Crowdfunding rewards: how to optimize your tiers
Should you bundle your rewards? If so, what's the optimal strategy to use?
As an indie comics creator, I’m forever thinking about how to best optimize my crowdfunding campaigns, and one of the goals of this publication has been to dive deep into the scientific literature to learn about what works (and what doesn’t).
A key, and under-researched areas, has been the rewards section for sites like Kickstarter and Crowdfundr. This is where creators get to set their pricing structure for the campaign, and ultimately this may be where success and failure is realized. Sure, the other elements are important too. If you’re trying to sell a comic and don’t have completed pages to showcase, then I’m going to be hesitant about backing you. But for the most part, these are often in place, and there’s a compelling campaign story, videos, etc. So what else can separate your project from another?
The price.
As I tried to argue last time, the data suggests you need to think about reward pricing in relation to the total goal you’re trying to raise. This may mean having slighlty higher prices than other (similar) campaigns, if that’s what you need.
But another critical factor that’s relatively unique to crowdfunding, is the ability to bundle rewards together to create reward tiers. If you have backed any crowdfunding projects, you’ve probably seen such bundles, some of which may have been of interest, and others of no interest to you at all.
I often struggle with my own campaign bundles. It can be hard to decide how many tiers to include, and what price points you should set them at, so I dug further into the literature to see if I could find some advice. There’s actually a dearth of research in this area, which is not surprising, as it’s technically quite challenging, but I did find a very promising paper - Modeling Menu Bundle Designs of Crowdfunding Projects - which tries to address these questions.
I’m not going to lie to you; this is a very complicated paper to read, and I had to go through it several times to really appreciate their methods. The key results though are very useful for creators to consider, so I hope you’ll stick around.
The authors basic premise is that they’re trying to model reward bundling. What is reward bundling, I hear you ask. Well, if you’ve ever used a reward-based crowdfunding platform, you’ve already got some experience with this. These platforms offer rewards which are arranged in reward tiers: typically arranged on-screen from cheapest to most expensive.
The two main decisions for creators when building their reward tiers are: 1) what does the tier include?, and 2) what is the price of the tier? The paper refers to these as bundling (what is in the tier) and offering (what is the price of the tier).
If you’ve never used a crowdfunding platform, let me give you an example of how it works. Say, I’ve written a book. I decide to offer an electronic PDF of the book as one tier - since there are few costs associated with this, my instinct is to make it cheap, so I’ll set the price at $5. For my next tier, I want to offer a paperback version of the book. As I have to print and ship the book after the campaign, I need this tier to be priced at a higher level, so I’ll set the price at $20.
This is exactly how it works in online stores.
However, in crowdfunding, I typically want to make the higher priced tier more attractive to backers, so I won’t just offer the paperback book at $20, but I’ll also include the electronic PDF. This is then a bundled reward rather than a single reward.
If I keep going, I could then imagine offering a higher-priced bundle of a hardcover book plus PDF at $40, BUT I could also offer a hardcover book plus softcover book plus PDF at $45 to target collectors. The reward tier price goes up, as does the number of items in the bundle, but you can see there is also extra value for backers.
The question for me is whether to offer the $40 or the $45 tier? Or both? The latter sounds great, but if I have another five or six reward tiers, each bifurcated in this manner, I quickly end up with a mass of rewards that become incomprehensible to backers - and that is likely to negatively impact my campaign.
Returning to the paper, what the authors want to know, is whether the choice of price (offering) and the choice of what goes into the bundle (bundling) has an impact on the success of a campaign.
Spoiler alert: yes, it does.
The authors analyzed a small dataset of around 15,000 Kickstarter projects between January 1st, 2014 and June 30th, 2014, which had almost 150,000 reward bundles between them. A review of all rewards allowed them to be categorized by around 11,000 words.
General themes arising from analysis of the data include:
On average, successful projects offer more bundles than unsuccessful projects.
Projects with bundled rewards tend to have a higher success rate than those without bundles.
The more items in a reward bundle, the less likely it is to be selected. However, when comparing bundles of the same price, more items led to more backers.
One reason for the above findings is likely perceived backer value, which increases as projects offer more bundles, with more rewards. It’s likely that offering more bundles means that backers will pledge more funds, ensuring those projects are more likely to hit their goals. Similarly, bundled rewards offer more value to backers than individual rewards, thus bundles are preferred. Finally, a high number of items in a bundle typically means a high price-point, which may simply be too high for some backers. However, if the reward cost is held constant, more items increases the perceived value, and therefore backers would be more likely to back bundles with a higher number of items.
But now we get into the real meat of the paper, as the authors construct a Menu-Offering Bundle (MOB) using Bayesian analysis to construct distribution probabilities for several variables, such as offering topic, bundling topic, project category, bundle price, word variable etc. This is a complicated model, which is too mathematical to be meaningful for our discussion. However, what is important is that the authors compared their MOB model against other mathematical models to best model reward cost, and reward type, and found it to be superior.
Having built a model, they can now try to answer their initial question hypothesis. Basically the MOB model generates relevant groupings from within the data that allow for deeper analysis.
Offering topics
Recall the offering is basically the price (or cost) of a particular tier. The MOB model actually identifies a menu of ten reward tier prices - not just a single reward - which is shown in the figure below. There are ten different menus identified, which are listed according to statistical weight rather than dollar amount.
For example, offering topic number 4 basically has rewards of $0, $5, $25, $50, and $150. Information about the nature of the rewards is not included, it’s just the cost. However, such a combination of rewards is often seen for products such as book and comics, which typically don’t have high-priced rewards.
Bundling topics
The MOB further generates a menu of ten bundled items (again arranged by statistical weight) and shown in the figure below.
It’s important to note that these menus are constructed from the entire project dataset, and not broken down by different categories. Thus, for example, it’s hard to see how an author selling a book would also be able to offer a producer credit (c.f., #4 above). However, it provides a useful understanding of how different items are grouped together.
Interactions
Neither of these menus exist separately of course, as they need to interact with one another, and this is where things become interesting. The interactions identify the most common associations for a specific set of outcomes.
Success rate
The pairing with the highest success rate is Offering menu #5 and Bundle menu #7 (which has a success rate of 80%).
The reward prices are grouped at the lower end (under $100) with one reward as a high-priced outlier. The reward words here suggest Free (likely a $0 award) and Early Bird Access rewards (with one highly exclusive award being the outlier). The authors suggest such interactions are typical for art and design campaigns, which rings true from my experience using the platform.
Interestingly, analysis shows Offering menu #3 with Bundle menu #7 has a success rate of only 40%. Why would this be? Well, the reward tiers are now spaced at: $50, $75, $100, $250, $1500; the outlier tier is much cheaper than the one from menu #5, but the lower tiers are far more expensive. The authors suggest such rewards are common in games and technology projects.
What this suggests, is that for a given grouping of reward items, there is an optimal reward cost. The challenge for creators is to try to best match their proposed rewards with the right pricing structure.
Backers
The optimal combination to generate the highest number of backers is Offering #6 with Bundle #3 i.e., tiers at $10, $45, $200, $1000, and $10,000 with words: Thank, Name, Invitation, Everything, Party. This suggests the items would be access to some exclusive party, but the authors provide an example of a project using this specific combination.
One of the project using this strategy is a publishing project, Artificial Intelligence for Humans. It offers12 rewards ranging from $1 to $250, with most prices around $10 to $50 and 11 out of the 12 reward bundles include download of ebooks. Due to the affordable price range, the project is able to attract backers to all of the offered reward bundles, with the most popular reward bundle being supported by 146 backers.
For those of you who don’t have any reward tiers that approach $10,000, the authors of the paper note that the second most optimal combination for backers was Offering #2 and Bundle #5. This means prices: $40, $50, $100, $250, $500, with words: Thank, Choice, Monthly, Sticker, Package. This wouldn’t be out of place for some arts projects.
Design suggestions
As I already alluded to, the challenge for creators is to optimize the alignment between bundles and reward levels. The authors attempt to show how this works by comparing successful and unsuccessful campaigns for projects with similar offering menus. They argue that the unsuccessful projects had shorter reward lists, over narrower price ranges, compared to the successful projects.
They then go on to examine the same situation for projects with similar offering menus, but with different bundle structures. The unsuccessful project had fewer bundles, which were not actually the product (but experiences), and the bundles did not add additional items from previous levels.
This final section reinforces some of the initial findings: namely that more rewards, and more bundles, are more strongly associated with success. However, it also suggests that reward tiers should have a large range of prices (to appeal to a wide variety of backers) and should also be hierarchical, including all the rewards from the previous tier.
Returning to my book example, this suggests the hardcover tier should build upon the existing tiers, thus I should have:
$5 PDF
$20 Paperback (plus PDF)
$45 Hardcover (plus paperback and PDF)
Which suggests a strategy based on offering topic #4 and bundle topic #4, where the high-cost reward is $150 (which could be something like):
$150 Exclusive one hour zoom with the creator (plus hardcover and paperback and PDF).
For most of us who regularly use crowdfunding platforms, the above may not seem like important information - it’s just how most people seem to do it. I hope this analysis has helped show why this strategy seems to work, and how it can be helpful when it comes to designing your next campaign.
What do you think? Did this resonate with you? What other questions do you think are important when it comes to rewards and tiers? Let me know in the comments.