How a TikTok Shop Cold-Start Ignition Works | MomentIQ

Week 0: Everything starts at zero

A new TikTok Shop has no affiliates, no content, and no sales history for the algorithm to learn from. There is nothing wrong with the product — there is simply no evidence yet, and TikTok distributes on evidence. This is the state every ignition begins from, including the ones that end up working.

What actually happens here: Before anything is spent, the catalogue, pricing, and affiliate plan are set up so a creator who wants to post can actually do it without asking anyone for permission.

• Affiliates posting: 0 (illustrative)
• Videos live: 0 (illustrative)
• Attributed GMV: $0 (illustrative)

Weeks 1–2: Seed creators, deliberately

Outreach fans out to creators, and the targeting is the whole game. The largest accounts rarely accept an unknown brand, and the smallest rarely convert. The workable band sits in between — creators big enough to have an audience that buys, small enough to still say yes to a product they have not heard of.

What actually happens here: Outreach goes wide because acceptance rates are low by nature, and samples are dispatched fast — a slow sample is the single most common reason a seeding push stalls before it starts.

• Creators contacted: hundreds (illustrative)
• Samples shipped: dozens (illustrative)

Weeks 2–5: Volume comes before velocity

Posts begin appearing. Most of them do very little, and that is the expected shape rather than a failure — you are buying attempts. The angle that works is discovered, not designed, and it is discovered by having enough different creators try enough different framings that one of them lands.

What actually happens here: Nothing is optimised yet. The job in this window is to keep supply moving and to watch which framings and hooks recur among the posts that do better than the rest.

• Posts live: tens (illustrative)
• Posts that outperform: a handful (illustrative)

The turn: One video catches

A single post behaves differently: it keeps being served, it keeps converting, and the shop feels it. This is the moment the whole cold start exists to produce, and it is also the moment most brands under-react to — they celebrate it instead of chasing it.

What actually happens here: The response is mechanical, not emotional. The creator is re-briefed and re-sampled, the format is handed to other creators to run in their own voice, and the product page and stock position are checked before the traffic arrives, not after.

• Winning formats identified: 1–2 (illustrative)

After the signal: Live and paid layer on — in that order, and only now

With a proven format and a product that converts, live selling and paid spend stop being experiments and start being multipliers. Run before the signal, both are expensive ways to learn what a few hundred dollars of samples would have taught you.

What actually happens here: Paid budget goes behind creative that already converts organically. Live sessions are scheduled around the products the data has already picked, not around the ones the brand likes most.

Where you would be: How long this takes, honestly

Timelines vary by category, price point, stock depth, and how quickly samples get out of the door. No two ignitions are identical, and any agency quoting you a fixed number of days to first GMV is quoting you a number they cannot know. What is consistent is the ORDER: foundations, then attempts, then a signal, then amplification — skipping a step does not accelerate the sequence, it just makes the expensive parts happen before the cheap parts have done their job.

What actually happens here: The right question to ask a partner is not 'how fast', it is 'what happens in week three if nothing has caught yet' — because that is the week most cold starts are abandoned.