Kickstarter Campaign Page: Draft the Story and Reward Tiers With AI

Write the campaign page around one promise and one proof (a working prototype), then set each reward tier from real costs: unit cost, packaging, shipping, and the platform's 5% fee plus payment processing. Use AI to draft story, video script and tier wording, but keep the figures, the claims and the honest AI disclosure in your own hands.

Most weak campaigns fail in the spreadsheet before they fail on the page. A lovely story with a $39 early-bird tier that loses $4 per order only creates a bigger problem after it funds. The maths comes first, on one illustrative product, and then comes how to have AI draft each part of the page so the numbers and the words agree.

Follow me on Instagram@sagnikteaches

What Kickstarter asks about AI, and what that means for your drafts

Kickstarter's AI policy took effect for projects submitted from 29 August 2023. In short: AI is not banned, but funded projects must include real human creative input, you must credit and have permission for any artist's work you reference, and you must be specific about how you use AI. The submission questions ask whether the project uses AI to generate output, whether it mainly develops AI technology, and whether you have consent from the owners of works used to produce AI material. Accepted projects show AI components in a "Use of AI" section on the project page. Improper disclosure can lead to suspension, and deliberate misrepresentation can bar you from submitting again. The current rules page adds that excessive AI use, where a project has little or no human thought or involvement, is prohibited.

Connect on LinkedInSagnik Bhattacharya

For a physical product, that gives you two practical rules. First, if AI-generated images or renders appear on your page, say so, and label which images are concept art and which are your real prototype. Second, if AI only helped you draft or edit text, check that your answer to the submission question is accurate, and word it plainly. The rules for design and technology projects matter just as much here: a prototype demonstration must show the product's current state, with no CGI or special effects demonstrating functions that don't exist yet. Reread the rules page before you film or submit, because fee and rule details change.

Subscribe on YouTube@sagnikteaches

Start with the tier maths, not the story

Here is an illustrative product: a foldable desk lamp, made by a manufacturer with a minimum order of 1,000 units. Every figure below is made up to show the method; replace them with yours. The costs per unit are:

• Landed unit cost: $14.50 (factory price plus freight and duty to your warehouse; the landed cost calculator tutorial shows how to build that number).
• Per-order costs: $0.80 box and $2.50 pick and pack, charged once per parcel however many lamps are in it. A bigger multi-unit tier gets a second box ($1.60 in total).
• Shipping cost to you: $6 for one lamp, $8 for two, $12 for four (a domestic assumption for the sum; real rates vary by destination).
• Fees: Kickstarter's platform fee is 5% of funds collected on a funded project, and payment processing adds between 3% and 5%. Its fees page for USD projects listed processing at 3% plus $0.30 per pledge in October 2026, with a discounted micropledge rate of 5% plus $0.08 on pledges under $10; each country's page differs. I'll model a flat 8% plus $0.30 per pledge. Shipping collected during the campaign counts towards your goal, so apply the fees to the total the backer pays, shipping included.

The tiers, with shipping charged separately at what it costs:

"Build and pack cost" is the unit cost times the lamps in the tier, plus the box and pick-and-pack charge. For the early bird: $14.50 + $0.80 + $2.50 = $17.80. "Left over" is what remains to pay for everything that doesn't scale with the order: tooling, certification, video, ads. Per lamp, the four tiers leave $17.30, $26.50, $24.32 and $23.03, so the early bird is the weakest and the standard price the strongest, and the bigger packs earn slightly less per lamp because the discount outweighs the saved packing.

From tier margins to a funding goal

Assume the fixed costs are tooling $12,000, certification $1,500, video and ads $3,000, which is $16,500. Now assume a backer mix: of 1,000 backers, 300 take the early bird (a limited slot), 450 the standard, 150 the pair, 100 the four-pack. That mix pledges $70,900 in total and leaves $33,623 after fees, manufacture and shipping, which is 47.4 cents of every pledged dollar.

Fixed costs of $16,500 divided by 0.474 give a break-even of about $34,800 pledged. Round up to a $35,000 goal and the campaign pays for itself if the mix holds. Two things the sum shows that a goal chosen "because it feels achievable" never would:

• Minimum order trap: with a 1,000-unit minimum, you'd need about $48,900 pledged at this mix to sell 1,000 lamps. At exactly $35,000 you'd sell around 716, leaving some 284 lamps paid for and unsold, about $4,100 of stock at landed cost. Either negotiate a lower minimum, lift the goal, or have a plan (pre-orders, retail) for the surplus.
• Early-bird depth: the early bird leaves $17.30 a lamp, $9.20 less than the standard tier. If 600 backers grab it instead of 300 because it was too generous, and those extra 300 would otherwise have paid the standard price, you earn about $2,760 less than modelled. Cap the slots and say the cap on the page.

Build this in a spreadsheet with the mix as input cells so you can try an optimistic and a gloomy mix. If the numbers go in a tool rather than your head, an assistant can help build the sheet and check the formulas, as covered in the ChatGPT in Excel guide.

Get AI to stress-test the tiers

Once the table exists, use AI as a sceptical reviewer, not an author. Paste your inputs and ask it to find holes. Use the prompt below.

I'm launching a physical product on Kickstarter. Here are my costs and tiers (table below).

Task: find problems, do not rewrite the page.
1. Recalculate each tier's leftover per order and show your arithmetic.
2. Flag any tier where shipping charged is lower than shipping cost.
3. Tell me what happens to the funding goal if (a) landed cost rises 15%, (b) 5% of pledges fail at collection, (c) shipping to the furthest destination is double.
4. List what I have not included in costs (returns, replacements, customs, platform fee on shipping, currency).

Assume fees of 8% + $0.30 per pledge on the full amount including shipping.

[paste table and fixed costs]

A typical reply (illustrative) confirms the leftover column, notices the four-pack ships in two boxes and asks whether $12 covers it, and points out that a 15% rise in landed cost takes $2.18 off every lamp, so the break-even goal moves from about $34,800 to roughly $38,400. Check its arithmetic yourself: models are decent at suggesting which scenarios to run and poor at carrying a long sum without a slip.

Shipping by tier, and the destination problem

Kickstarter lets you set shipping by destination for each reward. The mistake is to set one figure to "everywhere else". Pull actual quotes for three or four destination groups from your fulfilment partner and enter each. For the lamp, an illustrative table might look like this:

Taxes and duties need their own line on the page. Kickstarter's guidance is to keep sales taxes out of reward prices; if you use its Pledge Manager, it collects tax on eligible pledges after the campaign ends. Customs duties are separate: under DDP terms you or your fulfilment partner pay them up front, and under DDU the backer pays on arrival, which Kickstarter's fulfilment guide warns often frustrates backers. Say on the page who pays what, and ask your accountant before launch. For the delivery side, 3PL pricing logic is in the 3PL pricing calculator tutorial.

Draft the story with a brief, not a blank page

AI writes a flat, interchangeable story if you ask for "a compelling Kickstarter story". It writes a usable one when you hand it the raw material. Collect these first, in your own words, even as rough notes:

• The moment you noticed the problem (one real scene).
• What existing products do wrong, in specifics.
• What your prototype does, with one measurable detail (it folds to 2 cm; it runs 8 hours).
• What the backer receives and when (month and year, with a buffer).
• The risks you actually face (tooling, certification, shipping).

Write the "Story" section of a Kickstarter page for a physical product.

Facts you may use (nothing else):
- Problem: [one real scene, two sentences]
- What competitors do wrong: [three specifics]
- What our prototype does: [measurable details]
- What backers get and when: [tiers and delivery month]
- Risks: [list]

Rules: first person plural, plain words, short paragraphs, no superlatives, no promises I haven't listed. End with a "Risks and challenges" section that uses my risks. Mark any sentence you are unsure is supported by my facts with [CHECK].

Before and after

An illustrative first draft from a vague prompt: "We're revolutionising the way you light your workspace. Our award-winning, patent-pending design will transform your productivity." Nothing in it can be checked, and "award-winning" and "patent-pending" were invented. After the brief above, the opening reads: "Our desk lamp folds to 2 cm so it can live in a laptop bag. Every folding lamp we tried either wobbled on a hinge or needed a separate power brick. This one has a weighted base and runs eight hours on one charge." Same product, but every sentence is something you can prove. Delete any adjective you can't test.

The video script: 95 seconds, five beats

Keep the video short, and assume the first 15 seconds do the most work. Ask AI for a shot-by-shot script, then film it yourself with the real prototype.

1. 0 to 10 s: the problem on screen, one line of voiceover.
2. 10 to 30 s: the product in use, hands visible, no render.
3. 30 to 60 s: how it works, one demonstration of the measurable claim.
4. 60 to 80 s: what you have built so far and what is left (tooling, testing).
5. 80 to 95 s: the ask, the tiers in one sentence, delivery month.

Turn these five beats into a 95-second video script with a column for voiceover and a column for shot. Voiceover: 150 words maximum. Only show what exists today: if a shot needs something we haven't built, replace it with a drawing and mark it "concept, not final". Facts: [paste].

Check the output for shots that need things you don't have, such as "the lamp glowing in a bedroom". If it's a render, label it as one, and if it was made with AI, put it in your Use of AI answer. A render can illustrate a finished colour or finish, but under the rules it can't demonstrate a function the prototype doesn't have. Keep an eye on the same trust issues covered in the tutorial on AI mistakes that hurt trust.

Writing the tier descriptions

Each tier needs a name, a price, what's in the box, a delivery month, and a limit if there is one. The common AI error is vague bundles: "everything you need to get started". Give it the box contents and ask for one line per tier. Illustrative output after a good prompt:

• Early bird, $39: one lamp, USB-C cable, carry pouch. Limited to the first 300 backers. Estimated delivery: March.
• Standard, $49: the same, no limit.
• Pair, $89: two lamps and two pouches; one cable per lamp. A saving of $9 against two standard.

Check that "a saving of $9" is arithmetic you've verified: two standard lamps are $98, so $89 saves $9. If it says $10, correct it. State the delivery month with a buffer you believe; the Estimated delivery date you pick is a promise the backer will remember.

The tier that lost money

An illustrative failure. A creator sets four tiers and a single shipping figure of $10 "to keep it simple". Lamps in the four-pack are cheap to send per unit, but backers in a distant market pay the same $10 as backers next door, and the actual cost to them is $36. A fifth of the four-pack backers are overseas, and shipping alone swallows $26 each. The fix is a per-destination table before launch, not a plea to backers afterwards. In general, run every tier against your most expensive destination before you publish it.

Other places AI helps, and where it shouldn't

Good uses: a FAQ draft from the questions you expect (warranty, power, delivery, who pays duties); a timeline wording that lists tooling, certification, production, shipping, each with its own date and a buffer; update posts; replies to backer comments in your voice for you to edit. Poor uses: stretch goals (you need real costs behind each); any statement about certification you haven't got; a testimonial, a review count or a press mention. Never have AI invent social proof.

When the campaign succeeds, the supplier brief and inspection step matter as much as the page; the supplier negotiation tutorial covers lowering a minimum order before you launch.

Run a gloomy backer mix before you trust the goal

The $35,000 goal above assumed the mix of 300, 450, 150 and 100. Backers rarely behave to plan, so change the mix and watch what moves. Suppose the early-bird slots sell out in the first day and then most late backers choose the single standard lamp, with few buying packs: 300 early bird, 600 standard, 40 pairs, 10 four-packs. The pledged total is 300 × $45 + 600 × $55 + 40 × $97 + 10 × $181 = $13,500 + $33,000 + $3,880 + $1,810 = $52,190. What's left after costs: 300 × $17.30 + 600 × $26.50 + 40 × $48.64 + 10 × $92.12 = $5,190 + $15,900 + $1,946 + $921 = $23,957. Minus fixed costs of $16,500 that leaves $7,457, from 1,020 lamps, just over the minimum run. That still works, but notice what happened: with only 50 fewer backers than planned, a mix with fewer packs earns about $9,700 less.

The lesson is that the funding goal should come from the weakest mix you consider plausible, not the average. Run three mixes (hopeful, expected, gloomy), then set the goal at the gloomy mix's break-even if you can bear the number, or add a pack incentive if you can't. A pack incentive might be a $4 saving per lamp on the four-pack, but only if the sheet shows the tier still leaves more than the single lamp does on a per-order basis.

Failed payments and other quiet leaks

Pledges are collected when the campaign closes, and some cards fail. Backers then have seven days to fix a failed payment before the pledge is dropped. Kickstarter takes fees only on what it collects, and you still receive the collected total even if dropped pledges take it below your goal. I'd model a few per cent loss as an assumption, labelled as yours, rather than a figure from anywhere. On a $35,000 goal, 3% failure removes $1,050 and takes the funded total just below the $34,800 break-even if you funded at the line. Other leaks to put in the sheet as their own rows: replacement units for damage in transit; a small percentage of unreachable addresses; currency movement between pledge and payout if you price in one currency and your supplier bills in another; and your own time, which is easy to leave out because it isn't invoiced.

A filled-in FAQ for a physical product

Backers ask the same eight questions on almost every hardware campaign, so draft them before launch and let AI tidy your wording. A filled example (illustrative, for the lamp), keeping the answer only to what you can promise:

• When will it ship? "We plan to ship from March. Tooling takes eight weeks, and we've added four weeks for testing and four for shipping. We'll post an update at each stage."
• What if it's late? "We'll tell you in the next update, with the reason and a new date. We can't promise dates we don't control, but we can promise not to go quiet."
• Who pays customs charges? "Charges depend on where you live and are set by your local authorities. They're not included in your pledge. Check before you back if you're unsure."
• Is there a warranty? "Twelve months from delivery against faults in manufacture. Write to us with your order number."
• Can I charge it from a laptop? "Yes, from any USB-C port. We'll publish the measured charging time once the production samples have been tested."

The brief for AI is the same as for the story: give it the answers in rough notes and ask it to turn them into plain sentences without adding any promise. Then read each line asking "can I afford to be held to this?" Warranty periods and delivery dates in an FAQ are things backers screenshot.

Update posts: the ones to write before you need them

During a campaign you'll write a few updates quickly. Draft templates in advance with AI: a "funded" message that thanks backers and states the next three steps with dates; a "halfway" message that shows what changed and what hasn't; a "shipping delay" message built to the pattern what happened, what it changes, new date, what you're doing. Keep numbers in all of them, such as "samples passed testing on 3 of 3 units", only when they're true. A backer who reads a specific, early bad-news update usually stays; a backer who reads a vague late one asks for a refund.

Pre-launch check, in the order to do it

1. Landed unit cost, with a 10 to 15% cushion.
2. Tier table with fees on the shipping too.
3. Goal at or above break-even, minimum order checked.
4. Shipping table by destination group.
5. Story written from your facts, adjectives you can't prove deleted.
6. Video shows the real prototype; concept art labelled.
7. Use of AI answered honestly and matched to what is on the page.
8. Risks section written by you, from your real risks.
9. Delivery dates have buffers and your supplier has agreed to them.

Questions before you launch a physical product

Do I have to tell Kickstarter I used AI on my campaign page?

Kickstarter's AI policy asks every creator whether the project uses AI to generate output, and projects with AI-generated content show a 'Use of AI' section on the page. Using AI to help draft copy is not banned, but answer the submission questions honestly, because undisclosed use can lead to suspension. Read the current rules before you submit.

Should I include shipping in the reward price?

Usually not. Kickstarter lets you set shipping per destination, so each backer pays what delivery to them costs. Hiding shipping in the price makes nearby backers overpay and distant ones underpay. Whichever you choose, remember fees are taken on the shipping you collect as well.

How much should I set the funding goal at?

Set it at the amount you need to build and ship the minimum run at the prices you've offered, then add a cushion for failed payments and rising costs. Funding is all or nothing, so a goal you can only just meet is safer than one that leaves you making a loss on success.

When do I get the money?

Only if the goal is met, and not instantly: funds are released after the campaign ends and pledges have been collected. Plan your supplier deposit around that gap, not around the last day of the campaign.

Further reads

• Pre-Order Policy Checklist: Draft Yours and Check Delivery Rules — Pre-order rules for selling on after the campaign.
• Importer Product Compliance Checklist: AI Flags Missing Documents — Check the paperwork a first production run needs.
• Pre-Shipment Inspection Checklist: AI Drafts Your Inspector Brief — Inspect the first batch before it ships to backers.
• Shopify Pre-Order App: Which to Pick, Then Set Ship Dates With AI — Keep selling after the campaign closes.
• How to Write Product Descriptions With AI That Actually Sell — Turn the campaign copy into a lasting product page.
• AI Product Photos vs a Photographer: Cost and Quality Compared — Decide who shoots your campaign images.
• Alibaba Sourcing With AI: Vet Suppliers and Plan a Sample Order — Find and sample a manufacturer first.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Kickstarter's AI policy announcement (August 2023) and AI guidelines; its rules page (AI use, prototype demonstrations); its fees page (5% platform fee, processing and micropledge rates); help centre articles on shipping options and dropped pledges; its guides to tax collection and fulfilment. Fees and rules change, so check Kickstarter's own pages before launch.