TikTok Shop today documented Product Diagnosis, a new layer inside the Product Analytics section of its US Seller Center that tells sellers which products need attention, why each product's sales moved and what to change, measuring every key number against both the seller's own previous period and the top-performing products in the same category.

In Short

TikTok Shop has added a diagnosis tool to the product analytics pages US sellers already use, which flags products with falling sales and lists up to 10 reasons behind each change. It matters to brands and sellers who sell through creators, livestreams and search on TikTok, because the tool splits each product's sales across five channels and compares them with top products in the same category. The cause of a drop is now ranked inside Seller Center, but TikTok Shop does not explain how it picks the products in its benchmark, and it says the AI assistant's output is for reference only.

A diagnosis layer on existing data

The guide, titled "New Product Diagnostics Feature Launch" and carrying the Seller Compass label, sits in TikTok Shop's US Academy under Feature Guide, Seller, Analytics and Product analytics. It is marked as updated today, 11/12/2026. The launch timing it gives is simply November 2026, and its market line says the feature is currently available in the US. No author is named, and the guide gives no day within the month on which the tool reached sellers.

"Product Analytics now answers three questions, not one:" the guide begins. The three are which product needs attention, why its gross merchandise volume (GMV) moved, and what to do next. A comparison table in the guide sets the section's earlier limitations against the new features, one of which is a split of each product's GMV into five sales scenarios.

No new data feeds the tool. The guide addresses the point directly in its FAQ, asking whether this is a different data source from what sellers see today. "No. It is the same product analytics data, organized around a diagnosis," according to TikTok Shop. "The difference is the added period comparison, benchmark and factor ranking."

Four changes make up the release. The Product Performance list gains a Need attention flag. A Product Diagnosis view for each product explains how its GMV moved, which sales scenarios were affected and which factors mattered most, with an entry point to a "Chat with AI" assistant. Every key metric now carries a benchmark, showing both the change against the seller's previous period and the gap against strong performers in the same category. And the factors themselves link out to the Seller Center pages where the underlying problem sits: listing, stock, price, ratings, the affiliate program, product opportunities, SKPP and account health.

Where it sits in Seller Center

Sellers reach the feature through Seller Center, then Product Analytics. The default landing view is Product Performance, a shop-level product list. From there a seller can select "Diagnosis" on any product row, or open the Product Diagnosis tab and search by product ID or name. Filters set on one view carry over to the other.

One default is worth noting. When no product has been selected, Product Diagnosis opens on the product with the largest GMV drop in the chosen period. The tool, in other words, starts from the worst news.

Product Performance is organised into five modules. Filters come first, with a default date range of seven days and a Compare to Previous Period option. Key metrics follow: eight default core metrics, each showing its value, its gap against the benchmark and its change against the compared date. Sellers can swap metrics in and out, and the selection is remembered. The guide does not name the eight defaults anywhere. A performance breakdown shows GMV as a pie chart by scenario and, for content-driven scenarios, the top three content items by GMV. Traffic analysis covers impressions, clicks and orders by SKU, along with click-through rate (CTR) and click-to-order rate (CTOR), viewable by page views (PV) or unique visitors (UV). The product list itself is searchable by product ID or name, with sortable columns and a Diagnosis button on every row.

The single-product view is denser. Its header shows the product image, title, ID and category, a GMV range, a New product tag where relevant, price, ratings, opportunity match, listing status, stock, SKPP status and affiliate program status. The guide does not define what the GMV range refers to. Below the header come a Summary, key metrics with a benchmark line overlaid on the trend chart, a scenario summary, traffic analysis and a performance breakdown by creator and by SKU. In the diagnosis view, traffic analysis splits Shop Tab & Search into its two parts, Shop Tab and Search; the performance breakdown is hidden for that scenario.

Two yardsticks, one of them fixed by TikTok

Every diagnosis rests on two comparisons. The first, labelled "vs Compared Date," sets the current period against the seller's own previous period of the same length. The second, "vs Benchmark," sets it against other products. "Benchmark shows the performance of top-performing products in the same category," the guide states.

Sellers have no control over that comparison group. The benchmark is "platform-defined," and "Custom benchmark dimensions are not supported at this time." A question in the FAQ about whether the group can be changed gets a similar answer, with a rationale attached: "The group is fixed so that comparisons are consistent across sellers and over time."

TikTok Shop also addresses the obvious competitive concern. "Benchmark values are aggregates. They do not expose another seller's or another product's identifiable data," according to the guide.

Top by what measure, and how many products? The guide does not say how "top-performing" is defined. No percentile, sales threshold or minimum number of products is given. Nor does it specify which level of the category tree applies, a distinction that would matter for a product sitting in a broad category such as beauty against one in a narrow subcategory. The benchmark is described as top performers on the single-product view and "strong performers" on the shop-level list, without a statement on whether the two are the same group.

Gaps are presented in one of two ways. For amounts, the tool shows an absolute difference. For rates, it shows percentage points, so a rate of 3% against a benchmark of 2% appears as plus one percentage point. The choice keeps gaps on low-base rates looking small: by PPC Land's calculation, a product converting at 2% against a 3% benchmark trails by one point in the tool's display, but by a third in relative terms.

Five ways one product sells

GMV for each product is split into five scenarios: Creator LIVE, Creator Video, Seller LIVE, Seller Video and Shop Tab & Search. An "All" view combines them. A scenario appears only if the product had activity in it during the selected period; where it did not, the tab is disabled with a message saying the product has no data in that scenario.

The split mirrors how sales on the platform actually arrive. A product can sell because an affiliate creator featured it in a livestream or a short video, because the seller's own account did the same, or because a shopper found it through the Shop Tab or search. Those routes depend on different people and different rules. Affiliate creators, for instance, carry a daily Promotion Performance Score from 0 to 5 that gates product reach and badges, while a seller's own access to the affiliate program depends on a separate shop-level score.

The scenario summary ranks the scenarios by their contribution to the GMV change. A product whose sales fell because two creators stopped posting will look different, under this structure, from one whose search conversion slipped, even if the headline decline is identical.

Two conditions before a flag appears

The Need attention flag needs two signals. "The flag appears only when both conditions are true:" the guide states. The product's real-time GMV change rate must be negative, and the offline diagnosis must also return a verdict of Need attention.

A short entry in the FAQ explains why a product with falling sales can go unflagged. "The flag requires both a negative real-time GMV change and an offline diagnosis of Need attention," according to TikTok Shop. "A single day of decline, or a decline that the diagnosis attributes to normal variation, is not flagged."

The guide is careful about what the absence of a flag means. A product without one has "no confirmed decline," in the guide's words, which is not the same as "verified good." The distinction matters because the flag sits on the list sellers see first. A clean list says only that the two-part test was not met.

How long a decline must last, or what the diagnosis treats as normal variation, is not specified. The two-signal design appears to trade sensitivity for fewer false alarms, and the guide does not publish the thresholds behind that trade.

Risk, AA and AB: how factors are ranked

Each diagnosis produces factor cards in three types. The guide defines them in three lines:

  • "Risk: blocking issues such as listing status, stock and violations. Address these first."
  • AA: "factors identified from the change versus the seller's own previous period of the same length."
  • "AB: capability gaps identified against the benchmark for similar products."

The abbreviations AA and AB are never expanded; the guide defines them only by what they measure. A factor that qualifies as both AA and AB appears once, carrying both labels. And there is a cap. "The Summary shows up to 10 factor cards; not every identified factor is shown," according to the guide, which does not say how many factors a diagnosis may identify in total or how the ten are chosen beyond the ranking.

The Summary itself has four parts: the GMV movement, a breakdown by scenario ranked by contribution, the top factors, and a set of actions. Those actions let a seller copy the Summary, mark it useful or not useful, hide or show cards, and open the AI assistant.

Five diagnostic paths

The guide sets out five playbooks, each tied to a pattern in the data.

When overall GMV is down, the guide starts at the Need attention list, opens the product with the largest decline and asks whether the cause is less supply or lower output per item. When the drop comes from affiliate scenarios, it points to the number of active creators and content pieces, GMV per video or stream, the collaboration rate against the target collaboration rate, and whether refundable and free samples are available. A decline in Seller LIVE or Seller Video begins with posting cadence, then moves to product mix, price and stock.

The Shop Tab & Search playbook reads like a conversion funnel. The guide looks for the weakest step: a weak CTR points to the product card, a weak CTOR to the product detail page (PDP) and readiness at checkout, and a weak average order value (AOV) to the mix of SKUs on offer. The fifth playbook covers products whose sales have not fallen but which trail the benchmark; there, the guide prioritises the factor with the largest benchmark gap.

From finding to fix

Each factor type has a destination. Listing status, stock and price lead to Product management, with the product ID carried over. Ratings lead to Product Ratings. A matched opportunity leads to Product opportunities. SKPP eligibility or task progress leads to the SKPP page, SKPP being the Strategic Key Product Pool, which the guide does not otherwise describe. Violations or account risk lead to Account health. Shop Tab & Search conversion factors lead to the relevant listing page or to Seller Academy guidance.

Not every factor has somewhere to go. "Some factors have no destination because they describe a condition rather than a page," the guide explains, citing out-of-stock (OOS) duration and a fulfillment impression rate as examples. No complete list of destination-less factors is given.

Every destination named is a page inside TikTok Shop's own Seller Center or its Academy. None is an advertising tool.

Data rules and the lag question

Times are in US Pacific Time, stated as PST, UTC-8, and amounts are in US dollars.

The most consequential rule concerns timing. "Diagnosis relies on offline data, so it can lag the real-time numbers shown elsewhere in Seller Center," according to TikTok Shop. The length of that lag is not stated. Because the Need attention flag needs both a real-time decline and an offline verdict, a product can show a falling real-time number for some period before the offline diagnosis catches up and the flag appears. That is PPC Land's reading of the two rules together; the guide does not describe the sequence.

Several smaller rules govern edge cases. If the diagnosis period ends early, the page shortens it and states the actual windows used. If benchmark dates are missing, it falls back to the most recent period of the same length. A value of 0 means genuinely zero; a single dash means data is unavailable or abnormal; a double dash means the denominator was zero. If the Summary fails to generate, a fallback card with a retry option appears.

The scope is single-product diagnosis plus the shop-level product list. Shop-level diagnosis continues to follow the existing Shop diagnosis logic. Sellers can export the filtered report, including detail data, and past exports sit under Export Records.

The AI assistant and the disclaimers

The "Chat with AI" entry opens TikTok Shop's assistant with the product context already filled in. The guide hedges its output in a single sentence: "AI-generated content is for reference only and should be checked against page metrics and the seller's own business context before action."

A feedback mechanism sits on each Summary. Sellers who mark one "Not useful" can choose a reason from three options (data inaccurate, attribution wrong, attribution incomplete) and add up to 500 characters of notes.

Two further statements close the guide's account of availability. "The live version of Seller Center is the authoritative source for features, metric definitions and page displays," it says. "This material explains product capabilities only and is not a commitment of business results."

What the guide leaves out

For a document of this length, several gaps stand out. The eight default core metrics are not named. The selection method for the benchmark group is not described, beyond "top-performing products in the same category." The GMV range in the product header is undefined. AA and AB are not expanded. No figures are given on how often products are flagged, how accurate the diagnoses have been in testing, or how long the offline lag runs. And the launch timing goes no further than the month.

The guide applies to the United States only. Nothing in it addresses other TikTok Shop markets.

Why this matters for sellers and marketers

Product Diagnosis is the latest in a run of analytics additions for US sellers this autumn. On November 2, TikTok Shop listed nine new Compass capabilities for US local sellers, among them audience insights in LIVE Diagnosis and an Order Analysis module inside AI Report that traces GMV to SKUs, creators and content types. The same day, TikTok Shop documented Brand LIVE Impact, which credits livestreams with sales on a brand's own website over 14 days. Two days later, a Seller AI guide described tools that attach an estimate of additional orders to each recommended listing fix. Product Diagnosis moves the same logic into the product list: rather than recommending opportunities, it explains declines.

The category-relative approach is not new on the platform either. In June, TikTok Shop introduced a Voice of Customer index that rates each SKU Good, Fair or Poor against its category average, with Poor and Fair products potentially facing enforcement. That tool measures complaints. Product Diagnosis measures sales. Both rest on a category comparison that sellers cannot inspect.

Several of the factors the tool surfaces connect to rules with real consequences. Affiliate program status is one of the header fields, and a Shop Performance Score of 3.5 unlocks affiliate marketing and campaigns, while a drop below 3.0 terminates both. Account health is one of the destinations, and an Account Health Rating of 150 blocks new listings and mega-campaign enrolment for seven days. Price is a factor too, in a marketplace where Smart Promotion charges 3.5% of store GMV in regular periods and 4.5% in major campaigns. A diagnosis that points to price or affiliate performance is, in practice, pointing into those systems.

For agencies and brand teams running creator programmes, the five-scenario split is the most practical change. It separates sales that came from affiliate creators from sales generated by a brand's own content and from sales that arrived through search or the Shop Tab. Whether that split changes how budgets are divided between creator commissions and other spend will depend on how reliable sellers find the attribution. Two of the three reasons offered for marking a Summary not useful concern attribution, one for attribution judged wrong and one for attribution judged incomplete.

Timing adds pressure. Black Friday falls on November 27 and Cyber Monday on November 30, as PPC Land noted in its coverage of the Seller AI guide. That leaves 15 days between today and Black Friday, by PPC Land's count. A tool with a seven-day default window and an offline lag of unstated length will be read most closely during exactly the weeks when a delayed signal costs the most.

The broader comparison is with Amazon, which in September gave its Seller Assistant persistent memory and automations that run while sellers are logged out. TikTok Shop's tool stops short of that. It diagnoses, ranks and links, and the seller still makes the change. The scale behind the effort is considerable: EMARKETER data cited by TikTok put the platform's 2025 US e-commerce sales at $15.82 billion, up 108% year over year.

Timeline

Summary

Who: TikTok Shop, for sellers in the United States using Product Analytics in Seller Center, with implications for brands and agencies that sell through affiliate creators, livestreams and search on the platform.

What: Product Diagnosis, a layer on existing product analytics data that adds a Need attention flag (shown only when real-time GMV is falling and an offline diagnosis agrees), splits each product's GMV across five scenarios, compares key metrics with the seller's previous period and a platform-defined benchmark of top-performing products in the same category, ranks up to 10 factor cards of three types (Risk, AA and AB) and links factors to Seller Center pages. A "Chat with AI" assistant is attached, with output described as for reference only.

When: The US Academy guide is dated today, November 12, 2026, and gives the launch timing as November 2026, 15 days before Black Friday on November 27 by PPC Land's count.

Where: In Seller Center under Product Analytics, on the Product Performance list and the Product Diagnosis tab, documented in TikTok Shop's US Academy under Feature Guide, Seller, Analytics, Product analytics. It is available in the US market only.

Why: According to TikTok Shop, Product Analytics now answers three questions: which product needs attention, why its GMV moved and what to do next. The guide does not disclose how the benchmark group is selected, how long the offline data lags or which eight metrics are shown by default.