About the Client
Nebius is a global AI cloud company delivering infrastructure and services to accelerate the full AI pipeline from foundational model builders to enterprise solutions.
Nebius Group also includes other companies growing under distinctive individual brands:
- Avride (autonomous vehicles and delivery robots),
- TripleTen (edtech and reskilling for tech careers).
- In addition, Nebius owns stakes in companies including ClickHouse and Toloka.
Headquarters in Amsterdam, Netherlands. Data centers, commercial and R&D hubs across Europe, the US, and the Middle East. Clients worldwide.
A leading AI infrastructure company building a full stack of optimised hardware and software to support AI development, backed by strong partnerships (including NVIDIA and Accel) and complemented by additional lines of business such as Avride and TripleTen.
Goals and Objectives
As part of the multichannel campaign for Nebius, the agency was tasked with the following, broken down by promotional direction:
Make starting with Nebius as simple and attractive as possible for new users of AI infrastructure — emphasizing access to NVIDIA H100 GPU at $1.50/hour for the first 1000 GPU-hours per month, with a low entry threshold ideal for testing and prototyping.
Increase awareness of AI Studio as a convenient, accessible tool for launching open-source models — low latency, flexible cost/speed balance, up to 3× savings on input tokens, and no need for MLOps. Aimed at ML engineers, data scientists, and AI startup founders.
Create a viral effect and maximize organic reach for the Deepseek post via a mass, simultaneous repost strategy with relevant AI influencers, synchronized with a moment of heightened topical interest.
Promote the AI Discovery Award within medicine, biotechnology, and research communities — engaging thought leaders across LinkedIn, Telegram, X, and YouTube to encourage nominations and raise awareness in scientific circles.
Strategy and Approach
To effectively promote Nebius, we developed a comprehensive strategy that covered multiple directions, taking into account the specifics of the products, audience, and communication channels.
Powering AI Momentum with Influence
Explorer Tier Pricing — a plan with a low entry threshold for beginner and advanced users of AI infrastructure. We focused on attracting AI developers, startups, and researchers who value accessibility and favorable conditions for testing and prototyping.
AI Studio and Inference Service — ready-made tools for fast and cost-efficient work with open-source models. Here we targeted a technical audience of ML engineers, data scientists, and AI startup founders. The main emphasis was on advanced video content that clearly demonstrates the platform’s benefits: flexible speed/cost balance, resource savings, and no need for complex MLOps setup.
Influencer marketing on X — to promote Deepseek and new releases, we launched a large-scale campaign with native opinion leaders in the AI community to generate a viral effect through synchronized reposts and the creation of a social media “event.” The key was to carefully select relevant influencers and time the release to peak audience interest.
Soperator product launch on Twitter and Reddit — we focused on DevOps professionals, HPC specialists, and the open-source community by highlighting a unique solution for Kubernetes and Slurm. We used formats popular in niche communities and delivered deep technical content aimed at experts.
AI Discovery Award in Healthcare — a brand campaign to promote the award among medical and biotech professionals by identifying and engaging thought leaders on various platforms (LinkedIn, Telegram, X, YouTube). The objective was to encourage nominations and raise awareness of the award in scientific circles.
Creative and Insights
- For TikTok and Reels, we adapted content to the behavior and interests of a younger audience, emphasizing visual dynamics and simple explanations of the Explorer Tier’s value.
- For technical channels (Twitter, Reddit), we used in-depth expert materials and native formats to hold the attention of professionals and drive engagement.
- Mass reposts with opinion leaders created a “social proof” effect and increased reach without additional spending on targeting.
- We used look-alike audiences and behavioral targeting, which helped save budget and improve conversion rates.
Media Plan and Technology
- Focused on organic reach in social networks, backed by targeted advertising.
- The funnel was built to reflect the stages of awareness → consideration → trial.
- Analytics and end-to-end tracking were implemented to monitor campaign performance in real time.
To improve campaign efficiency, we introduced automated optimization algorithms. Based on a data-driven approach — systematic analysis of user behavior, engagement, and conversions by each campaign element — the team monitored key metrics in real time and made decisions based on actual data. This allowed us to flexibly reallocate budgets toward the most effective combinations of creative, channel, and audience, and scale up the formats that showed the highest performance.
ExecutionHow It All Unfolded
The campaign began with a preparatory phase at the end of Q4 2024, including the development of creative concepts, content verification, and selection of relevant influencers. We simultaneously launched multiple product directions: Explorer Tier, AI Studio, and the release of Soperator.
Key Stages:
- Preparation: audience and channel analysis, script and video prototyping, coordination with the Nebius team.
- Testing: January–February 2025 — first posts on Instagram and TikTok, targeted reposts on X.
- Scaling: March–April 2025 — expansion of release volume, including simultaneous reposts on Twitter, Reddit launch, and second round of creatives for AI Studio.
Creative Examples
Videos and Stories on Instagram and TikTok conveyed the core USPs of the products (access to NVIDIA H100, no need for MLOps, token savings) in the format of native recommendations.
Collaboration with the Client
The process was built on regular synchronization with the Nebius marketing and ML teams:
- Creative approvals followed a centralized workflow via a Content Approval spreadsheet.
- All scripts and videos were reviewed for technical accuracy and positioning.
- Edits were made within 48 hours following feedback from the client or influencer.
- Campaign reporting and monitoring were done weekly via a live document and working meetings.
Adjustments During the Project
- After the first launches and CPM analysis, we reallocated budget toward videos with high organic performance (especially for AI Studio).
- Some bloggers did not pass content verification — their launches were stopped to maintain a unified tone of communication.
- Scripts for Stories were refined: terminology was simplified to avoid technical overload.
Based on the first months of the campaign, together with the Nebius team, we identified the most effective influencers — in terms of both engagement and business results (clicks and registrations). Several bloggers were signed for long-term cooperation (6 months), involving multiple posts per month and deeper product integration of Nebius across various formats.
ResultsKey Metrics
The campaign delivered strong performance across all key indicators, significantly exceeding client expectations in several areas.
Across key social networks (TikTok, Instagram Reels, Twitter, Reddit)
On TikTok and Reels — significantly higher than the industry average of 1.5–2%
On TikTok/Instagram Reels — 37% lower than the initial forecast
To AI Studio and Explorer Tier landing pages, exceeding the 2.0% target benchmark
For the AI Studio direction, thanks to organic reach and repeated influencer posts
Twice as high as Nebius' typical landing page CR
Decreased by 44% compared to previous experimental launches without influencers
Relaunches and Scaling: several influencers signed 6-month contracts — Nebius and the agency continued regular collaboration with bloggers who showed the best engagement and conversion metrics. Successful video formats and scripts were scaled in the second wave of launches (Q2 2025), including in new geographic regions.
InsightsConclusions and Insights
Why It Worked:
- Clear product positioning allowed us to quickly communicate Nebius’s value to different audiences — from AI developers and researchers to DevOps engineers and members of the academic community.
- The short video format on TikTok and Reels proved optimal for promoting the Explorer Tier — native delivery and visual explanation of benefits helped reduce the entry barrier for new users.
- A data-driven approach to optimization enabled real-time budget redistribution, scaling of effective combinations, and increased ROMI at every stage.
- Engagement of micro- and nano-influencers with loyal technical audiences ensured trust and organic reach — especially on X and LinkedIn.
- Timely adjustments to creatives and scripts, based on feedback and results from early launches, boosted the effectiveness of the main campaign wave.
What Could Have Been Done Differently
- In some cases, the overly technical presentation in the initial creatives reduced engagement — some materials had to be simplified and adapted to short-form video formats.
- Certain platforms (e.g., Reddit) showed good reach but lower engagement depth — further A/B testing of formats will be needed in future campaigns.
- Not all bloggers were ready to work with technically complex products — the selection process should begin with deeper casting, including mini-briefs and test scripts.
Lessons for Future Campaigns
- Flexibility is key: rapid adjustments, resource redistribution, and close collaboration with the client allowed us to exceed expectations.
- AI infrastructure requires not just reach but a close match between the product and influencer profile — this ensures not just clicks, but real product engagement.
- Demonstrating interfaces and technical use cases builds trust and lowers the entry barrier — especially when communicating with developer and ML engineer audiences.
