AI Deployment
Feature 41  ·  Discovery  ·  Edition Q1 2026

Social is where
Gen Z finds products now.

Seventy-three per cent of US Gen Z say social media is their main source for learning about new products. Not a channel in the mix - the main one. Read alongside what this edition documents about search, it describes a generation whose product discovery never passes through the surfaces most marketing budgets are still organised around.

The headline figure is a claim about primacy, not participation, and that is what makes it consequential.

73% of US Gen Z say social media is their main source for learning about new products.1 Not one of several. The main one.

Around it: 79% of Gen Z and Millennials integrate social media into their shopping journey, and Gen Z are 42% likely to buy holiday gifts through social media against a 20% overall average.1

Figure 01
Social in the purchase journey
Gen Z against the general population.
Bar chart: 73 per cent of Gen Z say social is their main source for new products, 79 per cent of Gen Z and Millennials integrate social into shopping, 42 per cent of Gen Z buy gifts via social against a 20 per cent overall average.
Source: social commerce and consumer behaviour research, 2026. These figures come from different studies with different samples and question wording, and the first two use different populations - Gen Z alone, then Gen Z and Millennials combined. All are self-reported, which measures where people believe they learn about products rather than where they actually do.

The 42% against 20% is the one to hold. Gen Z are twice as likely as the average buyer to complete a purchase inside the same surface where they discovered it.

What "main source" displaces

A discovery channel becoming primary means something else stopped being primary. Two other desks in this edition describe what.

The Content desk's Feature 21 documents Google search traffic to publishers falling 33% globally and 38% in the US, with roughly 60% of searches ending without a click, rising to 77% on mobile. Feature 22 documents ranking and citation decoupling.

Those are usually read as a search story. Set beside this figure they read differently: the discovery layer did not disappear, it relocated - partly into generated answers, and partly onto surfaces where the answer arrives without anyone having asked a question.

Figure 02
Two discovery models
Which describes the surface, and what each requires of the buyer.
Search
The buyer initiates
Requires a formed intentSomeone must already know they want something and be able to name it.
Serves existing demandExcellent at capture, structurally poor at creation.
Measurable by queryWhich is why the discipline built itself around keywords.
Feed
The surface initiates
Requires no intent at allThe product arrives in front of someone who was not looking for it.
Creates demandWhich is what "main source for learning about new products" describes.
Not measurable by queryThere is no query. The keyword apparatus has nothing to attach to.
This framing is ours. It is an argument about what each surface can do, not a measured comparison, and the two are complements rather than substitutes for most categories.

The consequence for measurement is awkward and largely unacknowledged. A discovery event with no query leaves no trace in any keyword-based system. An organisation whose demand-generation reporting is built on search volume is measuring the channel that serves demand and not the one increasingly creating it.

Concentration, and what it costs

The behaviour is also consolidating onto fewer surfaces. Over half - 51% - of US social buyers will shop on TikTok in 2026, crossing the threshold where a single platform represents the majority of social buyers.1

That is a real efficiency for anyone selling: one integration, one creator marketplace, one set of formats reaching the majority. It is also a concentration risk of a specific kind, and Feature 39 of this desk sets out why it is not hypothetical.

A regulatory cascade is underway across more than a dozen countries. A platform holding the majority of a generation's product discovery, in a category subject to active legislation on minimum age, is a single point of failure for any brand whose audience skews young.

What concentration buys
51%
Majority of US social buyers on one platform. One integration, one creator pool, one format set.
What it costs
One
A single regulatory, algorithmic or ownership event moves the majority of your discovery at once - see Feature 39.

The limitation worth stating plainly

Every figure in this feature is self-reported, and self-report is weak evidence for discovery specifically.

People are poor witnesses to how they came to want things. A buyer who saw a product three times on social, once in a friend's hand and once in a search result will often attribute the decision to whichever they remember most vividly - and vivid is exactly what a video is designed to be.

So 73% is a reliable measure of where Gen Z believe they discover products, and a weaker measure of where discovery actually occurs. The direction is almost certainly right. The magnitude is probably overstated, and we found no attribution study that would let us say by how much.

Whose interest this serves

These figures come from social commerce research published by agencies, platform partners and seller tools. A finding that social is the primary discovery channel is directly useful to everyone who produced it. The figures are also broadly consistent across independent compilations, which is worth something - but consistency among parties with the same interest is not independence. Marketing Legendary operates a social practice and benefits from the same conclusion.

What this does to the brief

There is an operational consequence that follows from the search-versus-feed distinction and lands squarely on how creative gets commissioned.

Search-led marketing produces content that answers. Someone has asked a question, and the job is to satisfy it better than eleven alternatives. That shapes everything downstream - the format, the structure, the comprehensiveness, the fact that it must be findable by someone actively looking.

Feed-led marketing produces content that interrupts. Nobody asked. The job is to be interesting enough that a person who was not looking for you stops, and then to make the thing available while they are still stopped.

These require different work, different skills and different measurement, and most organisations brief them from the same document.

Search-led brief
Answer
Comprehensive, structured, findable. Judged on whether it satisfies a stated question better than the alternatives.
Feed-led brief
Interrupt
Immediate, specific, worth stopping for. Judged on whether someone who was not looking pays attention anyway.

Feature 13 of the Ads desk found what consumers actually respond to in the second category: warmth, entertainment, emotional payoff and craft - properties of the idea rather than of the optimisation. That finding sits awkwardly with a content function organised entirely around keyword coverage, and the 73% figure is the reason it can no longer be ignored.

What to do about it

Stop reporting discovery through query volume alone. If a growing share of discovery involves no query, a search-shaped measurement system will report a shrinking market that is not shrinking.

Separate demand capture from demand creation in your plan. They now run on different surfaces with different mechanics and different measurement. Most budgets still treat them as one line called "acquisition".

Do not generalise Gen Z figures to your buyer. The 42% versus 20% gap is the whole point - Gen Z behave very differently from the average, and a brand whose buyers are 45 should not plan against a 73%.

Price the concentration risk explicitly. If a majority of your social discovery sits on one platform in a regulated category, name what happens if that platform's availability changes. Feature 39 documents twelve-plus countries legislating.

Run one attribution study rather than citing survey figures. Self-report tells you what buyers remember. A holdout or geo test tells you what actually moved, and it is the only way to size this for your own business.

Figure 03
Social discovery, 2026
All figures self-reported.
MeasureValueGrade
Discovery
US Gen Z: social is main source for new products73%Self-reported
Gen Z + Millennials integrating social into shopping79%Self-reported
Purchase
Gen Z buying gifts via social42%Self-reported
All buyers, same measure20%Self-reported
Gen Z multiple of average2.1×Our arithmetic
Concentration
US social buyers shopping on TikTok, 202651%Projection
Context from other desks
Google traffic to publishers, global−33%Content F21
Zero-click searches, mobile77%Content F21
Not established
Actual versus perceived discovery source-No attribution study located
Whether the figures share a sample-Different studies
Figures come from different studies with different samples and populations. The 2.1× multiple assumes the 42% and 20% are from the same instrument, which we could not confirm.

How we did this

Where this comes from
From a company that sells into this market: social commerce and consumer behaviour compilations from agencies, platform partners and seller tools. No Tier 1: no survey instrument, sample size or question wording obtained for any figure.
Self-report
Every figure is self-reported. For discovery specifically this is weak evidence - buyers attribute to what they remember, and video is designed to be memorable.
Our own arithmetic
The 2.1× multiple is arithmetic, and assumes the two gift-buying figures come from one instrument. We could not confirm that.
What's ours, not the source's
The search-versus-feed framing, the no-query measurement gap and the concentration-risk argument are ours.

What this doesn't prove

  • Where discovery actually happens. Self-report measures belief. No attribution study we found tests it against behaviour.
  • That social displaced search. The two are complements for most categories. Both trends are real and the causal relationship is not established.
  • That the figures describe one population. Gen Z alone and Gen Z plus Millennials are different groups, from different studies.
  • That 51% concentration persists. It is a projection, in a category under active regulatory pressure.
  • Anything about older buyers. The 42% versus 20% gap establishes that Gen Z are atypical, which limits how far any of this generalises.
  • Anything outside the US. Every figure here is US-sampled.

Sources for this feature

  1. Gen Z social discovery and shopping behaviour, 2026. ringly.io, sqmagazine.co.uk, postplanify.com From a company that sells into this market - agency and platform-partner compilations
  2. Features 21 and 22 of the Content desk, and Feature 39 of this desk. Another feature in this edition
SL
The practice behind this desk

Social Legendary

We separate the platform that holds the audience from the platform that closes the sale, because on this desk they stopped being the same thing.