A survey-based investigation examining 251 marketing professionals across five industrial sectors in China has revealed that adopting low-code artificial intelligence tools significantly improves corporate marketing innovation performance. The empirical study, conducted by researchers in collaboration with Alibaba Cloud, establishes that low-code artificial intelligence platforms support both operational workflow enhancements and commercial offering developments. By analyzing firms utilizing the AliYi AI Copilot platform, the investigators traced how combining modular algorithmic components with natural language copilots allows non-technical marketing personnel to bypass traditional programming bottlenecks. This technological democratization cultivates dynamic organizational capabilities, specifically marketing ambidexterity, which enables enterprises to balance the exploitation of existing operational assets against the proactive exploration of emerging consumer opportunities.
In dynamic digital environments, enterprises face constant commercial pressure to anticipate shifting consumer preferences and execute marketing innovations rapidly. However, conventional artificial intelligence deployments frequently suffer from steep technical learning curves, high computational infrastructure costs, and persistent interdepartmental friction between technical and commercial divisions. Marketers often lack programming and data science expertise, while technical teams may prioritize system perfection over rapid market responsiveness. Low-code artificial intelligence addresses these operational constraints by embedding pre-built machine learning algorithms and drag-and-drop interfaces into generative artificial intelligence assistants powered by large language models, allowing professionals to construct predictive models, launch customer chatbots, and automate sentiment analysis using natural language commands.
To evaluate the structural mechanisms connecting low-code artificial intelligence with innovation outcomes, the research team grounded their analysis in dynamic capability theory and the organizational ambidexterity framework. Within marketing functions, ambidexterity requires the simultaneous pursuit of exploitative activities, which center on efficiency, refinement, and incremental optimization of current market assets, and explorative activities, which focus on search, risk-taking, and discovering novel market trajectories. The authors hypothesized that low-code platforms simultaneously reinforce both exploitative and explorative capabilities by reducing technical overhead and fostering cross-functional alignment.
The study surveyed enterprise clients actively utilizing the AliYi AI Copilot platform, which integrates Alibaba’s Qwen large language model to interpret text and voice instructions into functional code workflows and digital system integrations. Sampling criteria restricted participation to organizations operating within China that had utilized low-code artificial intelligence tools within the preceding three years. The authors distributed 400 survey questionnaires via email to platform-registered enterprises, collecting 275 total responses for a response rate of 62.8 percent. After eliminating incomplete submissions, the researchers retained a final validated sample of 251 marketing professionals, encompassing roles such as marketing managers, brand operators, product managers, and market analysts across diverse commercial domains.
The demographic and operational profile of the sample captured substantial industrial heterogeneity across automotive, aviation, software, retail, and financial services. High-tech manufacturing firms, categorized according to the official classification standards of the National Bureau of Statistics of China, accounted for 27 percent of participating organizations. In terms of organizational scale, less than 1 percent of enterprises employed fewer than 50 workers, while mid-sized companies employing between 101 and 500 individuals formed the median tier, comprising 35 percent of the total sample. Furthermore, 89 percent of surveyed enterprises had maintained commercial operations for more than five years, approximately 50 percent had utilized low-code artificial intelligence systems for more than six months, and 23 percent of individual respondents reported that their primary job scope explicitly required technical information technology skills.
Survey constructs were adapted from established academic literature using five-point Likert scales, followed by extensive back-translation between English and Chinese and pretesting on a focus panel of 33 senior users. To control for exogenous market environments, the investigators incorporated industry-level concentration indicators, including the four-firm concentration ratio and the Herfindahl-Hirschman index, derived from the CSMAR database using 2023 financial reports. Firm-level control variables included organizational age calculated relative to 2024, high-tech sector status, organizational size categories ranging from under 50 to over 2,000 workers, and cumulative low-code usage tenure. An individual-level control tracked whether a respondent’s job duties formally necessitated technical programming expertise.
To evaluate potential systematic survey errors, the researchers conducted tests for common method bias and non-response bias. Harman’s single-factor test using confirmatory factor analysis showed that the first extraction factor accounted for only 29 percent of the total variance, safely below the standard 40 percent critical threshold. Collinearity assessments confirmed that all variance inflation factor scores between construct pairings remained beneath 3.3, indicating that common method variance did not distort the findings. Non-response bias was tested by dividing respondents into early and late response cohorts using the average completion duration of 263 seconds as a split threshold, with independent t-tests showing no statistically significant discrepancies between the 171 rapid respondents and the 80 delayed respondents.
Hypothesis testing was executed using SmartPLS software applying Partial Least Squares Structural Equation Modeling, a variance-based statistical technique suitable for exploratory models that do not require multivariate normal distributions. Bootstrapping resampling procedures confirmed that the structural model accounted for significant variance across endogenous constructs, with explained variance coefficients reaching 27.6 percent for exploitative capabilities, 17.6 percent for explorative capabilities, 16.7 percent for process innovation performance, and 9.0 percent for product innovation performance. Reliability diagnostics demonstrated acceptable internal consistency, with construct item loadings exceeding 0.7, Cronbach’s alpha and composite reliability values exceeding 0.6, and average variance extracted scores exceeding 0.5 across all scales.
The structural equation model confirmed that low-code artificial intelligence adoption exerted a direct, positive, and statistically significant effect on both marketing explorative capabilities and exploitative capabilities. When evaluating innovation dimensions, the researchers found divergent mechanistic pathways. Both explorative capabilities and exploitative capabilities generated statistically significant positive effects on marketing process innovation, supporting the role of ambidexterity in optimizing operational workflows and resource management. Conversely, marketing product innovation performance was significantly driven solely by exploitative capabilities; the direct path connecting explorative capabilities to product innovation was not statistically supported, demonstrating that radical exploratory search alone did not translate into immediate commercial product gains.
Investigation of indirect pathways revealed that low-code artificial intelligence significantly improved overall marketing innovation performance through ambidexterity, although mediation for product innovation operated exclusively through exploitation mechanisms. Incremental product adjustments aligned with existing consumer ecosystems proved far more effective in translating artificial intelligence adoption into commercial product performance than speculative, open-ended experimentation. Furthermore, moderation modeling established that an innovation-oriented organizational culture significantly amplified the positive influence of low-code artificial intelligence on both exploitative and explorative capabilities. In corporate environments that actively reward experimental mindsets, cross-functional knowledge sharing, and iterative trial-and-error, marketers integrated automated tools into daily workflows with substantially greater speed and organizational efficacy.
To verify that the empirical findings were not artifacts of statistical specification or endogeneity, the authors executed secondary robustness evaluations. A moderated mediation analysis utilizing Hayes PROCESS Model 8 replicated the directionality and statistical significance of all structural paths while controlling for the full covariate set. To address potential endogeneity stemming from unobserved worker technical aptitude, the researchers estimated an instrumental variable two-stage least squares regression. They employed the historical frequency of corporate IT skills training conducted prior to low-code implementation as an instrument, reporting a sample mean training score of 3.83 and standard deviation of 0.90. First-stage regression F-statistics exceeded 10 and Kleibergen-Paap identification tests were significant, confirming instrument validity and demonstrating that the primary findings remained stable when controlling for potential skill confounds.
The authors contextualized their findings by identifying operational and structural limitations that warrant future scientific inquiry. The survey sample was primarily drawn from Chinese enterprises operating in non-high-tech sectors, meaning that firms with mature internal artificial intelligence engineering units may experience different organizational dynamics. Additionally, third-party low-code deployment introduces corporate risks regarding proprietary data security, customer information protection, and vendor lock-in dependencies on commercial cloud platforms. The authors concluded that while cultivating an experimental internal culture allows firms to maximize the strategic value of low-code intelligence tools, sustainable long-term governance requires robust data protection architectures and regulatory safeguards across commercial digital ecosystems.
Subject of Research: Low-code artificial intelligence adoption and marketing ambidexterity in enterprise innovation
Article Title: Low-code AI enabling marketing innovation: The mediating role of marketing ambidexterity
Article References: Wang, X., Wang, Y., Li, N., & Zhao, Y. (2025). Low-code AI enabling marketing innovation: The mediating role of marketing ambidexterity. Journal of Digital Management, 1(1), Article 12. https://doi.org/10.1007/s44362-025-00015-4
Image Credits: AI Generated
DOI: 10.1007/s44362-025-00015-4
Keywords: low-code AI, marketing innovation, marketing ambidexterity, dynamic capabilities, process innovation, product innovation, generative AI copilot, artificial intelligence, organizational culture, Alibaba Cloud, AliYi AI Copilot, partial least squares SEM, structural equation modeling, digital marketing, enterprise software, two-stage least squares
Cite Scienmag News
Blake Davidson. (November 21, 2026). Low-Code Artificial Intelligence Enhances Marketing Innovation Through Ambidextrous Capabilities. Scienmag. https://scienmag.com/low-code-artificial-intelligence-enhances-marketing-innovation-through-ambidextrous-capabilities/
Blake Davidson. "Low-Code Artificial Intelligence Enhances Marketing Innovation Through Ambidextrous Capabilities." Scienmag, 21 November 2026, https://scienmag.com/low-code-artificial-intelligence-enhances-marketing-innovation-through-ambidextrous-capabilities/. Accessed 21 November 2026.
Blake Davidson. "Low-Code Artificial Intelligence Enhances Marketing Innovation Through Ambidextrous Capabilities." Scienmag. November 21, 2026. https://scienmag.com/low-code-artificial-intelligence-enhances-marketing-innovation-through-ambidextrous-capabilities/








