TikTok Smart+ and Advantage+ Campaign Build

TikTok Smart+ hands targeting, bidding and creative selection to the model, which means the model needs conversion volume before it has anything to learn from. We do not launch Smart+ Web on an account producing fewer than about 50 attributed conversions a week; below that the campaign spends its budget discovering what it cannot know, and you will conclude that Smart+ does not work when what did not work was the timing. Where volume is short, we build to a mid-funnel event first and move the optimisation target once the data supports it. What we build: Account structure that will still make sense in nine months. One naming and UTM taxonomy applied across Meta and TikTok so your warehouse can join them without a lookup table. Campaign-level budget with the automated campaigns sized so they are not competing with your manual campaigns for the same catalogue at the same moment, which is the most common self-inflicted wound in accounts we inherit. Catalogue work, which is most of the job: For fashion and marketplace inventory the feed is the campaign. Product set structure built on margin bands and sell-through, not on the site's category tree, using custom_label_0 through custom_label_4 for what they are actually good for. Availability and price synchronisation checked against real disapproval rates rather than assumed from the feed spec. Variant handling, because a colourway split across twelve item IDs teaches the model twelve weak lessons instead of one strong one. Where you sell on TikTok Shop we set up GMV Max separately and keep it out of the same optimisation pool. On Meta: Advantage+ shopping with the existing-customer budget cap set deliberately rather than left at default, exclusion logic that survives audience refresh, and placement decisions made rather than accepted. Asset requirements handled properly: 9:16 at 1080 by 1920, sound-on assumption for TikTok, safe zones checked against the current UI overlay rather than last year's template, and a minimum runtime per format that we specify per campaign type. What we refuse: We will not build on an account whose event data we have not verified, so either we run the measurement engagement first or you show us a recent EMQ and postback audit from someone competent. We will not build Smart+ and a manual prospecting campaign against the same catalogue and the same audience and then attribute the result to either. We will not accept a brief that says test everything; a build with fourteen simultaneous hypotheses is a build that answers none of them, and we will make you choose three. What you receive: The account built and documented: structure diagram, naming convention, and the reasoning for each optimisation target written down. Feed specification with the mapping applied and the disapproval rate measured before and after. A launch checklist your team can run again next season without us. Two weeks of post-launch supervision, which is the period where automation makes its ugliest decisions and someone needs to know which ones to leave alone. Not included: Creative production, ongoing management, feed engineering inside your PIM, and content for TikTok Shop listings. Media budget. Who this is not for: Advertisers whose first campaign this is. Automated campaign products are the wrong first structure for an account with no conversion history, and we would be selling you a build that cannot succeed. Teams who want the account handed over with no documentation because it is faster; the documentation is what makes the build worth this fee.