How do we use incentives to drive product growth?

Tagging @Finance for review prior to vote, please.

Appreciate the positive initiative here but this post leaves me with quite a lot of questions…

First up. Before coming to the forums with a ~$5m a year proposal how much consultation has there been with Treasury/Finance? As a representative of growth & marketing WG, it would have been appreciated to have some opportunity for input.

Are the sums here based off of the recent analysis carried out by AWG (Harmonic Liquidity)?

Is this taking into consideration the future moves to L2’s where depth of required liquidity is reduced?

How does PWG propose to raise the potential $4.8m a year to fund this?

What considerations are we taking together in exploring alternatives to expensive incentive schemes? A deliberately exaggerated hypothetical: what if we spent $4.8m acquiring users by investing in a self-hosted “learn to earn” program? (Evidence suggests coinbase spent just 1% of year 1 revenue on L2E, our collective LM programs to date are >100% in contrast)

Over the last few Q’s there has been a growing unease/perception among many that PWG has a tendency to act with a unilateral mindset, without objectivity and has a somewhat undefined remit. I fear posts like these will only amplify such concerns.

Our collective repeated failures on liquidity to date suggest to me we should be concentrating on agreeing on a solution/compromise there before rushing to commit yet more resources to LM incentives. We know that LM can be incredibly wasteful if not fully understood and correctly managed.

The product pipeline and prioritisation schedules have been previously criticised by many methodologist’s (both internal and external) for lacking objectivity, direction and even bias. I fear a flat $400k price tag on every product will actually create more friction at this point as it will serve as an easy excuse to NOT launch products. The barrier to launch is already very high with many feeling it is too high which is leading to delays, lack of agility, unwarranted risk aversion, and an unwillingness to experiment with new ideas and products.

To be clear I am all for investing in growth initiatives. But there are many issues that need solving before we talk big numbers.

7 Likes

Agreed!

Maybe my message was unclear - if DATA were to be allocated $400k from IC, I would not want to use it on 2 months of LM rewards without careful consideration of how else that capital could be most effectively deployed. I think it is unlikely an LM campaign would be the best use of resources.

4 Likes

One important but tangential point is dropping the community reluctance to indices with overlapping compositions. On the DG1 proposals I’ve noticed questions about composition overlap but if we launch 1 product per month and restrict ourselves to the top 50-100 most liquid tokens then the products will inevitably have substantial overlap. Otherwise we’d be looking at a hard ceiling of 10-20 products. This will be even more acute if we move to new chains with even lower component liquidity. Ideally, the cannibalism would be offset by the new AUM. It will be key to manage Methodologist expectations around overlap though since they wouldn’t benefit from the new AUM.

3 Likes

Fully agreement here. I am particularly keen to dispell the overlap / cannibalisation narrative. TVL = TVL

4 Likes

Really appreciate all the thought and analysis that has gone into this recommendation. You raise all the right questions about how to make sure our products are primed for success right out the gate. As I’ve proposed in IIP-XY: Forming a Liquidity Pod, I’d like to see a group of specialists take ownership of handling liquidity for new products. I envision PWG / Composite Indices Pod being a part of this group.

I will say that I’m not a fan of a standard $400K for every product. It may be the case that we want to throw a lot of weight behind one product, in which case $400K won’t be enough. I’m usually a big fan of simplicity, but in this case I like having some freedom of movement.

6 Likes

There’s a tonne of technicalities that are beyond my current understanding here - however, as someone with strategic new product and service growth experience (consultancy), I have a few questions, based simply on logic. I love the intention here but there seems to be a number of [logical fallacies]

Given the requested budget is 4.8M I don’t see how ‘limited capital’ [red herring] substantiates the argument that methodologists pay seed liquidity. That said, methodologists contributing seed liquidity seems a logical idea - an investment in your own product.

Given the above quoted premises, logic shows that IndexCoop should take the initiative/responsibility to provide at least some portion of seed liquidity, as a sensible investment, especially being this “…cost can be recuperated with minimal loss…”

Q: What split is being considered here?

Arguments in Favor of Liquidity Mining

It is clear that liquidity mining is the preferred (only) strategy presented here. Yet given the amount of savings it has recently provided imo there needs to be a whole lot more supporting evidence as to why this is the best option “amongst others”. Imo, as a business proposition, this proposal seems to favour a product growth strategy of sunk costs over investment.

Given the kind of target assumption of $50M in 2-months plus the complexity and diversity of projects.
Q: What’s the actual horizon here on a hard stop.
Q: What is the likelihood that funding 12 products w. LM return no clear answers?
Q: Has anyone yet considered the ROI on 4.8M

Control Study

From a research perspective (PMF), I would highly recommend PWG take a close look at the success and natural growth trajectory of $DATA. @Thomas_Hepner @Kiba are imo obviously doing something right, there must be lessons we can learn from existing products first. Thomas has outlined a number of learning throughout the forums.

I would be super keen to see a comparative analysis between products that received LM funding DPI, MVI vs DATA. How can DATA growth be optimised ( without LM, yet) to develop a comprehensive control study for these ‘other options’ against which the performance of any future LM projects could be baselined.

Worth reading Liquidity miners: hands of 💎 or 🧻? — Alex Kroeger
.

8 Likes

Hey, @DocHabanero thanks for the post - wanted to give feedback earlier, but alas, here we are. Sorry for the delay!

I think this all hinges on whether we have strong confidence that 2-month LM programs (at ~$400k) can build ~$1.5-2m in sticky liquidity?

$400k in incentives over 2 months would essentially be incentivizing $1m in liquidity at 20% APY for just two months - so we are making the assumption that 20% APY means much to people chasing yield and that that liquidity / growth would remain sticky. Otherwise, you are back to the same problem where your liquidity is not quite effective enough and you have to deploy from your Treasury.

So, if anything, I would be keen to try an experiment of this size with one product before we commit to doing something like this for all future products.

If the argument for incentives (which is really the only reasonable argument I have heard) is that short-term boosts can build hype and buy-in, then I don’t think $400k is getting you that type of reaction, unless it is for just a few weeks.

The Coop should take a stronger position on providing seed liquidity for our products. And we should be ok with slower, organic product growth at the onset. $DATA is a good example - overall AUM is not large yet, but we are seeing increases in growth metrics. We are in the business of creating products that people want to hold because they are great products - we should lean into that.

For smaller AUM products, I don’t think you can effectively gauge PMF when there are incentives involved anyways. PMF will be watching $DATA grow to $10m AUM even if we do no incentive programs.

I am not saying I am opposed to occasional growth experiments, but I believe it would be a mistake to commit to a base $400k in incentives for every product launched. I would like to see us commit to $500k in seed liquidity for every product - focused on managed, concentrated liquidity. If, between the Coop, Methodologist, Wintermute, and others can build ~$1.5m within +/- 50% of the price on a concentrated liquidity pool, then you can facilitate a $25k trade at less than 2% price impact. My point being - there are more sustainable ways to provide effective liquidity at launch.

If we go this route, we eventually find ourselves with $5m in protocol-owned liquidity after 1 year (as opposed to spending $4.8m and being unsure how much we have to show for it) that we are migrating between our products and new product launches. As successful products no longer rely on protocol-owned liquidity, we have assets ready to be deployed into new product liquidity.

Plus, with the improvement of exchange issuance - larger amounts of capital will still be able to enter new products even as liquidity grows on DEXes.

9 Likes

Just chatted to JD about this, correction for the record:
$400k over two months on $1m TVL = 240% APY not 20%

3 Likes

With these kind of numbers coming from an established team and backed by a token like INDEX, I would be pretty darn keen to ape in and then find out more about the product after the fact.

Putting a BD hat on, if we can grow AUM aggressively at launch, we can then pivot to some kind of incentives for providing liquidity through either Visor or Gelato pools. This gives us a means to withdraw some Index Coop liquidity and roll it into the next product. We simply don’t have enough capital to PCL every product for extended durations of time. To do that means slowly ship product.

We collectively are starting to get really good at integrations and creating extrinsic use cases for our products. If we can transition from Index Coop staking incentives, to a burst of LP incentives and then into extrinsic use cases, we will create a really sticky product roadmap. Once we have liquidity and AUM, we can utilise our existing networks/partnership to drive extrinsic use cases.

I’m less set on pinning down a universal budget per product, as I don’t think one shoe fits all. I do like the concept of having a spend estimate from launch to say 6 months post launch. We receive feedback that some folks are seeking more productive assets from us and without intrinsic productive, we must accelerate the time from launch to extrinsic productively. A Go To Market roadmap and budget is key.

Organic growth is amazing! Slow burn organic growth is boring. We operate in a fast paced, short attention span industry. We need to go from zero to hero, fast. We need to capture market surface area, tap into every segment of the market we can and then capture as much capital as possible within that area of the market. In some ways, we minted a governance token to bootstrap a community via bootstrapping a product into success. The more market dominating products we have, the stronger we are as a community.

To me the challenge always has been, how to get from launch to domination mode. Do it once, do it again, then again, just faster and more efficient. That takes skills, planning, hustle and spend.

6 Likes

Thanks for the feedback everyone, and apologize for bad timing while I was out for holiday.

A key theme in the comments seemed to be around the duration of the project.

I propose we do this for $GMI, $LDI and $DATA. The products are relatively diverse and $DATA has already been in the market for a few months so it’s a good indicator of how necessary incentives are.

We know that $DATA has/had roughly 30% APY so is 240%, or 100% appropriate. There are also many different incentive options, such as single sided staking, that we’ve never explored.

I agree with @jdcook’s point that we should be building products people want. But that doesn’t always determine market success. There are crappy products with great branding that beat out better products. I was thinking about this while doing black Friday shopping. People are obsessed with getting good deals, sometimes over buying the best product.

@lee0007 brings up a good point about building case studies and including as much external information as possible. I think that’s a critical part to ensuring we’re collecting and acting on the right data. Your input on some of those parameters would always be welcomed!

@edwardk I agree with your point but it becomes difficult to create criteria for how much money each product should get. I prefer something to the effect of: each product gets up to $400k and if certain thresholds are met we increase of decrease. This comes from optimization. If products don’t hit a certain AUM in a certain time period that is likely an indicator of product-market fit. We can decide to pull back spending at that point.

Note: Seed liquidity will be handled separately and is the main focus of IIP - 110 - forming a liquidity pod. This pod would manage the strategies and execution of capital for this project.

@mel.eth I’d like to run a poll for a vote on allocating up to $400k for $GMI, $LDI and $DATA to test the effectiveness of incentives on growth.

If passed either 1 of the 2, whichever happens first

  1. These products (excluding DATA) will have these terms included in their DG2 snapshot
  2. I’ll work with our internal liquidity experts to create a plan for incentive strategies and request a IIP snapshot
1 Like

Thoughts on Incentives

At this point, I definitely view incentives as a separate tool from liquidity.

MVI and DPI both had really big “pops” in their first 30 days of launch in terms of users. DPI went from 0 to ~4,500 addresses in its first 2 months of its LM campaign; MVI went from 0 to ~1,200 users in its first month when incentives were highest. Incentives did attract liquidity, but arguably, more importantly, they created a critical mass of enthusiastic users.

DATA is growing robustly (60% address growth in November) but did not have a big pop in its first 2 months (we just crossed ~312 addresses on Ethereum mainnet today) like DPI and MVI did. Is that because its not a good product, or because we simply did not offer enough incentives to attract a critical mass of users post-launch? For comparisons sake, both the DATA and MVI launch videos received close to the same numbers of views on Twitter, so why did they not receive similar numbers of investing addresses post-launch? It’s not a certainty, but it’s definitely plausible (maybe likely?) that $400k in MVI incentives from IC vs. ~$13k in DATA incentives from Sushi is the primary driver (to be clear - I’m not pointing fingers or blaming anyone! I’m not upset about this).

Obviously, I am extremely biased here, but running a big $400k LM campaign for DATA is a way we could see if incentives do give products a big pop or not.

Thoughts on Liquidity

Liquidity is closely related to incentives, but is a separate tool to solve the problem: Can users buy the amount of the product they want at a price they are willing to pay?

The reason we get big spikes in total addresses when we launch on Polygon (DATA launched on Polygon less than a week ago and already has ~20% of all DATA addresses) is because anyone who wants to invest less than $5k is pretty much priced out of Ethereum mainnet at $150 per DEX swap.

On the other hand, users who want to purchase >$50k of DATA cannot do it with just $2m of Sushi liquidity. On DATA launch day, I bought ~$63k of DATA tokens from Wintermute (IC’s market maker) who sold them to me at just 50 bps of price impact (~$300). That same value trade on Sushi now has 576 bps of price impact (~$3,600)!!!

Right now, because of liquidity issues, DATA users who want to buy/sell >$20k from a DEX (as opposed to KYC through Wintermute) are priced out of the product.

2 Likes

I will say that I think its a mistake to treat every product as if they should get similar programs.

For one, I think its fair to say products like DPI, MVI, and GMI had better demand signals than most pre-launch

I think in DATA’s case, I am against that level of rewards for a few reasons

  • Aforementioned lack of stated demand for theme / product
  • Structurally inequipped to handle high throughput growth program (staking) due to outdated DEX architecture choice
  • Liquidity mining on that older architecture is wasteful to the end of liquidity
  • Liquidity mining on v3 would be a fine but hilarious use of the funds (and if we’re moving to PCL, what’s the point?)
  • at $2m AUM, the product earns $13,300/yr for the Coop before costs, making back $400k seems pretty tough.
  • DATA’s case uses appeals based on DPI and MVI but fails to recognize that another composite product, $BED (why no retroactive rewards for BED?) is performing better in AUM and users despite $0 in incentives. Part of this has to do with dex choice, ofc.

I would be pretty staunchly against giving a program of similar size to the launch budgets of newer ones, especially if the product doesn’t have the organic traction we’re hoping for. If DATA methodologists were at something closer to 90/10, I think it could make economic sense for the Coop. Otherwise, I would be willing to try something like $40k for a month-long program.

But this $400k rewards program, which I believe you are modeling off of my request for GMI, I think should be given on a discretionary basis at product launches should the product clear some demand signal hurdle.

1 Like

Hey @LemonadeAlpha - totally appreciate your points and value your opinion greatly, but want to clarify a few points you made that I think are unfair or incorrect.

We launched DATA on Sushiswap because they were offering incentives via Onsen and Index Coop and Uniswap were not. We did not know what the incentives were until AFTER the product launched (we were never told by Sushi team pre-launch despite requesting this information multiple times). We had to make a judgement call and I thought having incentives was better than no incentives with greater capital efficiency (UNI-V3.), especially given that were putting our own capital at risk of impermanent loss, and given that IC provided no seed liquidity for DATA either.

To be clear - We are not in favor of launching an LM program for DATA on vanilla Sushi or UNI V2. If we were to do a program like this, we would want it to do it on something like Visor + UNI-V3 or Gelato + Sushi Trident.

This was true of both DPI and MVI before incentives were started.

Some of my response here will sound spicy, but I promise this is not my intention!

Firstly, I don’t think AUM is the right metric. DATA is generating ~2.5x daily revenue of BED for Index Coop.

Secondly, for fairness sake, if we are doing these kind of analysis, we should be doing cohort analysis. BED has been around twice as long as DATA (launched 2 months before DATA on 7/21) and at the bottom of the bear market when BTC closed at $32k and ETH at ~$1.9k, so BED is getting a 2x in TVL vs DATA just off capital gains growth alone. DATA top holdings would have all benefited from similar gains if it had launched on same day (7/21); LINK was ~$15 (now $25), GRT was $0.56 (now $0.92), and FIL was $46.5 vs $53.5 now. BAT did a 3x over that time period and LPT a ~4.5x. Just from that very biased launched day, DATA’s TVL would likely be ~2x greater due to appreciation alone.

No disagreement here! Per this post, users can purchase $30k of BED vs. just $8k of DATA at 1% slippage. Would be more than happy to discuss moving DATA to a Visor + UNI V3 style liquidity solution for improved capital efficiency!

4 Likes

Nothing spicy about your take.

I will say I was pretty surprised the day of launch to see we weren’t doing so on v3. I think a quick look at MVI vs BED liquidity profiles prior to launch would have shown you that v3 is at least 4-5x as capital efficient as the v2 architectures.

Fair point on the LM options presented, think those are definitely solid options but I will maintain that the choice at launch was a misstep which could have given you a better liquidity profile to start (ofc assuming the v3 pool could attract suitable liquidity)

This was true of both DPI and MVI before incentives were started.

This is true but there is no counterfactual while DATA’s is clear - I’ll take that back a bit, I will grant DPI before incentives is at least somewhat similar to DATA, but IC’s recent / current reach is a great improvement on that moment in time.

You’re right that BED has appreciated more in price, but we have a measure called $NSF (net dollar flows) to account for inflows, BED is outperforming DATA on that metric at the same time in its lifecycle (contingent on verification of my belief $DATA’s has a bug on Dune).

RE: revenue, I think you’re absolutely right, but let’s not forget that BED had its fee beaten down by the madness of crowds, with many claiming outright disgust that we would charge a reasonable fee by all customer accounts. In addition, BED has much lower costs to maintain than other products.

In spite of that, BED is a vault for DPI which accrues revenue to IC, so I think its fair to say some of that revenue can be attributable.

It was not a decision taken lightly. We knew about 4-5x capital efficiency of UNI-V3 after research from @overanalyser , @jdcook , @verto0912 , and others when we made the decision.

Would we have decided to launch on UNI V3 if we had known how relatively small the Onsen incentives would be? Maybe. Still, given that all liquidity until the week of launch when Wintermute stepped in was being provided by Titans of Data and Index Coop community members, we did not want to put retail investors/supporters of DATA in the harms way of UNI-V3 impermanent loss (you really should not be doing this unless you are a professional market maker or a automated solution like Visor). Impermanent loss on UNI-V3 is a FAR greater risk for a product like MVI/DPI/DATA than BED (this was the best research we had at the time).

I still maintain no misstep was taken given the knowledge we had at the time. Obviously, we know more and can do more now. We worked within the constaints we had and in hindsight might do things a bit differently given the knowledge we have now.

Worth noting a Uniswap V2 / Sushiswap transaction has done increased 25x from October 2020 to ~$150 now. Small buys of DPI were very feasible then; small buys of DATA not until we launched on Polygon several days ago.

Yes, there is definitely a bug in Dune for the DATA $NSF charts and your counterpoint is definitely correct. BED definitely outperformed DATA in both users and $NSF in a cohort analysis of first 2 months - no doubt about it.

I’m fine with saying another product outperformed DATA in various dimensions - I just want to make sure we’re making reasonable analyses in our comparisons, and I don’t think comparing a 4 month old product on UNI-V3 to a 2 month old product on Sushi is an accurate representation of much of anything.

2 Likes

I guess my point would be that the growth of BED organically has been solid and a counterpoint to the idea that incentives are necessary to achieve a critical mass of users post launch

I also just think its worth noting that BED is the only composite index that hasn’t received nor requested incentives.

A bit unrelated but I am also wondering why DATA is being considered for something like this while BED is not and I suspect will not be - revenue would be a pretty good answer, I concede :sweat_smile:

I would support moving DATA to a PCL pool on v3 - and your points are fair around user safety - and trialing some kind of small incentive program (would be happy to share why I’m hyped about single-sided staking!), but I do think there is something lost by separating the incentive program and the launch hype.

2 Likes

I don’t think it’s necessary. I do think it helps achieve this:

Of course it comes at a significant financial cost. I’m not totally convinced that it’s worth the cost vs. just “slow” organic growth over time (is 60% MoM address growth slow?!) and I don’t think @DocHabanero is either which is why he wants to run an experiment?

Not true - Titans of Data has never requested LM incentives from Index Coop for DATA. They were offered/suggested in this post without us asking for them. This post was my first time hearing about possibility of INDEX LM incentives since post-launch.

That’s a good question. I honestly have no idea. @DocHabanero?

Agree - wish we could go back in time and change this. That being said, we (Titans of Data), are planning a “2nd launch” (stay tuned!) with a video that we could strategically time simultaneously.

I had no part in this decision. Based on our analyses, we believe index products that have a 10 bps mint/ 20 bps redeem fee with a 50 bps streaming fee (this could be even lower for BED) would generate ~50% more revenue and allow us to further lower streaming fees for users by generating revenue from arbitrageurs. DATA passed DG2 with this condition, but launched with standard 95 bps streaming fee due to engineering constaints. We’re still waiting to hear when this can/will be prioritized by IC.

If I recall correctly, the idea to do 10 bps mint / 20 bps redeem may have originally been your idea or you at least inspired me. :smiley:

1 Like

To your point about about product-market fit. This needs to be a better exercise in the product onboarding process. We should feel confident that all of our products at least of a chance at good product market fit. I think this program helps identify that early. As an example, if we expect most products to reach XAUM/X# of wallets/etc in a certain time frame, but a specific product doesn’t reach that, then we remove incentives and think about removing the product.

1 Like

We really just didn’t want to go too far back, and BED has had relatively good traction. If you feel BED should be included we can include the forth product but that ultimately has to be decided by the community.