| Episodes | 246 |
| Mentions | 701 |
| Cited here | 52 |
| First — last | #5 — #725 |
| Top guests | Jeri Ellsworth, Ian Danaher, Andrew Witte |
| Related | crowdfunding · open source hardware · arduino · prototype · component sourcing |
Kickstarter is an all-or-nothing crowdfunding platform on which backers pledge money toward a project’s stated funding goal: a card is registered at the time of backing and charged only if the campaign reaches its target, so an unfunded project costs its backers nothing.[27] The platform retains roughly five percent of the amount raised, with a further three to five percent consumed by payment processing, before the creator receives anything.[325] Its value to hardware creators derives principally from the audience already browsing it for projects to fund, an audience that regional and competing platforms have been unable to replicate.[98][87] After early disputes over undelivered projects, the platform publicly defined backing a project as distinct from buying a product and introduced rules — most prominently a requirement that campaigns demonstrate a working prototype.[114][142]
Funding model
A pledge on Kickstarter is not a purchase. The transaction completes only if the campaign reaches its stated goal within its time limit; otherwise no money changes hands.[27] Of the total raised in a successful campaign, approximately eight to nine percent is deducted before the creator sees it — five percent to the platform and three to five percent to the payment processor.[325] A further fraction never arrives at all, because some pledged cards decline or are maxed out in the interval between pledging and collection, and a reserve must be left with the processor against chargebacks; the headline campaign total therefore overstates what is available to spend.[173]
Funds are released only after the campaign closes and a further holding period of about two weeks, so parts must be ordered against the creator’s own money if delivery is not to start weeks late; platforms that release funds as soon as the target is met remove that gap, and the difference in payout timing is a substantive reason projects run late.[184] Long-lead components must be committed to before the campaign even launches if the promised delivery date is to be met, exposing a creator to thousands of dollars of expenditure before any money is raised. On Dave Jones’s μCurrent campaign, a four-thousand-dollar reel of components was purchased in advance for this reason.[175]
Backers are not investors and receive no equity; offering equity in exchange for a pledge would be a securities matter, and the belief that a pledge buys a piece of the company is a persistent misunderstanding among people new to the model.[260] The boundedness of the transaction — a defined product delivered for a defined sum, with no continuing claim on whatever the founders do afterwards — is cited by creators as the reason for choosing crowdfunding over an accelerator or investment. Ian Daniher gave this as the rationale for funding the Nonolith Lab instrument through a campaign rather than taking a stake-based arrangement.[87]
Several campaign parameters, including the receiving bank details, are fixed at launch and cannot be changed afterwards.[177] Only backers can comment on a campaign, so posting a public objection on a project’s own page costs the price of the smallest pledge.[108] Campaign creation was restricted by country for years, tied to the payment processor, so creators outside the permitted countries needed a partner resident in one to take legal responsibility for the campaign; Australians, for example, could not launch directly.[35]
Platform rules and oversight
Following complaints about non-delivery, the platform published a public position titled “Kickstarter is not a store”, stating that backing a project is not buying a product. Rules introduced at the same time required creators to answer what risks and challenges the project faced and what qualified them to overcome them.[114] A working prototype subsequently became mandatory, ending campaigns built on renderings alone; projects unable to meet that bar moved to platforms without it.[142] The prototype rule has been enforced retroactively against large campaigns: one project was suspended after raising about four million dollars, having reached the top tier of funding on the site, on the ground that it had no working prototype.[271] The rule leaves the substantive question open, since a project’s stated purpose is usually to fund the work of making an impractical demonstration practical, and no definition settles how closely the prototype must match the campaign’s claims.[271]
Hardware campaigns undergo more review before approval than other categories, with a nominal approval time of about a week.[176] Review does not scale with volume: a campaign explicitly claiming perpetual motion and infinite energy was reported and returned as not violating any guideline, indicating that reviewers work a checklist of required elements rather than assessing physical plausibility.[434] Whole product categories are excluded by rules that are not obvious in advance — home-improvement devices among them — and campaigns have been rejected under criteria the creators did not know applied to them.[175]
The terms of use require a creator who cannot deliver to make a good-faith effort to refund backers, but the obligation is contractual rather than practically enforceable, and by the time a project fails the money has usually been spent.[278] The refund requirement collides with the sequence of events: money is committed to tooling and parts before the shortfall becomes apparent, so a project in trouble has neither the product nor the funds.[113] Refunds also become mechanically difficult with time, since payment systems stop allowing a transaction to be reversed after a period, leaving individual repayment as the only route.[155]
The platform later partnered with distributors and manufacturing consultancies on a programme intended to reduce hardware failure rates, providing manufacturing guidance to selected projects.[345] That programme is curated and requires a previous successful campaign, so it is closed to exactly the first-time creators whose failure rate motivated it; its practical target is projects already prototyped and close to manufacturable.[358]
Campaign dynamics
The pledge curve of a campaign has a characteristic shape: a large surge in the first days from people already watching the platform, a long flat middle, and a rise at the deadline. Press coverage during the middle produces only small blips, and adjusting the campaign’s presentation does not lift it off the floor.[173] Third-party tools that extrapolate a campaign’s final total from its current rate are misleading precisely because the curve decays; an early projection can be an order of magnitude above the outcome, as on Jeri Ellsworth’s castAR campaign, where a first-day projection ran near ten million dollars.[173]
One of the strongest predictors of a campaign’s outcome is the size of the audience the creator already has, which makes audience building part of the engineering project rather than something separable from it.[151] About a third of pledges come from repeat backers, who arrive after a campaign has visible momentum rather than creating it; the observed failure rate rises sharply for projects that attract almost no backers at launch.[350] The conventional sequence is therefore grassroots outreach to friends, family and interested parties, then a mailing list, then launch, so that the campaign opens with momentum the platform’s own audience can amplify; a campaign posted and left alone is the failure case.[350] No large campaign has succeeded on the merits of the idea alone — each had a deliberate outreach effort behind it, whether or not anyone on the team was called a marketer.[350]
Running the press side of a live campaign is itself a full-time job. On the castAR campaign, Ellsworth’s schedule ran to three interviews or appearances a day in the first week, dropping to two a week and then one a day, on top of a backer correspondence load that did not decline.[173]
The recommended preparation for running a campaign is to back a number of small projects first and follow their updates from the backer’s side, then to ask creators who have run one what they know now that they did not know then.[350] A creator cannot prove in advance that manufacturing will go well, since even experienced teams fail; documenting the process publicly as it happens is the closest available substitute, and openness correlates with lower risk.[350] Backers frequently cannot distinguish a proof of concept from a looks-like or works-like prototype or a production-ready design, so creators are encouraged to state explicitly which stage they are at; even a project in final production can still fail.[350] For products whose function is easy to fake in a video, an unedited demonstration filmed in full from multiple angles is the practical verification, which is why some categories are asked for it.[350]
Publicly promised timelines lengthened as the model matured, with one-year delivery dates becoming common where earlier campaigns had promised months.[325] David Kronstein applied the inverse test before launching the Chronos high-speed camera campaign: he deferred the campaign until the design was essentially finished, asking whether he would be confident selling the product in its current state, inverting the usual order in which money is raised to fund the remaining development.[325] In the model’s early period the platform enforced little, and a campaign could be launched on the strength of being able to make circuit boards, without the accounting of time and cost that the work actually required; Colin O’Flynn describes his own early campaign in those terms.[693]
Economics and failure modes
Almost all of a campaign’s proceeds are already committed to building what was promised, so the sum arriving in the account bears no relation to what the creator has earned.[445] Even a campaign raising several hundred thousand dollars from a few thousand backers can leave its creator unable to say whether it was profitable once building, testing, shipping and supporting the units are counted, and the result is likely to be poor against an hourly rate; Jones has said of his own several-hundred-thousand-dollar campaign that he does not know what profit he made.[682]
Underpricing is endemic: campaigns are priced without accounting for cost of goods, tooling and non-recurring engineering, so a creator can raise half a million dollars and still make nothing, having left no margin.[113] For a given product there is a price and volume combination below which the campaign cannot succeed regardless of effort: at a low enough unit price, a creator paying themselves and their helpers nothing still runs out of money before delivery. Stretch goals move a campaign in the wrong direction on that curve.[215] Because asking a fixed sum in exchange for a fixed reward price removes any later opportunity to correct the numbers, the price must carry a buffer; the advice given repeatedly to first-time creators is to double it.[350]
Outcomes divide into three bands. A small campaign that barely funds is a manageable side project; a very large one funds a full-time company. The middle band — too much work to do alone and not enough money to hire anyone to help — has been called the “pit of despair”.[104]
The most common failure pattern is a team that reaches the last stretch of development, is funded well beyond its goal, and has no idea what to do next, because the remaining work is manufacturing rather than design.[113] Backers will not fund a realistic two-year schedule, so campaigns promise delivery in months; combining an unrealistic schedule with first-time volume manufacturing, direct customer contact and an inadequate budget describes most of the failures.[113] Lateness is the norm rather than the exception — about eighty percent of campaigns deliver late — and it follows structurally from having to commit to a delivery date before the decisions that determine the schedule, such as who will be hired to do the work, can be made.[147] Cash management, not technical failure, is the usual proximate cause of a funded project collapsing.[417]
Over-funding must be planned for as carefully as under-funding: a campaign that raises several times its goal has committed its creator to several times the work at the same margin, and success at that scale is itself the hazard.[438] Both directions of funding error destroy projects, including those run by established companies — too little to meet the commitment, or so much more than expected that spending expands past what the project can support.[282] Scale is no protection: the largest European campaign of its time, the Zano drone project raising over two million pounds, ended in bankruptcy with backers receiving nothing.[282] A frequent post-funding failure is treating the raise as a company-building event and hiring executives and marketing staff against it, a course that accelerator and investor expectations tend to encourage.[209]
Campaign-era hardware should not be shown to later investors once better prototypes exist. On the castAR programme, a series B pitch was made with the crowdfunded units, which most of the firms approached had already seen; the single opportunity each firm allows was exhausted and could not be retried with the working prototypes.[394]
A recurring form of campaign misrepresentation is presenting off-the-shelf hardware and existing firmware as years of original development, including fabricated photographs of development boards. Reselling a commodity product is a legitimate business; the misrepresentation is in the claimed provenance.[221]
Manufacturing and fulfilment
The platform originally had no mechanism for collecting shipping addresses, only a single one-time survey, so a campaign with tens of thousands of backers had to build its own web application to gather and correct addresses; Andrew Witte’s team did this for the Pebble watch campaign.[175] The backer survey can be sent only once, so an error in it or a wrong address supplied by a backer cannot be corrected through the platform. Its export is otherwise usable: results download repeatedly as a spreadsheet, sortable by reward level and by whether an item has shipped.[182]
Capping the number of units a campaign will sell bounds the manufacturing commitment deliberately. Marcus Schappi’s Ninja Blocks campaign capped quantities on the reasoning that the purpose of a campaign is to get the project off the ground, and any quantity beyond that adds risk and strains the assembly partner without changing the outcome.[189]
Orders with a new manufacturer are escalated in stages: a small pre-production run built by exactly the methods intended for volume, evaluated, then a larger order, then larger again as confidence in the supplier’s process accumulates. On Simone Giertz’s product programme the first step was fifty units built by production-identical methods; the failure this avoids is opening ten thousand boxes to find every unit damaged, as happened when a run of finished products was scratched by the racking used to store the boards.[592] Skipping the pre-production step changes the exposure by an order of magnitude: on the µSupply project, a planned pre-campaign lot of forty units for field feedback was overtaken by schedule pressure, so the first several hundred campaign units became the test lot instead.[377]
Contract manufacturers offer end-to-end campaign fulfilment covering sourcing, assembly, programming, test and shipping, with the test fixture either supplied by the customer or designed for a fee, because a defective product reflects on the assembler as well as the creator; Macrofab operates such a service.[243]
Role in product development
Crowdfunding does three separable things: it moves money from the point of sale back to the point where design and parts must be paid for, it performs market research on whether the product is wanted, and it assembles a community with a stake in the outcome. A project that already has money and a community is buying only the market research.[105] The cost of the model is the commitment it creates: backers reasonably feel they have bought the product described, so the ordinary development practice of changing direction as the design teaches something reads as a bait and switch.[105] An unsuccessful campaign is a cheap negative result — if the interest does not exist, nothing happens and no inventory was built — and that property is a large part of the model’s value to a small hardware project.[161]
Where prototypes and small runs can be bought for a few hundred dollars, the funding function matters much less than the publicity, so the reason for running a campaign shifts from capital to attention.[244] A creator who already has an audience can take pre-orders from their own site to the same effect; the platform’s distinct value is to someone with no audience, who gets access to people browsing for projects to fund.[108] Regional crowdfunding platforms existed but did not substitute, for the same reason: a campaign hosted anywhere else is closer to selling from one’s own website.[98]
A completed campaign functions as evidence for later investors in a way a plan cannot: working hardware plus people who have already paid for it converts a pitch from a claim about the future into a demonstrated fact.[268] The same evidence works with institutions unfamiliar with crowdfunding: the transaction record of campaign proceeds arriving is documentary proof of where a business’s money came from, which Jones used when accounting for the origin of his business’s funds.[391] A campaign can also function as a volume purchase rather than a funding event: aggregating demand into one order is what allows a product to be priced at its intended level, with the same items sold immediately afterwards in small quantities costing roughly twice as much.[517] The educational market is reachable for a small hardware vendor provided the administrative burden is accepted, purchase orders included, and provided the product is not otherwise available.[189]
Structuring a launch as a single crescendo, where the product either succeeds or fails on one day, works against the iterative approach that produces good hardware companies; the alternative is shipping something merely good enough early and improving it, rather than perfecting a product before anyone has used it.[402]
Alternative platforms
Indiegogo, the main competing platform, offers flexible funding — the creator keeps whatever is raised whether or not the target is met — alongside an all-or-nothing option. It predates Kickstarter and applies looser rules, which makes it the destination for projects rejected elsewhere, including those displaced by the prototype requirement.[142] Crowdfunding audiences do not transfer between platforms: the smaller ones lack the numbers and the flow of high-profile projects, so the choice of platform materially changes the amount raised for the same project.[87] Running the same campaign simultaneously on two platforms became standard practice, which increases reach and correspondingly increases the number of backers exposed if the project fails.[455]
Crowd Supply, a curated platform for open hardware, treats the campaign as a product launch rather than a funding event: it selects the projects it works with, provides fulfilment and marketing, and continues to stock and sell the product afterwards, so delivery is the start of the relationship rather than its end. It was founded in direct response to the non-delivery problem and to the larger platform’s public position that it was not a store, by people who had built products themselves.[314]
Notable campaigns
The Pebble smartwatch campaign asked for one hundred thousand dollars, enough for roughly a thousand watches, and raised about ten million from tens of thousands of backers, converting a build that could almost have been done by hand into a mass-production programme overseas.[175] A successful hardware campaign need not be large: Michael Ossmann’s HackRF campaign drew about 441 backers and fifty-odd thousand dollars against a sixteen-thousand-dollar goal, enough to make manufacture worthwhile, and continuing sales after the campaign were what enabled him to leave employment.[161] The Nonolith Lab campaign for a measurement instrument aimed at six thousand dollars, enough for a first board run, and tripled it after coverage on hardware blogs and interest from the educational market, taking the team from proof of concept and rendered mock-up through a production run of a few hundred units.[87]
The castAR augmented-reality campaign raised just over a million dollars from about three thousand backers; it was the company’s first outside money and funded its first hires, with the team still working from a house.[394] Keyboardio’s first hardware product committed its makers to about two thousand keyboards with a milled wooden enclosure and individually sculpted keycaps, after visiting roughly a dozen candidate factories and taking bids from half of them.[450] Ken Burns’s single-person project set a ten-thousand-dollar goal and raised about eleven times it, the point at which a project becomes a business and the creator inherits the manufacturing problem.[458] Christina Cyr’s phone project found that paper and cardboard prototypes — the textbook method for validating a product before building it — failed to elicit useful reactions; only a working device assembled from other people’s boards, able to make a call and receive a text, produced people asking to buy it, and the campaign followed from that reaction rather than preceding it.[475]