[Interview] Yuwai's Tanaka and Anagram's Nishimura Discuss the Present and Future of Shopify Ad Management

As 2025 draws to a close, the world of advertising in commerce is changing at an unprecedented pace. Generative AI is becoming practical, platforms are updating their specifications, and measuring data in a cookieless era presents new challenges.

For this discussion, we welcomed Yuwai's Tanaka, an expert in performance advertising for retail and commerce, and Anagram's Nishimura, a specialist in vertical video and social media advertising, to take a thorough look back at 2025 and discuss ad strategies heading into 2026.

From behind the scenes at Google Merchant Center to the current state of TikTok Shop, the art of creator assignment in third-party ad delivery, and the technical pitfalls Shopify merchants face — we bring you a rich collection of practical insights that can only be shared from those on the front lines.

Left: Hiroki Tanaka, Representative Director, Yuwai Inc.   Right: Nishimura, Anagram Inc.

Ad Management in the AI Era: "Conversion Data" Is the Starting Point for Everything

Kurose: Thank you both for joining us today. I'd like to start by looking back on 2025. This year felt like one where "AI-driven automation" truly took hold in the world of advertising. In this environment, what should advertisers have been doing to adapt?

Tanaka: Absolutely. AI has already deeply embedded itself within performance advertising systems. The key question is: how well can humans supply the data needed to drive that AI correctly?

For most advertisers, the goal of advertising is conversions — purchases, sign-ups, and so on. Google and Meta's automated bidding works by having AI predict who is most likely to convert and serving them ads accordingly. In other words, if accurate conversion data isn't being fed into the AI, automated bidding simply won't work properly.

But today, iOS's ITP (Intelligent Tracking Prevention) and regulations like GDPR have made cookie-based measurement extremely difficult. There's a real dilemma: AI needs massive amounts of data, yet the data we can collect is shrinking.

So before making incremental creative improvements or tweaking campaign settings, the top priority in ad management right now is building an environment where you can measure conversion data without gaps — for example, through server-side tracking.

Kurose: I see. Does the same apply to other platforms like TikTok?

Nishimura: Yes, the same trend applies to TikTok and Meta as well. The direction these platforms are heading seems to be: rather than having humans fine-tune targeting settings, just input the creative, budget, and goal — and let AI optimize delivery from there.

Accounts that have accumulated sufficient conversion data tend to benefit greatly from AI, while those that haven't struggle because the machine learning can't gain momentum. This polarization is visible in TikTok ads and Meta ads alike.

Kurose: In Japan, TikTok Shop is gaining traction. Does purchase data from TikTok Shop also feed into optimization for regular TikTok ads?

Nishimura: We haven't been able to confirm that at this point. As a premise, the promotional tools within TikTok Shop and the ad formats that drive traffic to external sites like Shopify are clearly separated, and from our hands-on experience, we don't currently get the sense that TikTok Shop purchase data is influencing the targeting of regular ads.

Google Search's AI Transformation and How to Win at the "Shopping Graph"

Kurose: Google Search has also been transformed dramatically with the introduction of AI Overviews, completely changing the landscape of search results. The SEO industry has been shaken — but how should commerce businesses respond?

Tanaka: The features showing "popular products" or "recommended shops" in search results are drawing from Google Shopping's database, known as the "Shopping Graph."

The essential tool for feeding accurate information into this Shopping Graph is Google Merchant Center. In other words, in the age of AI-powered search, continuously supplying Google with accurate, rich product data through Merchant Center has become more important than ever for getting your products to appear.

Kurose: Specifically, what elements of the product feed should be prioritized?

Tanaka: The key elements Google focuses on are primarily the following:

1. Product identification information: JAN codes, brand names, model numbers (MPN), and so on. Without these, Google cannot correctly recognize your products.
2. High-resolution images: In addition to the main image, register sub-images up to the maximum of 10 (11 total). A size of 1024px or more is recommended.
3. Accurate titles: Appropriate product names that match search queries.
4. Google product category: Correct category mapping.

Beyond these, the required fields vary by industry. For apparel, for example, spec information such as size, color, and material is also essential.

Kurose: When it comes to images, what kinds tend to perform best?

Tanaka: Recently, Google has been offering a generative AI feature for apparel called "Virtual Try-on," where users can upload a photo of themselves and see the product composited onto them so they can visualize how it would look.

To support this feature, there are requirements such as having flat-lay images and ensuring that a mannequin's hands aren't overlapping the clothing. Amazon's product image guidelines are also a useful reference, but the key to improving click-through rates is providing images where people can easily imagine themselves wearing or using the item.

The Japan-US Gap in TikTok Shop and TikTok-Native Product Development

Kurose: On TikTok Shop — the integration with Shopify is getting stronger, but how do you see the current state in Japan?

Nishimura: According to publicly available data, Japan's gross merchandise value (GMV) for the three months from July to September 2025 is said to be around 4 billion yen. Meanwhile, in the US, which launched earlier, it generated around 44 billion yen in GMV in just the first month after launch. Even accounting for differences in market size, Japan has not yet seen a rocket-start.

Kurose: Why did it spread so explosively in the US?

Nishimura: I think there are two main reasons.

The first is the penetration of TikTok itself as a platform. Political debates aside, American users are extremely active, and buying things on TikTok has become deeply embedded in their daily lives.

The second is the existence of brands and products optimized specifically for TikTok. In the US, many D2C brands have emerged that were designed from the product planning stage with TikTok Shop in mind — calculating how to sell there and how to make things look good on video. For example, they create compact packaging with no shipping cost to make gifting to creators easy, and design products with clear before-and-after transformations that look great on video.

In Japan, most cases are still just listing existing products on TikTok Shop, whereas in the US, brands built to fit the platform's context are leading the market — and that's the major difference.

Leveraging Third-Party Ad Delivery and the Art of Creator Assignment

Kurose: In social media advertising, we're seeing more cases where creator posts are used as ad creatives, rather than just content from the brand's official account. Meta's Partnership Ads, TikTok's Spark Ads — so-called third-party ad delivery. What are the key points for making this work?

Nishimura: Absolutely. Right now, platforms themselves strongly recommend running ads using creator content. Traditional influencer campaigns often ended with a single post and were hard to measure, but with third-party delivery, you can freely set the destination link, use targeted delivery, and measure conversions — making it possible to generate sales while keeping results visible.

Kurose: What's the key to making it successful?

Nishimura: The biggest factor is creator assignment. A common mistake brands make is choosing creators based on follower count — but follower count doesn't matter in third-party delivery.

Today's social media algorithms determine reach based on content quality (engagement), not follower count. You should choose creators who have a strong affinity with your product and achieve high engagement from both followers and non-followers alike, even if their follower numbers are small.

Kurose: What about content production after the creator is assigned?

Nishimura: I believe the golden rule is for brands not to over-control the content's direction and expression. When companies give detailed instructions about the content, it tends to come across as overly promotional, and users pick up on that and skip it. Respecting each creator's worldview and tone — even if the result is a bit rough around the edges — and letting them make a video that feels like them: that authenticity is what blends into the timeline and makes users stop scrolling. That said, it's also important for the brand to define the product's key features and the elements that must be communicated, and to convey those to the creator. The key to success is striking the right balance between what the brand needs to nail down and what's left to the creator's discretion.

Tanaka: I completely agree. TikTok and Instagram are places where users passively enjoy content, so when corporate advertising enters as a foreign element, it becomes noise. By riding the creator's narrative, you can soften that noise and deliver your product in a natural way.

Ad Management Pitfalls That Shopify Merchants Fall Into

Kurose: At StoreHero, we support Shopify merchants. Could you share any ad management pitfalls that are specific to Shopify?

Tanaka: This year, there are three pitfalls I felt particularly strongly about.

① Migration Issues with Checkout Extensibility

The migration to Shopify's new checkout environment (Checkout Extensibility) means the checkout page has become a secure environment (sandboxed), and the traditional method of embedding GTM tags or ad tags directly no longer works.

Major platforms like Google and Meta can still measure through their dedicated sales channel apps, but the problem lies with domestic Japanese media such as Yahoo! Ads and LINE Ads. These often don't have official Shopify integration apps, meaning if you do nothing, conversion tracking will stop working entirely.

To address this, you need to implement a measurement solution like StoreHero's, or set up server-side tracking. The rollout is planned to expand to plans beyond Plus in the summer of 2026, so urgent action is needed.

② Data Source Proliferation Caused by the Migration to Merchant API

This is a spec change on Google's side, but the mechanism for connecting Shopify with Google Merchant Center is migrating from the traditional Content API to the Merchant API.

This migration often happens automatically, but we've confirmed a phenomenon where changing market settings or language settings on the Shopify side causes a large number of unintended data sources (feeds) to be generated in Merchant Center.

If you had previously customized product data using supplemental feeds or feed rules, when the data source changes, those settings get dropped — leading to incidents like "the product titles we had optimized had reverted to the originals without us noticing" or "sale prices aren't being reflected." Rather than leaving everything to apps, it's become essential to have an operational process that regularly checks the contents of Merchant Center.

③ Disapprovals Due to Mismatches Between Site Prices and Feed Prices

This is also a Shopify-specific pitfall that occurs when you're using Shopify apps to dynamically rewrite prices or shipping costs displayed on the site. The API sends regular-price data to Google Merchant Center, but when Google's crawler actually visits the site, it may see member-only pricing or dynamic shipping costs applied by an app.

When this happens, Google determines that the price in the feed doesn't match the price on the landing page and disapproves the product. In the worst case, there is also a risk of account suspension. Especially if you're using subscription apps, member-only pricing apps, or shipping customization apps, you need to verify what price the crawler is seeing and take steps to ensure consistency.

Outlook for 2026: The Backlash Against AI Fatigue and a Return to Human Warmth

Kurose: Finally, could you share your outlook for 2026?

Nishimura: In the video space, videos produced with generative AI like Sora will flood advertising and content. But as a counterreaction, I expect something like "AI fatigue" to set in among users. As everyone becomes able to easily produce polished videos, content will become saturated, and users will quickly learn to spot "this was made by AI."

As a backlash to that, the value of content with human warmth — real voices from real people, a handcrafted feel, honest reviews — will rise in relative terms. The words of a brand's own operators, or highly passionate reviews from customers: information with a human temperature that AI cannot replicate will become more decisive in driving purchases than ever before.

Tanaka: I agree completely. No matter how beautiful a banner or video you create with generative AI, there's always some lingering sense of something off, or a lack of sizzle. I was talking about this elsewhere earlier today, but generative AI can't eat sushi, so it can't truly express what sushi tastes like (laughs). The look on someone's face when they take a bite, the genuine emotion — only a human can convey that. And as we saw when Coca-Cola's AI-generated holiday campaign video went viral for the wrong reasons — scenes where the tire count kept changing — large companies using AI also face risks around copyright, trademarks, and likeness rights.

In 2026, I think we'll see a return to the essence of creativity — using AI for efficiency, but ultimately adding the human touch and human emotion at the end.

Kurose: It's very interesting that the more technology advances, the more paradoxically human qualities gain value.

Tanaka-san, Nishimura-san, thank you so much for your valuable insights today!