JI
JI
EXPERIMENT · LINKEDIN ADS

LinkedIn Ads Experiment: 42 Split Tests That Changed Our CPQL Model

A structured experiment log of 42 LinkedIn Ads split tests across creative, audience, and offer variables — with lifts, losses, and no cherry-picking.

#linkedin#paid#experiments#b2b
By Junaid Imtiaz · Published February 28, 2026 · Updated March 30, 2026
SECTION 01

Executive summary

Problem — Public LinkedIn Ads guidance is dominated by winner-only anecdotes. Operators have no honest ledger of what actually beats control.

Why it matters — LinkedIn is now the most expensive B2B channel — decisions must be data-backed, not vibes-based.

Who should read this
  • Paid media leads
  • Demand gen managers
  • Growth consultants
Business implications
  • Only 38% of tests beat control at p < 0.05. Half of what teams ship is noise.
  • Creative accounts for 62% of the total observed CPQL improvement — audience tweaks are ~2× less impactful than teams assume.
  • Video ads outperformed static in 71% of paired tests — but only when ≤15 seconds.
TOP FINDINGS
1. Win rate
38%

38% of tests beat control at p<0.05.

2. Creative dominance
62%

62% of aggregate CPQL improvement came from creative variables.

3. Video vs static
71%

Video ≤15s beat static in 71% of paired tests.

4. Doc ads outperform
-34%

Native document ads had a median 34% lower CPQL than single-image.

5. Job title vs skills
8/12

Skills targeting beat job-title targeting in 8/12 head-to-heads.

6. Frequency ceiling
+22%

CPQL degraded 22% once frequency exceeded 8 impressions per user in 14 days.

7. Landing page role
21%

Landing page changes contributed 21% of aggregate CVR improvement — not the ad.

SECTION 02

Research objectives

OBJECTIVES
  • Publish an honest, pre-registered LinkedIn Ads experiment log.
  • Attribute lift by variable category (creative / audience / offer / landing).
RESEARCH QUESTIONS
  • What variable class contributes the most CPQL improvement?
  • How often do LinkedIn Ads changes actually beat control?
  • Where are the frequency and creative-fatigue ceilings?
SECTION 03

Methodology

SAMPLE
42 experiments across 4 SaaS accounts
PERIOD
March 2025 – February 2026
CONFIDENCE
95% confidence per test
METHOD
Sequential split tests with statistical significance thresholds
DATA SOURCES
  • LinkedIn Campaign Manager conversion data
  • Downstream CRM opportunity data (HubSpot)
SELECTION CRITERIA
  • Minimum $8k spend per variant
  • Minimum 100 conversions per variant before significance testing
STATISTICAL METHODS
  • Two-tailed z-test for proportions
  • Bonferroni correction for multi-variant tests
VALIDATION
  • External review by 2 independent paid media consultants
BIAS CONSIDERATIONS
  • Learning-curve bias — later tests benefit from earlier ones.
SECTION 04

Data & visualizations

Aggregate CPQL improvement by variable class
016314762CreativeLanding pageAudienceOffer/CTAShare of total CPQL improvement (%)
Test outcomes
010192938Won (p<0.05)Lost (p<0.05)Inconclusive% of tests
SECTION 05

Tables

TOP 10 WINNING VARIANTS (MEDIAN LIFT)
#Variant classChangeMedian CPQL liftp-value
1CreativeVideo ≤15s vs static-38%0.002
2FormatDocument ad vs single image-34%0.004
3CreativeFace in first frame-28%0.01
4LandingAbove-fold value prop rewrite-24%0.02
5CreativeBold data point in headline-22%0.03
6AudienceSkills > Job title-18%0.04
7Offer'Free audit' vs 'Book demo'-16%0.03
8CreativeCarousel vs single image-14%0.04
9LandingRemoved nav on LP-11%0.04
10AudienceCompany size 200-1000-9%0.04
SECTION 06

Analysis & insights

  • 01Most LinkedIn teams over-invest in audience tinkering and under-invest in creative velocity.
  • 02The 'won' bar is lower than everyone thinks — expect ~4 in 10 tests to move the needle.
  • 03Landing pages are the second-largest lever and the least tested.
SECTION 07

Recommendations

Ship 4 creative variants for every audience change
High priority

The 62% creative contribution is the clearest signal in the dataset.

Add landing page tests to the roadmap
High priority

21% of aggregate improvement is currently untouched at most companies.

Retire underperformers within 14 days
Medium priority

Frequency-driven decay is real — the 22% CPQL degradation is not recoverable.

Limitations
  • Sample is B2B SaaS only. Consumer verticals will differ.
  • Learning-curve bias may inflate later-test win rates.
SECTION 08

Download the full research package

SECTION 09

FAQ

Are the raw experiment logs shared?

Yes — the CSV includes all 42 experiments, including losers and inconclusives.

Why publish losers?

Because a 38% win rate is the honest number. Winners-only libraries lie to operators.

STAY IN THE LOOP

Get the monthly research digest

One email per month with the latest SaaS marketing research, benchmarks, and experiments. No spam.

Join the newsletter
KNOWLEDGE GRAPH · LINKEDIN ADS

Continue exploring

Recommended next steps chosen by topical relevance — not popularity.

Apply a framework
Related articles
Related frameworks
Related tools
Related assessments
Related templates
Related research