CIP-74 Retrospective: Aligning Rewards with Revenue

General
1

CIP-74 Retrospective: Aligning Rewards with Revenue

Following the introduction of the 2 bps volume fee, the Core Team has completed a 6-week analysis of the protocol’s economic health. Per our mandate to iterate on these fees, we are now sharing our findings and the subsequent adjustments.

The primary objectives of this experiment were, as follows:

  • Incentive alignment: Transitioning to a dynamic reward cap for solvers that is directly tied to fees generated. This protects the protocol from losses on small batches while uncapping solver upside on high-value, high-fee auctions

  • Toxic flow mitigation: Filtering out low-value “surplus farming” (e.g. 1:1 correlated asset trades) that adds volume but loses money for the DAO, and increases risk for solvers

  • Economic sustainability: Ensuring positive gross margin on every network and eliminating cross-network subsidies

  • Direct monetization: Validating the impact of a transparent volume fee

Overall, we see the experiment as a qualified success - particularly in relation to the four objectives listed above. However, CoW Protocol volume dropped in this period by more than expected, and we received valuable feedback from partners that is leading us to recommend lowering volume-based fees on correlated asset pairs going forward.

**
Unit Economics & Profitability**

The experiment was a success for protocol health. CoW Protocol moved from being profitable only on trades over $100k to achieving profitability on trades as small as $1k. Although solver rewards increased and price improvement fees dipped, our overall revenue line remained stable - meaning we are now earning more from a broader range of trade sizes. This was parallel to a Fee Rate (Protocol Fees in USD / Volume USD) improvement from ~2bps to 5.6 bps.

Volume Impact & Market Realignment:

As anticipated, volume dropped following the fee implementation. The data reveals that this was largely the result of “pruning” low-margin correlated asset bots, particularly on L2s like Arbitrum and Base. While the raw volume is lower, the quality of the remaining flow is significantly higher.

Additionally, DEX Aggregator volume dropped significantly in December due to a broader market contraction. Focusing on the most relevant networks, we saw significantly lower volumes on Arb1 and Base compared to the broader market, and less volume for variable token pairs on Mainnet. (Remember that some of this was expected due to the intentional reduction in activity from correlated asset bots).

Transparency vs. Obfuscation of Fees in Quote value

In line with the CoW DAO’s commitment to transparency the volume fee has been explicitly displayed on the interface rather than being “hidden” within the price improvement /surplus picking. While this led to a perceived drop in competitiveness on aggregators like DefiLlama for correlated assets, we believe radical transparency is the superior long-term strategy for user trust.

Solver Ecosystem Health

It has been observed that the new model puts pressure on solver ROI for small orders. Some solvers noted that rewards on these trades sometimes fall below their operational overhead. While the protocol’s revenue remains stable, a slight tightening in gross margin is still posible to ensure our solver landscape remains diverse and competitive.

Next steps

To regain competitiveness without inviting toxic flow back, the Core Team will exercise its mandate to implement a tiered fee structure starting February 3rd, 00:00:00 UTC:

  1. Correlated Assets: The volume fee will be reduced from 2 bps to 0.3 bps for all correlated asset pairs, including stable-to-stable token pairs.

  2. Dynamic Registry: We are launching a dynamic list of “Reduced Fee Tokens” (including major Stablecoins and RWAs) based on price volatility whose volume fees will also be reduced from 2 bps to 0.3 bps.

Thank you and we welcome comments.

5 Likes
CIP-74: Align Solver Rewards with Protocol Revenue and introduce a volume based fee
2

It is disappointing that this is the case and that these broad volume fees being charged, whereby this is not related to the quality of solution, or more intrinsically aligned with the effort being expended by the protocol and/or solver.

tl;dr: There is a disconnect between the fee charged to the user, and the effort expended when solving.

Additionally, where is this dynamic registry and correlated assets? I’d expect that this should be notified to the market - timely communication for something that is already in effect.

3

@mfw78 , think you might have disregarded the topic of “toxic flow mitigation”? There is a very clear reason for a flat volume fee: To avoid consumption of DAO resources (in rewards, infrastructure, etc) from toxic flow.

Also you might be undervaluing the effort solvers input even in simple swaps (e.g. even deposit actions), which still need infrastructure, leads to gas fees and token harvesting, solver rewards administration, etc.

My concrete suggestion here is: Since you disagree, what would be your counterproposal to this second testing phase? Given the topic that 0 volume fee would invite gaming of correlated assets, leading to losses to the DAO and extra risk to solvers.

Regarding the token list, I think it’s a fair comment - will see in case the list can be made public.
Thank you for the reply.

4

The element of gas fees is irrelevant as these fees are already noted to the user, and are in addition to the protocol fees that are being charged, eg:

image
image527×585 35.6 KB

In regards to toxic order flow, there is likely a more broad topic as this is going to impact on things like same input token as output token (I’m aware that this functionality is currently in development). It’d be pretty silly to have the same input token as output token and charge a percentage fee on this, when essentially this is a gas fee and should not exceed such.

I’d sooner see appData sandwiched in so as to flag a “like for like” (in which case can have a condition that input must equal like output in number, therefore stopping the 1:1 correlated pair arbitrage, which would still have the percentage fee attached to it if the amounts didn’t correlate 1:1). The primary concerns for this flat rate bps fee is for all swaps between USDC / USDS / DAI which may be done for only gas fees, but yet charging any bps on this makes CoW Protocol non-competitive when Spark Finance can do it for free (gas fee only). It would be an annoying flow for anyone moving reasonable money around to have to leave CoW Swap if they are using it in any treasury management capacity, purely to do this aspect of a leg, then to switch back.