are inside the targeted audience.
A boosted post starts as organic content on your page and receives paid distribution afterward. A dark post is created inside the ad manager from the beginning and never appears on your page organically.
No. Dark posts use the same auction pricing as any other paid social ad. Cost is driven by targeting, quality score, and competition, not by whether the post is published on your page.
Start with 3 to 5 variations per ad set. Test one variable at a time (headline, image, or CTA) so you can attribute performance changes cleanly. Kill the bottom 40% after 3 to 5 days and reallocate budget.
Create dark posts inside Meta Ads Manager by choosing "Create Ad" and then "Create New" at the creative step. Select the connected page but do not publish to the page. The resulting post ID can be used across multiple ad sets.
Related glossary terms
Sources
- [01]Meta Business Help — Unpublished page posts
- [02]LinkedIn — Direct Sponsored Content
- [03]WordStream — What are dark posts
- [04]Internal audit: 40-variant creative test on a B2B SaaS client — Mar 2026
