AI Performance Measured in Metrics! Appier's Chih-Han Yu: Integrating AI into OKRs, Per-Employee Quarterly Gross Profit Up 38% YoY

The adoption of AI in enterprises has evolved from 'Can employees use it?' to 'How much work output does AI actually generate?' Appier (TSE: 4180), an AI software company, co-founder and CEO Chih-Han Yu revealed today (13) during the earnings call that the company has incorporated AI usage outcomes into employee performance evaluations and OKRs, even establishing a performance assessment concept of '50% AI results, 50% core business.' With Agentic AI fully integrated into R&D and daily workflows, Appier achieved a per-employee quarterly gross profit of ¥10.25 million in Q2, up 38% year-on-year, marking a historical high.

Founded in 2012, Appier is an AI software company headquartered in Taipei, primarily using AdTech and MarTech technologies to help enterprises conduct AI-powered advertising, customer analytics, and marketing decision-making. The company is listed on the Prime Market of the Tokyo Stock Exchange under stock code 4180 and operates 17 offices across Asia-Pacific, the U.S., and EMEA.

Unlike many companies that still view generative AI as an auxiliary tool, Appier has advanced to formally incorporate AI usage outcomes into performance management. Yu stated that the company tracks AI tool usage across different workflows and the resulting functional developments, model improvements, and actual work outputs through internal systems.

Does Appier meticulously calculate how many tokens each engineer uses or how much AI computing power is consumed? Yu clarified that the company 'doesn’t go down to the token level,' instead evaluating overall input and output—such as computing resources consumed, features developed, and model improvements—continuously monitored via online systems.

In other words, Appier measures not 'who issued the most prompts,' but whether AI has genuinely increased work output.

The integration of AI into OKRs extends beyond engineers—Yu himself actively uses it. Yu pointed out that Agentic AI is not limited to R&D; the company is advancing toward 'company-wide AI adoption,' with employees across various functions, including himself, extensively using AI tools to improve work efficiency.

He also revealed that Appier now incorporates AI-driven outcomes into performance evaluations and OKR design, forming a '50% AI, 50% core business' evaluation concept, aiming to transform employees from mere AI tool users into producers of measurable productivity.

These efforts are already reflected in financial metrics. In Q2, Appier’s overall per-employee quarterly gross profit reached ¥10.25 million, up 38% year-on-year, a significant increase from the ¥7 million range last year. Yu clarified that this metric is calculated as 'total company gross profit divided by number of employees,' not limited to R&D or engineering output.

Agentic Engineering in R&D: Gross Margin Breaks 60% for the First Time

One of the earliest and most direct areas where Appier adopted AI is R&D. Yu stated that the company has introduced Agentic Engineering into its R&D process, enabling AI agents to act as R&D assistants for engineers, supporting feature development, model testing, and algorithm improvement, shortening development cycles previously done manually.

This process has become a key factor behind Appier’s improved gross margin in Q2. The company’s gross margin reached 60.1% in Q2, breaking the 60% threshold for the first time, and 61.3% under constant currency. Core business gross profit grew 43.5% year-on-year, outpacing the overall gross profit growth of 33.5%.

Yu noted that while increased use of Agentic AI in R&D does raise some AI computing and development costs, the acceleration in model improvement and client ROI is even greater, resulting in a net positive impact.

The underlying logic is that AI shortens the R&D cycle, allowing Appier to improve model accuracy and product performance faster. Clients, achieving better advertising or marketing ROI, are willing to allocate more budget to Appier’s platform, further expanding revenue, data scale, and gross profit, which in turn allows more resources to be reinvested into AI model development.

The company calls this cycle the 'AI flywheel'—AI efficiency gains drive model improvements, higher client ROI expands budget and data advantages, which then feed back into strengthening AI performance.

AI is Also Changing Software Companies’ 'Build or Buy' Decisions

The impact of Agentic AI extends beyond labor efficiency, beginning to reshape Appier’s M&A strategy. Yu pointed out that in the past, a software company lacking specific technology or product capabilities would typically need to acquire them through M&A. But now, with AI, certain functions can be developed faster internally, 'so it’s not necessarily required to merge with another company.'

Yu emphasized that Appier has historically 'acquired products, not revenue through M&A,' and any future acquisitions will still prioritize targets that highly integrate with core business and generate product synergies.

Appier stated that future capital allocation will continue to focus on four areas: vertical AI and Agentic engineering R&D, expansion into key markets and enterprise clients, selective M&A with core business integration synergies, and gradually increasing shareholder returns through dividends or share buybacks as core free cash flow improves.

Yu believes the next stage of AI Agent value isn’t simply 'replacing employees,' but enabling one employee to manage more digital agents. Humans can act as supervisors, delegating parts of development and execution to AI, even allowing tasks to continue during employee downtime.

During the meeting, he described, 'Humans need to sleep, but AI doesn’t.' As employees shift from mere executors to AI Agent managers, the production unit of software companies may gradually evolve from 'one employee' to 'one employee plus a group of AI Agents.' Appier’s 38% year-on-year increase in per-employee quarterly gross profit in Q2 is the company’s first step in validating this productivity transformation through financial results.