A video shows the standard machine. The caption describes a capability available with an optional module. Both assets may be accurate in isolation. Presented together, they can imply that the module is included in the machine on screen. This is a combination problem, not necessarily an exaggerated claim in either original file.

TikTok's Smart+ upgrade documentation, marked as updated in October 2026, describes ads generated from combinations of videos and ad text. It also describes ads as asset groups with their own destination URLs. Availability can vary by account. Before using the available automation, an overseas B2B marketer needs to know which assets can truthfully appear together.

Build groups around compatible conditions

Organizing a library by product family is convenient, but it is not enough to establish compatibility. One family can include different sizes, market versions and service packages. A demonstration of a compact model may not support the same application statement as a larger model. A price for the equipment alone may not describe the installed system shown in a video.

For a campaign promoting products made in China, the product owner and the local marketing team may hold different parts of this information. The factory knows the configuration; the marketer knows the language and destination page. Neither should have to infer what the other intended from a filename. The review needs the visible model, the scope of the text claim and the offer reached after the click.

Mark the combinations that must not run

A useful preparation step is a compatibility matrix. Put videos in rows and candidate text in columns. At each intersection, identify whether the combination is valid, needs an explicit qualification or should not run. Write the reason in product terms: the accessory is absent, installation is excluded, or the application statement belongs to another version. This makes the decision inspectable without creating another vague approval label.

Then group assets whose conditions genuinely match. If a sentence is true for only some videos, separate those videos or rewrite the sentence accurately. Do not rely on the buyer to discover the limitation after clicking. Equally, do not interpret the system's preference for a combination as evidence that its product statement is correct. Delivery performance concerns audience response; it does not validate the offer.

Neutral text is not automatically harmless. A phrase such as “complete system” can expand what the pictured configuration appears to include, even without a numerical claim. Read the caption as a statement about the object currently on screen. If its meaning depends on a hidden exception, either make that exception visible or remove the pairing. A broad family slogan should not erase the distinction between the equipment shown and the package actually offered.

Check the path a buyer can actually see

Review the destination with the combination, not as an unrelated website task. If the ad shows one variant and the page opens on another default variant, the buyer has to reconstruct the intended product. A general company homepage is unlikely to resolve a configuration-specific claim. The page should identify the model or clearly explain the range covered by that ad group.

Revisions can change the compatibility decision. A crop may remove the small label identifying an optional part. A shorter voiceover may retain the benefit while dropping the condition. A translated caption may widen a narrow application statement. The original files can remain in the library while the current presentation no longer supports the same approval. Review the rendered version that will run, not just the unchanged name of the asset.

This does not require testing every imaginable combination in a large, unstructured pool. It argues for making the pool smaller and coherent before delivery. Start with the combinations the chosen account controls permit, then remove incompatible crossings. If the interface cannot express the necessary separation, simplify the asset group instead of assuming a restriction exists somewhere else.

Spark and non-Spark identity decisions remain a separate part of planning, discussed in our guide to organizing those assets before scheduling. Assets that share an advertising identity still do not automatically share product conditions. Rights, identity and factual compatibility answer different questions.

For the first review, select one product family rather than the whole library. Ask the product owner to identify version boundaries, the editor to revise sentences that are not universal, and the campaign operator to reproduce the intended grouping in the actual account. Keep the reason for each exclusion. The deliverable is a set of combinations that remain understandable together, not a folder in which every file happens to have been approved separately. Platform approval and campaign outcomes remain subject to TikTok's controls and actual delivery.

Sources