How to market a new product online in the first 90 days, on a small budget

The first two weeks go to the product page

Traffic you send to a weak page is money spent proving the page is weak. Before any ad, get the listing images right, because they are what people look at before they read a word. On a marketplace that means a main image on white that meets the platform’s rules (Amazon, for one, requires pure white for the main image), then secondary images that each answer one question: what it does, how big it is, what comes in the box, how it is used, and which detail shows the quality. On your own store you have more freedom, but the same questions apply.

Say you launched a line of waxed canvas tote bags. The page needs the bag on white, the bag on a shoulder to show scale, the inside pocket open, the stitching up close and the three colours side by side. Shoot them on your phone near a window, or start from one good photo and let the product listing image generator plan the set slot by slot. Compare every generated image with a real bag before it goes live; a strap that changes width between images is a small error that makes the whole page look less trustworthy.

If you sell on Amazon, our Amazon main image guide covers the rules for that first image and what gets a listing hidden from search.

Weeks three and four bring three ads with three openings

With the page ready, make a small set of video ads rather than one expensive one. Keep the product, the offer and the ending the same and change only the opening: one starts on the product in use, one on the problem it solves, one on a detail people don’t expect. When one opening clearly holds attention better, you have learned something you can reuse in every later video, and a single ad can’t teach you that.

Make them vertical at 9:16 for TikTok, Reels and Shorts, plus a square or 4:5 still or two for feeds. The AI ad generator plans several ads for one product at once, with different hooks and the same offer. If your product page is already live, URL to video reads the name, price and selling points from it; you still upload a clear photo of the exact variant, because that photo is what the video is built from.

Before you spend, check whether your category has its own advertising rules. Supplements, CBD, alcohol, lingerie and a few others do on Meta and TikTok, and our ad policy guides collect them by category. An ad rejected in week three costs you a week on top of the creative.

Creator-style UGC that doesn’t pretend to be a customer

UGC ads for small brands work because they look like a person showing you something rather than a brand talking at you. That is also where small brands get into trouble. The FTC’s final rule on fake reviews and testimonials, announced in August 2024, covers testimonials that misrepresent that they come from someone who doesn’t exist, with AI-generated fake reviews as its example, and testimonials from people with no actual experience of the product. An AI presenter saying “I’ve used this every day for a month” is exactly that kind of testimonial.

There are two honest ways to use the format:

• AI presenters who show rather than vouch. A presenter can hold the tote, open the inside pocket and say what is happening on screen (“the strap adjusts, here”). The AI UGC ad generator writes the lines from the facts you give it. Leave out first-person results, and label the AI where the platform asks: TikTok Shop’s content guide requires creators to disclose AI use, especially for synthetic faces, voices and digital humans.
• Real creators who actually used it. Send the product to a few people whose audience fits and ask for their honest take. The FTC’s guidance for influencers says a free or discounted product is a material connection, the disclosure belongs in the video itself rather than only in the description, and nobody can talk about their experience with a product they haven’t tried.

Friends, family and the split between AI and real creators

Launch posts from people close to you count too. The FTC’s influencer guidance tells people to disclose any financial, employment, personal or family relationship with a brand, so your sister’s post about the tote needs the same plain disclosure, such as “ad”, that a paid creator’s would. It is an awkward conversation for about a minute, and it keeps the launch clean.

A workable split for a small budget: use AI presenters to find out which opening and which product moment hold attention, then brief one or two real creators with the winner. The AI version is quick to change and costs little per variation. The real person carries the trust, and for products that sell on fit, feel or taste, only a real person who has used the product can speak to that.

Days 31 to 60 are for reusing what worked

By the second month you should have one or two videos and a few images that clearly do better than the rest. Don’t make new content for every channel; cut what you already have. The winning video becomes the video on your product page, a listing video on marketplaces that accept one, and the hero of your launch email. A still frame from it, the moment the product does its job, becomes the email header and a feed post.

Email is easy to neglect when ads take all your attention, and it shouldn’t be. Everyone who bought or signed up in the first month has already said yes once. A short sequence is enough: the product in use, the close-up with the story of how it is made, an answer to the question you get most often in messages, and the other colours. None of that needs a new shoot if the listing set was done properly in the first two weeks.

Social posts follow the same rule. The openings that lost the ad test can still go out as organic posts to people who already follow you, since the audience is different. Change the caption and the first line rather than reposting an identical file in the same place.

What to measure, and what to ignore for now

With a small budget you won’t get statistically clean answers, so measure the few things that show where the path to a sale breaks. For each ad: how many people watched past the opening, how many clicked through, and what each click cost. For the page: how many visitors added to cart and how many bought. For the month: what each sale cost you across ads and samples, compared with your margin on one unit. Read the numbers as a chain:

• Few people watch past the first seconds: the opening is the problem, so test new openings before anything else.
• People watch but few click: the video entertains without making the product the point, or the offer is unclear.
• Plenty of clicks, few add-to-carts: the page, the price or the images are the problem, not the ad.
• Add-to-carts but few purchases: look at shipping cost, delivery time and the checkout itself.

Keeping the numbers honest with a simple log

Ignore follower counts and likes for the first 90 days. They feel like progress and say little about why sales are or aren’t happening. Change one thing at a time, write the date and the change in a simple log, and give each change at least a week before you judge it. A spreadsheet with five columns (date, what changed, spend, add-to-carts, sales) is enough.

The log matters more than any dashboard because it is the only record of cause and effect you will have. When a good week arrives, you want to know whether it came from the new opening, the creator post or the price change, and memory is unreliable after a month of launches.

Days 61 to 90 decide where to double down

By the third month the log should show a pattern. If one angle sells (say the tote’s inside pocket keeps winning), make five more videos about that angle instead of five about new ones, brief a real creator around it, and move the pocket shot up the listing gallery. If nothing has sold at a cost you can live with, look at the offer and the audience before making more creative; another round of ads rarely fixes a price or a product that isn’t landing. Talk to the people who bought, and to the people who asked a question and didn’t.

A realistic week in this phase, for a one-person brand that is also packing orders:

• Monday: read last week’s numbers and write one sentence on what changed.
• Tuesday: shoot or generate two new openings for the best-selling video.
• Wednesday: send samples or a brief to one creator, and answer messages.
• Thursday: launch the new openings against the current winner.
• Friday: send one email and post the openings that lost as organic content.

What the plan costs you

The plan is built so that most of the spend is ads and samples, not production. The listing set and the first videos come from your own photos, and everything after the first month is reuse. If you make the videos with Blurbly, know that there is no free tier: you can plan an ad and see its shot list before paying, and plans start at $9.90 a week, so you can price a week like the one above before committing to it.

The bigger cost is attention. Ninety days of one change at a time, logged and judged after a week, is slower than launching everything at once, and it is the only way a small brand ends the quarter knowing which two product moments actually sell.