Sell products on Instagram and TikTok

A social storefront is not a second website. It is a copy of your catalogue living inside somebody else's app, kept in step by a feed, feeding a set of surfaces — a shop tab, product tags on posts and video, ads that pull from the same catalogue. So the first question is not what to post. It is whether your product data can survive being read by a system that has never seen your store, cannot see your theme, and rejects anything ambiguous. Titles written for a human already browsing a category, prices that live in a plugin rather than in the product record, stock that refreshes once a night: all of these are invisible on your own site and hard failures in a catalogue. Content sits on top of that layer and cannot compensate for it.

There are three ways your products reach a social platform, and they fail differently. An official connector plugin maps WooCommerce fields to the platform's commerce catalogue and pushes changes on a schedule. A generated feed file — CSV or XML at a URL — waits for the platform to fetch it on its own cadence. Or you upload manually, which stops being viable somewhere around fifty products. Whichever you choose, the platform then reviews each item and can reject it quietly, one at a time, without emailing anyone. Sync is not instant either: a price change on your site can take hours to reach a tagged post, and a sold-out variant may stay buyable in the app until the next fetch. Plan around the lag rather than discovering it during a promotion.

• Choose deliberately: connector plugin, hosted feed file, or manual upload
• Check the rejected-items queue weekly — nobody notifies you
• Know your sync interval before planning a flash sale around it
• Map variants on purpose; colour and size become separate items

The hygiene a social catalogue needs is the hygiene Google Shopping needs, which is convenient: fix it once and both channels improve. Titles should read as a search query and a shelf label at the same time — brand, product, the one attribute that distinguishes it — not the padded string someone wrote for a category page. Prices must match what the shopper is charged, including currency and any sale price expressed in the field the platform expects, because a mismatch is one of the faster routes to a suspended catalogue. Images want the product on a clean background: no price stickers, no promotional overlay text, no watermark, and nothing that turns into nonsense when cropped to a square or vertical frame. If you have already done this for Google Shopping, most of the work is behind you.

• Titles: brand, product, distinguishing attribute, in that order
• Price and sale price identical to the amount actually charged
• One clean primary image with no overlay text or watermark
• GTIN or MPN wherever you have it; identifiers reduce rejections

Then the decision that shapes everything downstream: where does the money change hands. Buying inside the app is fewer steps, since payment method and address are already stored, and Baymard Institute's checkout research is a long argument that every removed step is worth something. What you trade away is ownership. The order lands in the platform's seller tools, the customer's email may be masked or missing, your list does not grow, your subscription and wholesale logic does not apply, and funds settle on the platform's schedule. Sending the shopper to your own store costs steps and returns the customer, the data and every rule your store already enforces. Availability moves too — native in-app checkout has opened and closed by market before, so verify what is actually available to you. If you send people home, that path should be as short as you can make it, which is what a native checkout handoff is for.

An ad or a tagged post that drops everyone on the homepage wastes most of what you paid for. The shopper saw one specific thing; they arrive somewhere that asks them to find it again, on a phone, in a session that began as entertainment. Every commerce surface on both platforms can deep-link to a specific product URL, and the page they land on should assume zero context: the item, the price, the variant they saw, and the buy control visible without scrolling. Social traffic is unusually unforgiving here because attention has to survive an app switch and a cold page load. If your template only makes sense to someone who has already browsed a category, the click is spent. Product page work is the highest-leverage fix, and it pays back on every other channel.

Now the part nobody enjoys. When someone taps a link inside Instagram or TikTok, your site usually opens in the app's in-app browser rather than Safari or Chrome. That environment keeps its own storage, so a returning customer can look brand new, a saved cart may be gone, and a logged-in session can vanish. On iOS, Apple's App Tracking Transparency framework means much of the click-level identity you would once have relied on is simply unavailable unless the user opted in, so platforms model conversions instead of counting them. Add a consent banner that holds tags until acceptance and you have three independent reasons the reported numbers will disagree with your order list. The answer is not more tags: it is server-side event forwarding where offered, consistent identifiers, and monthly reconciliation, which is how analytics is wired on a rebuilt storefront.

• Expect the in-app browser to look like a first-time visitor
• Reconcile platform-reported conversions against real orders monthly
• Use server-side event forwarding where the platform supports it
• Treat modelled conversions as a direction, not a receipt

A catalogue does not create demand. This is the common failure: a store connects the feed, tags every post, waits, then concludes the integration is broken because nothing sold. Usually the integration is fine and the content is not earning attention. Feeds make it easy to buy something a person already wants; they do not make anyone want it. If your posts get little watch time, few saves, and no comments that sound like questions from buyers, fix that before spending on plumbing. There is a second case for doing nothing. A catalogue of thirty products with steady demand elsewhere often does better with a link in bio and one strong product page than with a full commerce integration and its review queue. And if your margin cannot absorb platform fees plus a higher return rate, the honest answer may be that this channel is not for you yet.

The last cost is running it. This is a cadence, not a project: something has to be published often enough that the shop surface is not an empty room, and that work does not end when the integration does. Inbound moves to DMs and comments, which is now a support channel with same-day expectations and no ticket history unless you build one. Returns behave differently as well, since an impulse purchase from a video is decided in seconds and reconsidered later far more often than a purchase made after comparing a category. Check your returns policy and your margin against that before scaling spend. Decide who owns the reply queue, who reads the rejected-items list, and what the delivery promise says. A channel nobody owns quietly rots, and the catalogue keeps advertising products you no longer stock.