The first $10Blive-commerce year (and what the numbers don't say)

When industry reports said TikTok LIVE crossed $10B in GMV last year, the takeaway from most boardrooms was simple: more Mega creators. More 1M+ followings. More celebrity drops.

The takeaway is also wrong.

We sit in a fairly unusual position to check. Every live that runs on Tikee gets recorded server-side and scored — duration, peak viewers, mentions, GMV (when affiliate-attributed), gifts. That's 142,000 lives across 38 countries in 2025, on a roughly balanced creator mix from 1K-follower nanos all the way up to multi-million-follower megas.

What follows is what the data actually says — boiled down to the four findings most brands are getting wrong.

The Micro tier is doing the heavy lifting

If you cluster 2025 GMV by creator tier and normalize for the size of each tier in our dataset, the Micro tier (10K – 50K followers) produced 47% of all attributable revenue. Megas produced 11%. Macros, Mids, and Nanos make up the rest.

This isn't a fluke. It tracks something we've watched at the per-campaign level for two years: viewer retention at the Micro tier is much higher (mean 22 minutes vs. 14 for Mega), the audience knows the creator, and conversion intent is stronger.

Mega creators are good for awareness. They're not as good for moving units in a 45-minute window. The two motions confuse each other when CMOs report “creator influence” as one bucket.

Time-of-day matters more than you think

Across every region in our dataset, the “evening” GMV window — defined as 6pm to 10pm local time — outperformed the rest of the day by 2.8×. This isn't surprising in itself. What's surprising is how much brands are still scheduling lives between 11am and 2pm “to catch lunch breaks.”

Lunch lives convert at 0.6× of evening lives, on average. They look fine on a dashboard if you don't compare. They're a money loss on the merits.

The brands that win in live commerce treat their creator calendar like a TV schedule. Lunchtime is for talk shows. Prime time is for shopping.— Elena Marsh, Head of Insights

Niche overlap matters more than follower count

One of the most-shared findings inside Tikee is what we call the “niche overlap multiplier.” When a creator's niche tag set overlaps with the brand's tag set by 60% or more, GMV per live is 3.4× higher than when overlap is below 30%, holding every other variable equal.

Most agencies will say this is obvious. But agencies still pitch deals on follower count and “category alignment” — a much fuzzier signal. The hard overlap math doesn't usually make it into the deck.

• Beauty × beauty overlap drives 3.8× the GMV vs. beauty × general lifestyle.
• Tech × tech overlap drives 5.1× the GMV vs. tech × business.
• Food × food overlap drives 2.6× the GMV vs. food × general lifestyle.

The “verification effect” is real

Here's the one that surprised our own team. Campaigns where the brand requires Tikee's automated verification — recording + duration/mentions/GMV scoring — produce 18% higher GMV than equivalent campaigns where verification is disabled.

Our best theory: creators perform better when they know the live is being scored. Mentions land at the right beats. Demos are tighter. The 45-minute mark gets hit deliberately instead of accidentally. Lives stop drifting.

This isn't a Tikee sales pitch. The effect is small at the per-live level. But across a 20-live campaign, it's the difference between a 3.7× ROAS and a 4.4× ROAS — and that's what the data tells us.

What this means for 2026

The brands that under-performed in 2025 weren't bad at picking creators. They were bad at weighting creators — over-indexing on Megas, scheduling outside prime time, and stretching their niche definitions until they snapped.

For 2026, the playbook that's emerging on Tikee looks something like:

1. Build the creator slate from the Micro tier outward, with 1–2 anchor Megas for awareness only.
2. Schedule 70% of lives between 6pm and 10pm local.
3. Filter creators by hard niche overlap — discard anything below 50%.
4. Verify every live. Always.

It's not the playbook the industry reports are pitching. It is, as far as we can tell from 142,000 lives, the playbook that works.