Why Your Ecommerce Store Isn’t as Holiday-Ready as You Think

Every merchandiser has a peak season checklist. Inventory buffers, promotional calendars, search synonym lists, a landing page for every campaign. Most of those checklists were built for a world where a shopper typed a query into a box on your site and looked at what came back. That world still exists. It is simply no longer the only one. A fast-growing share of holiday product discovery now begins inside an AI assistant that reads your product data long before any shopper sees your storefront, and the assistant decides whether you are an answer. Syntheum.ai’s guide covers what still matters, what has quietly stopped mattering, and the layer most peak season checklists do not yet have.

The numbers behind that shift are no longer speculative. Adobe forecasts $275.1 billion in U.S. online holiday sales this season, up 6.7% year over year, with traffic from AI sources up 130% against last year. Salesforce expects 20% of holiday ecommerce traffic to arrive from AI chat agents. Whatever your merchandising plan says, one fifth of your visitors are being prequalified somewhere you cannot see.

1. Search still drives discovery, but it no longer starts on your site

Onsite search remains one of the highest intent surfaces you own. What has changed is how many shoppers reach it, and what has already happened to them before they do.

Salesforce reported that traditional search engine usage and marketplace usage each fell 15% year over year, while new channels including AI assistants and social surged 38%. During the 2025 holiday season, AI-powered search traffic doubled and converted at nine times the rate of social referrals. Half of shoppers now use an AI assistant somewhere in the buying journey, and that share rose sharply over the past year.

The practical consequence for merchandisers is uncomfortable. A shopper who once searched your site and left a logged query now asks an assistant instead, and your internal search reporting never records the intent at all. Your zero-result rate looks healthier than it is, because the hardest queries never reached your site.

Discovery Checklist

Pull your top 200 holiday queries from last peak and check zero-result rates
Identify queries with high volume and low add to cart, which are merchandising failures rather than search failures
Confirm synonym coverage for seasonal language such as gift, present, stocking stuffer, and under fifty dollars
Log zero-result queries daily through peak rather than weekly
Compare onsite search volume year over year, and treat a flat number against rising traffic as a signal that discovery moved off site

2. Your product data is read by machines before it is read by people

This is the layer missing from most peak season checklists, and there is now hard evidence of how badly it is being handled.

Adobe measured how much retail product content is actually readable by AI systems and found that fewer than two-thirds of product listings parse cleanly in any category. Cosmetics performed best at 63% readable. Electronics reached 56%. Sporting goods and apparel sat at 51% each. Grocery managed 48%, and furniture and home came last at 47%.

Read that again with a merchandiser's eye. In the home category, more than half of product listings cannot be reliably interpreted by the systems that an increasing share of shoppers are asking for recommendations. Ambiguous attributes do not merely perform poorly. They make a product ineligible to be surfaced at all.

The upside is equally measurable. AI-referred visitors spend 53% more time on site, view 23% more pages, and convert 54% better than non-AI traffic. That last figure is a reversal. In 2025 the same traffic converted at roughly half the rate of other sources. The audience has matured, and the retailers whose data is legible are the ones capturing it.

On-site Checklist

Audit your top 100 holiday SKUs for complete, machine-readable attributes covering size, material, compatibility, and age range
Remove merchandiser shorthand from public-facing attribute values
Verify structured data renders on product detail pages, not only on category pages
Confirm gift-relevant attributes exist as structured data rather than only as marketing copy
Spot check how your top ten gift SKUs are described when an AI assistant is asked to recommend them

3. Checkout friction is still real, and still mostly self-inflicted

Nothing in the shift toward assistant-led discovery has repaired the checkout. Baymard Institute, aggregating 50 separate studies, puts the documented average cart abandonment rate at 70.22%. Salesforce observed a higher 82% during holiday conditions, which is what peak traffic and heavier comparison shopping produce.

What makes this worth a line on the checklist is that the leading causes remain entirely within your control. Extra costs revealed too late account for 40% of abandonments. Forced account creation accounts for 18%. A checkout that is simply too long accounts for 17%. The leading causes are still configuration decisions, made once and never revisited.

Mobile raises the stakes further. Adobe expects mobile to carry 57.4% of seasonal spend and 63% on Thanksgiving itself, and Salesforce puts mobile at roughly three quarters of all online traffic. A checkout that is merely tolerable on a phone is a checkout that is failing most of your customers.

Checkout Checklist

Display full landed cost including shipping and tax before the final step
Enable and test guest checkout on mobile, not only on desktop
Bring mobile wallet options live before peak traffic rather than during it
Count your checkout steps and remove any that exist for internal convenience
Treat discovery absence as a funnel metric, because a shopper who never reached you did not abandon a cart; they were never offered one

4. Promotions built for humans confuse machines

"Buy one, get one half off on select styles" is perfectly legible to a shopper standing in front of a banner. It is close to meaningless to a system trying to determine what a product actually costs.

This matters more this year than last because discounting is deeper and more dynamic. Adobe projects peak discounts reaching 30% during Cyber Week, building from roughly 14% in early November. Every one of those changes needs to be expressed in data, not only in creative, or your catalog becomes less accurately representable at exactly the moment representation matters most.

Marketing Checklist

Express every holiday promotion in structured pricing data, not only in campaign creative
Define "select styles" at the SKU level
Verify sale pricing appears in product markup throughout the promotional window
Confirm expired promotions clear from structured data on schedule
Check that bundle and threshold offers resolve to a comprehensible unit price

5. Holiday content should outlast the season.

Most holiday content is written to be disposable. It is dated, campaign-bound, and effectively dead by mid-January. That is a choice rather than a constraint.

Content tied to a specific year earns traffic for roughly eight weeks and then becomes a liability, signaling staleness to shoppers and search / LLM crawlers alike. Content organized around a recurring season compounds annually instead. The same logic applies to the shopping calendar itself, which keeps starting earlier. Adobe recorded $95.8 billion in October online spending, up 8% year over year, which means a guide published in November is published late.

Content Update Checklist

Write peak season guidance without the year in the title or the URL
Separate evergreen guidance from campaign-specific creative so the former survives January
Date stamp substantive updates transparently rather than silently republishing
Publish seasonal guidance in early October, before the spending curve begins
Plan the January refresh before December rather than after

6. Returns are a merchandising signal, not a fulfillment cost

Returns are usually owned by operations and ignored by merchandising, which wastes the single most honest dataset either team will see all year.

The National Retail Federation put total U.S. returns at $849.9 billion, representing 15.8% of annual sales and 19.3% of ecommerce sales specifically, with roughly 17% of holiday sales expected to come back. Salesforce measured $181 billion in holiday returns, around 14% of purchases and up 10% on the prior year.

A return is a shopper telling you precisely where your product data and your merchandising diverged from reality. Sizing returns point at attribute gaps. Not as described returns point at imagery or copy that oversold. Wrong item returns point at search relevance. None of that reaches the merchandising team unless somebody deliberately routes it there.

Fulfillment Checklist

Capture return reasons as merchandising input rather than only as fulfillment data
Map the top five return reasons for your highest volume holiday SKUs to a specific content or data fix
Flag any SKU whose return rate exceeds category average before you merchandise it again next year
Review whether returns cluster around products with incomplete attributes

7. Post-peak is where next year's strategy gets written

The most valuable merchandising intelligence you will generate this year arrives in the two weeks after peak, when returns, search logs, and abandoned sessions reveal the gap between what you merchandised and what shoppers actually wanted. Most teams are too depleted to capture it, which is exactly why the teams that do pull ahead.

There is also a structural shift worth preparing for. Salesforce predicts one in three ecommerce sites will have shopper agents deployed by Cyber Week, and retailers who deployed branded agents last season grew sales 6.2% against 3.9% for those who did not. Whatever you learn this January becomes the input for that decision.

Post-season Checklist

Schedule the post-peak search log review before peak begins, while you still have the energy to commit to it
Record what sold without being merchandised, because unplanned winners are next year's strategy
Archive the full holiday query set for year over year comparison
Capture which products were recommended by assistants and which were invisible
Write the retrospective in January, not in March when the detail has faded

Frequently asked questions

When should holiday merchandising preparation begin?

Product data and search readiness work should be complete before traffic arrives, which in practice means early October. Adobe recorded $95.8 billion in October online spending this year, up 8% year over year, so the season now begins well before the traditional Thanksgiving starting line. Promotional configuration can follow. The sequence matters, because data quality problems discovered during peak cannot be fixed during peak.

Does AI-driven product discovery replace onsite search?

No. It adds a discovery path running parallel to your site rather than through it. Onsite search continues to convert strongly for shoppers already on your store. The new layer determines whether some shoppers arrive at all, which is why both need attention and why neither substitutes for the other.

What is the single highest-impact item on this list?

Product attribute completeness for your top holiday SKUs. It is the prerequisite for everything downstream, including search relevance, promotional accuracy, and eligibility in assistant-mediated discovery. Adobe's finding that under half of furniture and home listings are machine-readable suggests most categories have substantial ground to make up.

How is this different from standard SEO preparation?

Traditional preparation optimizes pages so they rank. This optimizes product data so it can be retrieved and represented accurately by a system answering a question somewhere else. The disciplines overlap but they are not the same, and merchandising teams own considerably more of the second than they typically realize.

Is it too late to act if peak has already started?

Sections 3 and 4 remain actionable at any point, because checkout configuration and promotional data can be corrected mid-season. Sections 1, 2, and 5 require lead time and are better treated as next cycle work. Section 7 costs nothing to schedule now and pays for itself in January.

The merchandiser's advantage

It would be easy to read a list like this and hand the AI layer to a technical team. That misses the point. Everything an assistant reads was written by a merchandiser, and the team that understands the catalog best is still the merchandising team. Handing that work away is how merchandising teams lose ground.

Everything the assistants are reading was written by merchandisers. The attribute taxonomies, the product copy, the category logic, the promotional structures. The discipline that determines whether a product is eligible to be recommended is the same discipline that has always determined whether it was findable, and the people who understand the catalog best are the ones already sitting in the merchandising function. What has changed is the audience, which now includes machines, and the cost of ambiguity, which has risen sharply.

That is the genuinely good news in a season of uncomfortable numbers. The work in front of you is not unfamiliar. It is the work you have always done, held to a higher standard of precision, with a far larger prize for getting it right.

Start with your top hundred SKUs, fix what the data says is broken, and go into January knowing what to do next.