Multiply raises $9.5M for AI systems that continuously optimize B2B ads
The startup ties sales calls and CRM outcomes to ad testing, reflecting the shift toward RevOps-linked media optimization and tighter GTM feedback loops.
Multiply has raised $9.5 million and emerged from stealth with an AI-driven approach it calls “self-learning advertising” for B2B marketers. The company is focusing first on Google Search and LinkedIn, using connected go-to-market data to generate and refine creative, targeting, and messaging.
The pitch maps to a real B2B pain point: performance declines when campaigns go stale, but keeping ads aligned to buyer reality is hard when sales insights, CRM data, and paid media workflows live in different places.
Short on time?
Here’s a quick look at what’s inside:
- What Multiply means by “self-learning” ads for B2B
- How the system connects sales calls, CRM data, and ad performance
- Competitive context vs Metadata, Directive, HockeyStack, and MadKudu
- Why this aligns with marketing and sales convergence
- What B2B marketers should evaluate before adopting
What Multiply means by “self-learning” ads for B2B
Multiply describes its product as an AI-native media agency model where ads improve continuously using a company’s internal data. Rather than relying on periodic creative refresh cycles, Multiply’s approach is built around ongoing structured experimentation across audiences, copy, and creative so performance improves week over week.
The company says it supports multiple “agents” across the workflow, including systems focused on extracting customer language from sales calls, refining ICP targeting from closed-won data, tuning search quality score inputs, refreshing creative, and running large volumes of A/B tests.
Multiply also emphasizes human oversight for brand safety and compliance, which is a practical requirement in B2B categories where claims, regulated language, and competitive positioning can create risk.
Patrecia Meliana
How the system connects sales calls, CRM data, and ad performance
The most distinctive element is the data loop: Multiply connects directly to sales call recordings, CRMs, and ad platform performance data to generate new messaging based on why buyers choose a product.
If it works as described, this can reduce a common lag in B2B growth: sales learns what language converts, but marketing updates ads weeks later after multiple handoffs. A tighter loop matters most in competitive categories where differentiation is subtle and objections evolve quickly.
This model also implies new operational requirements. To get value, companies need usable CRM hygiene, consistent lifecycle stage definitions, and permissioned access to call data, plus a governance model for what the system is allowed to generate and test.
Competitive context vs Metadata, Directive, HockeyStack, and MadKudu
Multiply sits in the B2B performance advertising segment where vendors blend software with execution. It overlaps indirectly with platforms like Metadata (campaign automation and experimentation), agencies like Directive (services-led performance execution), and analytics-focused tools such as HockeyStack and MadKudu that help teams connect marketing activity to revenue outcomes.
The differentiation Multiply is aiming for is “creative and messaging generation from first-party GTM truth,” especially sales calls and pipeline outcomes, combined with continuous experimentation. In contrast, many stacks either (a) optimize distribution and targeting but do not deeply operationalize sales conversation data, or (b) provide measurement without actually executing the iteration loop.
That said, the competitive bar is rising quickly as more tools add AI-assisted creative, audience automation, and measurement connectors. Multiply’s defensibility will likely depend on how reliably it can translate noisy qualitative inputs (calls) into compliant, testable ad variants, and whether it can prove lift beyond what strong in-house ops or a traditional agency can deliver.
Why this aligns with marketing and sales convergence
Multiply’s model reinforces a broader trend: B2B paid media is becoming an extension of revenue operations, not a standalone channel function. When ads are tuned using closed-won analysis and call-derived language, marketing and sales alignment becomes a data integration problem as much as a collaboration problem.
It also reflects the shift toward AI marketing automation that is workflow-native, meaning it plugs into existing systems (CRM, call recording, ad platforms) and produces continuous iteration rather than one-off “AI insights.”
What B2B marketers should evaluate before adopting
Before betting on a self-learning ads approach, performance teams should pressure-test fundamentals:
- Data readiness: Are call recordings accessible and labeled? Are CRM fields dependable enough to drive targeting and learning?
- Control and governance: Who approves new claims, competitive language, and visual direction, and how is that enforced in the workflow?
- Experiment design: How are tests structured to avoid false positives, overlapping variables, and budget dilution?
- Attribution expectations: What outcomes are promised and how are they measured (pipeline, meetings, CAC), especially across Google Search vs LinkedIn?
Multiply cited early customer results such as Vanta reporting 770% more sales meetings, and it said pre-launch partners generated more than $25 million in sales pipeline with reported 300 to 500% performance improvements. Marketers should treat these as directional signals and ask for methodology, baselines, and time windows before using them for forecasting.
