Pattern's Dave Wright: Solid Products and Proper Messaging Enable Brands to Truly Benefit from Technology

[Ebrun Original] On September 17, 2026, at the "Accelerate26 Cross-Border E-commerce Acceleration Conference · China Stop" hosted by Pattern and co-organized by Ebrun as a strategic partner, Dave Wright, Co-founder and CEO of Pattern, delivered a keynote speech titled "Technology Empowers Business, Unlocking Core Levers in Cross-Border E-commerce."

Dave Wright believes that China possesses outstanding manufacturing capabilities, product supply, and pricing advantages, but as AI and large language models make product information increasingly transparent, whether the product itself is truly excellent is becoming more important than ever. Technology cannot replace product strength; it should serve great products, helping brands better understand demand, build trust, and improve efficiency in traffic acquisition, conversion, and global execution.

He pointed out that brand growth cannot rely solely on increasing marketing spend. Companies need to evaluate organic and paid traffic together to determine whether product quality, pricing, content expression, and fulfillment experience can form long-term competitiveness. Meanwhile, leveraging machine learning, natural language processing, and large language models, businesses can transform massive consumer feedback, search terms, content attributes, and advertising data into actionable operational decisions.

"As long as the product is solid and the marketing messaging is right, brands can benefit from technology," said Dave Wright. The value of technology lies not only in providing analysis but also in continuously adjusting ads, content, pricing, and fulfillment at machine speed, enabling brands to achieve scalable growth in the complex global e-commerce market.

This article is compiled based on the guest's on-site speech, with minor edits that do not affect the original meaning.

Below is the full transcript of the speech:

01 Technology First Serves Great Products

I'm delighted to be here with all of you. We pay close attention to China because it is a huge market and home to many globally leading products. These products are made in China and sold worldwide.

My background is in technology. People in tech focus on numbers and data, which is what I do every day: I closely monitor how Chinese-made products perform and what Chinese companies are doing. I believe China has excellent companies and products, and there will be even greater success in the future.

Especially with the rapid development of AI and large language models, product experiences are becoming increasingly transparent. Consumers can directly access deeper information about products and understand the true value they offer. Companies must therefore think about how to accurately capture demand in the intelligent era.

Compared to brand buzz, we focus more on whether the product itself is outstanding, because ultimately, better products win the market. China has unique advantages in product supply, manufacturing scale, and pricing, but for these advantages to truly translate into growth, brands must also answer several questions: Is the price competitive? Can the customer experience win? Will technology and marketing investments yield corresponding returns?

I used to dislike marketing and oppose exaggerated marketing. When I first encountered e-commerce advertising, I saw a brand discussing advertising return on investment in spreadsheets. At that time, Amazon reported that its return rate was about five times the average, but when the company increased spending, the business did not improve; it lost even more money.

We quickly conducted a correlation analysis on the data, and the results made me question: Is advertising really effective? This experience made me realize that brands cannot just look at superficial advertising return figures or simply hand growth problems over to marketing. Brands need to use mathematics and technology to identify the variables that truly impact business results.

This also became an important starting point for founding Pattern. We want to help great products succeed rather than using technology to mask product problems. Technology support and marketing support are both indispensable, but the core premise remains the product itself.

For Chinese brands looking to enter the U.S. market, our data shows that product quality must first pass the test. Consumers read reviews, compare products, and if quality issues are exposed, technology can do nothing.

The second factor is pricing. Price is not about being as low as possible; it must match product quality and value. Brands do not need to rely solely on low prices to attract traffic; it is more important to find the right audience and price range.

The third factor is content and trust. Even if a brand has a good product and price, if the brand name, textual description, and visual content do not allow consumers to quickly understand, users may still skip it. Especially after platforms have aggressively cracked down on review fraud, the trust threshold for unfamiliar brands has further increased. Brands need to explain products with clear content and may need to reposition brand expression for target markets.

We once re-packaged the content for a sleep aid device brand. The product itself remained completely unchanged, but the proportion of consumers who felt the content reflected "high quality" rose from 15% to 84%, and the proportion who found the product "trustworthy" increased from 13% to 86%. This shows that good products still need accurate content to be seen and understood.

We also worked with a well-known nutritional supplement brand. Data analysis showed that its actual conversion rate was more than 6 percentage points lower than what the product page displayed. The brand's growth largely came from high marketing spend: the average marketing spend in the relevant category was about 13%, while it invested 19%. This growth model is not sustainable.

Many Chinese brands ask: "Our products are excellent, even better than competitors, and prices are lower. Why haven't we succeeded?" The problem often lies here: product strength is a prerequisite, but brands also need to build trust so consumers can quickly understand product value. Only when quality, price, and information expression are addressed can technology have a foundation to work on.

02 Growth Cannot Rely Solely on Increased Ad Spend

Global e-commerce is a very complex issue. Brands must not only reach consumers worldwide but also handle platform operations, content, advertising, logistics, pricing, and customer experience. Companies must ask: Can we reach more users at a lower cost? Can we speed up delivery? Can we improve returns when profit margins are already limited?

Team size is also a bottleneck for brand growth. Large brands may deploy hundreds of people on platforms like Amazon, while small brands' e-commerce teams may have fewer than 10 people. Without scalability and execution efficiency, companies struggle to maximize the value of each new product launch.

Among the factors affecting e-commerce growth, the first is traffic. Brands can find influencers to create content on demand, which may be a shortcut to traffic, but they still need to break it down: Can the conversion from traffic cover the cost? How should different marketing channels be integrated?

We hope to use machine learning, natural language processing, and artificial intelligence to help brands convert traffic into actual revenue, rather than relying only on a few influencers making videos. The second key factor is conversion rate. Whether traffic is effective ultimately comes back to conversion.

Brands need a clear understanding of advertising. To achieve growth, organic and paid traffic must be evaluated together: Is product information accurate? Are there long-term winning factors? How much does organic traffic contribute, and what traffic still needs advertising to supplement? Is current performance better than three months ago?

If product information and conversion have not improved and you simply increase ad budgets, you typically won't get better results. Companies need to first confirm whether the messaging is appropriate, whether the product fits the market, and whether this growth approach can hold up long-term.

Traditional advertising makes it difficult to precisely answer how each investment affects sales, but in the e-commerce environment, we can calculate using Return on Ad Spend (ROAS). Ad campaigns need to consider two parts simultaneously: the product itself and the conversion logic behind the product. With models, we can extract multi-dimensional features from user and product information and use counterfactual reasoning to predict click changes, determining which ads truly generate incremental value.

This is the biggest difference between technology and traditional advertising. The focus is not simply on compressing time or budget but on making every ad dollar more valuable. Companies can review traffic trends, observe advertising patterns, and continuously adjust campaigns based on product performance.

We took a smartphone brand as an example. Before launching a new generation, the brand might invest millions or even hundreds of millions of dollars in advertising to secure higher rankings across multiple e-commerce platforms. We thought about whether we could use large language models and machine learning to improve its traffic monetization approach.

To do this, we conducted a comprehensive analysis of the brand, continuously adjusted keywords and search terms, tested different content and campaign combinations, and ultimately found a more effective approach on Pattern's technology platform. What technology truly adds is not advertising buzz, but the ability to identify opportunities, validate results, and execute quickly.

This means brands cannot separate advertising from business operations. Traffic, conversion, pricing, fulfillment, and content collectively determine outcomes. Brands need to calculate every step, know where growth comes from, and recognize which growth is just short-term numbers bought with higher spending.

03 AI Must Move from Insight to Real-Time Execution

The value of technology is not just helping companies do a one-time analysis; more importantly, it is about continuously detecting changes and taking action. Our technology can handle about 20 million bid changes per day; humans cannot keep up with machine speed.

In a high-frequency changing trading environment, new keywords and phrases keep emerging. Once a keyword or keyword combination gains traffic and the product performs well, the system can continue investing; if performance declines, it can adjust or shut off budget. If a brand turned off ads for a product six months ago but hasn't kept tracking new search trends, it might miss new opportunities.

Content can also be analyzed through data. For example, with a creatine product from Thorne, we extracted a dataset containing 270,000 pieces of content, and the computational cost for a single analysis was about $1,700. On that basis, we summarized the different factors driving consumers to buy creatine and mapped them against sales share.

For instance, 48.8% of sales in this category went to products with the "gluten-free" attribute, while only 28.5% of products actually had that attribute. Other high-ranking attributes included improved athletic performance, high purity, and muscle gain. Such analysis helps brands discover which values consumers already recognize, which values are absent from content, and which opportunities require trade-offs.

Data provides evidence, but the final decision must still be made by the brand. For example, we once suggested that Thorne keep the attribute "helps muscle recovery" on two products, but its scientific team felt the attribute should strictly belong to only one product. Similarly, data showed that before-and-after comparison images performed well for conversion in the creatine category, but considering regulatory risks, the brand decided not to use that visual format. Technology cannot erase brand boundaries; it should instead make choices and their costs clearer.

Search term data can also help brands assess new product opportunities. "Women's creatine" has a certain search volume, but some more specific search terms still account for a low share of overall orders; "women's creatine gummies" has search demand but generates no orders because the brand hasn't yet offered a gummy format. Companies can use this to evaluate whether to launch new products while avoiding overinvesting in overly niche demand.

Consumer feedback can also be incorporated into decision-making through sentiment analysis. We found that competitor negative reviews regarding stomach acid and gastrointestinal discomfort were mentioned about 3% of the time, while for Thorne it was about 7%, which warrants further investigation. With the rise of large language models, such information will become even more transparent, and the result could be a further shift of consumers toward fewer, better products.

In the future, the application of AI-generated content will continue to increase. Currently, many brands remain cautious about this, especially opposing AI-generated human images, and we respect brand choices. But as platforms push dynamic content, when users search for different keywords, the image combinations on the page may also change in real time. In the past, mathematical methods were used to find the image with the best overall conversion; in the future, content might be dynamically matched based on different search intents.

Image generation also needs to solve product authenticity issues. General-purpose AI does not always accurately understand the spatial relationships and different angles of products. To address this, we use image data collection and LoRA training to build adapted models for specific products, generating more accurate product images under different lighting conditions; we can also complete collection of about 80 images in a very short time, which can then be used for e-commerce content creation.

We have also integrated years of technical capabilities into Pi (Pattern Intelligence). The platform is built on layers of execution, price changes, and data, with relevant information accessible and usable in real time. Its underlying infrastructure connects to more than 70 global markets, built on 91 trillion data points, and converts analysis into concrete actions through an execution engine.

From my perspective, brands must further understand themselves: what level of organic traffic can be achieved, what advertising truly brings, how to balance price and profit, what can be done internally, and where technology is needed to fill gaps.

We are looking for growth, and we hope to start a new growth journey with truly exceptional brands. Technology is built for the best products. The core premise remains the product itself and whether it delivers genuine value to consumers. On this basis, technology can help brands connect traffic, conversion, content, pricing, and fulfillment, allowing them to go further and more steadily in the global market.

This article was first published on the official website of Ebrun.