Templates & slots

How Catalog Video templates work — the AI Template Studio, typed slots, binding expressions, spoken personalization, hooks, and the robustness probe.

Last updated July 2, 2026

A template is an approved master video whose changeable parts are marked as slots. Bind a product’s data into the slots and you get that product’s video — same craft, personalized content.

The AI Template Studio

You don’t have to design templates by hand:

  1. Pitch first. Give a brief (or just your brand URL) and the Studio pitches three concepts — short descriptions with draft storyboard sketches. Concepts cost cents, not full generations.
  2. Pick one. The chosen concept is produced as a full master video on the normal generation pipeline — brand grounding, style, quality gates, everything.
  3. Slots are extracted automatically. Wavemaker walks the approved master and proposes slots: the headline, the price, the product image, the end-card copy, spoken lines. You confirm the proposal on a review card.

Templates are versioned and immutable — editing one creates a new version, so a running campaign never changes under you. Seasonal variants (holiday wardrobe) can be authored ahead and scheduled.

Slot types

KindExampleNotes
textheadline, subtextmax-length + fit strategy per slot
price”Now $39.99”must match your feed data — fabricated prices are blocked
imagethe product shotfed by the item’s prepared images
spokena re-voiced VO lineper-product voiceover in the master’s designed voice
endcard / logo / badgeclosing card fieldslogo slots never receive product photos — a row without its own mark keeps your brand logo
clipthe hero scenehero-tier products get a generated shot from their real photo

Binding expressions

Slots bind data with a small, safe expression language:

{display_title}
Now {price|currency}
{title|truncate:40}
{?sale_price}Was {price}, now {sale_price|currency}!{/?}
  • Formatters: currency, number, spoken, titlecase, upper, lower, truncate:N.
  • Spoken formatting turns “$29.99” into “twenty-nine ninety-nine” so voiceover reads like an ad, not a spreadsheet.
  • Null-safe conditionals{?field}…{/?} drops the whole segment when the field is missing, so “Was , now $29.99” can never ship. Rendering a raw empty value is a hard block, not a warning.
  • Fit strategies per text slot — shrink (responsive type), truncate (ellipsis), or wrap — declared once, enforced on every product.

Hooks: structural variety for free

A template can carry 2–3 hook variants — alternate opening scenes authored once. Products are assigned a hook deterministically, so a 10,000-product catalog isn’t 10,000 structurally identical videos, and fatigue rotation has pre-staged alternates ready.

The robustness probe

Before a template can be approved, it’s test-rendered against your catalog’s worst cases: the longest and shortest titles, the highest and lowest prices, items missing optional fields, the worst-scoring image, and non-ASCII text. Every probe must pass. This catches the classic catalog-ad failures — overflowing text, ugly image crops, blank interpolations — once per template instead of hundreds of times per catalog.

Routing multiple looks

One campaign can route different products to different templates with ordered rules — category = jewelry → template A, price > 500 → premium template, default → template B — so a multi-category catalog isn’t forced through one look.

The price-in-video policy

Ad platforms overlay live prices from your feed at serve time; a price baked into the video can go stale between syncs. Price slots are therefore flagged volatile: enabling one on a rarely-synced campaign warns you at save time, and evergreen (price-free) templates are the default for most catalogs. Spoken prices are best for stable-price catalogs or fixed-end-date offers.