Velora & Cowswap Integration Analysis

Introduction

This publication is intended as a quantitative analysis highlighting the impact of a 15 to 25bps Partner Fee implemented on Orderflow through Cowswap via aave.com, a domain owned by Aave Labs.

The objective is to contextualise the mechanical consequences it has on Aave users, broader market efficiency, Aave DAO and Aave Labs revenues.

Aave Users

At the time of writing, the stablecoin borrow rate on the Core instance is 5.21% USDT, 4.90% USDC and 4.73% GHO. With a strong GHO peg, the GHO Stewards are actively encouraging users to utilize GHO debt. The 48bps spread between USDT and GHO encourages users to migrate from USDT to GHO debt.

Screenshot 2025-12-14 at 14.32.39

For each unit of USDT debt that migrates to GHO, the users’ borrow costs are reduced by 48bps, and the Aave DAO’s Revenue increases from 52.1bps to 473bps per unit of debt (x9.07) due to the Reserve Factor increasing from 10% to 100% and the GHO borrow rate. When users perform the swap via the Aave.com frontend, the borrow cost reduction is offset by the following:

  • Price Impact on the swap;
  • Cowswap Fee; and,
  • Partner Fee.

The price impact for a 1M GHO to 1M USDT is ~0bps, and the Cowswap Fee for all swaps is 2bps, which offsets 15 days of borrow interest savings, equivalent to 15 days of payback period.

telegram-cloud-photo-size-4-5911451390192585486-y

The stablecoin Partner Fee adds an additional 3.5 to 4 months payback period to recover the 15bps charge (15/48x365days), extending the total payback period to 129 days.

This excludes gas, which is most impactful on small user holding balances and assumes the user is migrating from variableDebtUSDT to variableDebtGHO swap via debt_swap_adpter.

The 48bps borrow rate discount migration payback period:

  • 15 days payback period to recover 2bps Cowswap Swap Fee;
  • 114 days payback period to recover 15bps Partner Fee; and,
  • 129 days total payback period.

The frontend swap feature offers an improved UX experience by allowing them to easily navigate rebalancing an active position on Aave Protocol.

Notably, the additional fees friction is unlikely to affect more sophisticated users who can access alternative tooling. The user group most affected are those without the ability to bypass the fee, or are suitably captive.

The linear structure of the Partner Fee can present challenges when managing larger positions, as the absolute USD cost becomes more significant, particularly for fee-sensitive stablecoin users. That said, linear fee models remain common across many DEXs and aggregator platforms. As a result, participants deploying substantial amounts of capital may seek to minimize direct swap activity and instead explore alternative mechanisms for transitioning between assets. One such approach involves routing through peg stability modules, which can offer fee-free or lower-cost conversions, as seen within ecosystems such as Sky and Curve.

The table below compares the USDC to USDT swap fees across various destinations:

Swap fees comparison (+25bps) (1)
Reference: Cowswap Fee, Velora Fee, Aave Flashloan Fee and Velora Flashloan Fee Waiver

The swap fee of 0.775bs was calculated by taking an average swap fee for stablecoin swap across 0.5bps Balancer Swap Fee, 1bps Curve Swap Fee, 0.8bps Uniswap Swap Fee and 0.7bps Fluid Swap Fee whilst taking into consideration that Balancer is transitioning to 0.06bps Swap Fee.

In the context of the above, stacking DEX, Aggregator, Flashloan and Partner Fees has a notable impact on the overall fee size being incurred by users.

GHO

When GHO has deviated from its peg, increasing the borrow rate encourages users to migrate from GHO to a different debt asset. With the introduction of the Partner Fee, users repaying GHO debt via the swap feature are less likely to repay the debt due to the lengthy pay back period. The buy pressure on the peg resulting from users repaying GHO debt helps strengthen the peg. With the introduction of the Partner Fee, users are less likely to migrate debt via aave.com due to the lengthy payback period. The Partner Fee somewhat impairs the Gho Stewards’ ability to maintain the peg as a portion of the user base is discouraged from arbitraging borrowing rates. Going forward, users need to be accutely aware of the implications on the Borrow Rate for GHO as larger premiums are likely to be required to entice growth and contraction in GHO’s circulating supply when managing the peg.

Upon assessing the aave.com frontend, users with GHO debt seeking to migrate to non-GHO debt are not routed directly via the Gho Stability Modules (GSMs) and are routed through either Paraswap and Cowswap integrations. When GHO is at peg, users acquiring GHO with either USDC or USDT incur both Swap and Partner Fees instead of the zero fee from Minting GHO via either GSM. Whilst this is not a recent evolution, the addition of the Partner Fee impacts the UX and the introduction of a direct GSM integration would improve the GHO UX.

The effect of the fee is particularly visible in the case of GHO. With a 2bps Cowswap Fee, 15bps Partner Fee and 5bps Flashloan fee embedded in the Cowswap routing, an arbitrage trade aimed at restoring the peg becomes profitable only once GHO trades below 0.9978. This mechanically widens the no-arbitrage band around the peg and allows deeper and more persistent deviations from parity than would otherwise occur. Currently, the stataUSDC and stataUSDT GHO Burn Fee is 8bps and 10bps respectively, and both GSMs would be depleted prior to the arbitrage trade becoming economic. Previously, deviations closer to parity would have been immediately arbitraged. The presence of Fees shifts limits users reliant on the aave.com frontend to where corrective action is delayed until the peg has already materially weakened. The result is likely that GHO’s peg is more dependent on more sophisticated actors not reliant on the frontend.

Partner Fee

Paraswap, now Velora, started distributing part of the positive slippage from swaps via the referrer program during 2022. The Aave DAO has been a beneficiary of this referral program ever since and has accrued a total of $2,560,193 in revenue to date from this integration.

A quick summary of this mechanism;

  • Positive Slippage Capture: Partners (such as wallets or front-ends) can opt in to capture positive slippage, which occurs when a trade is executed at a better price than the user’s initial quote.
  • Partner Fee Sharing: When a partner sets an additional fee on trades routed through Velora, that fee is shared between the partner and the protocol.

The Partner Fee Sharing is currently configured to 0.00% on Velora and to support the DAO’s participation in the Positive Slippage Capture @BGDLabs developed the Paraswap Fee Claimer, whereby a bot calls a function on a contract triggering the transfer of accrued assets to the Aave DAO’s Treasury. Meanwhile, the Flashloan Fee was set to zero via Aave’s governance during October 2023 via AIP 338.

During July 2024, @BGDLabs pushed PR 2101 on Aave Lab’s repository to upgrade the Velora integration from v5 to v6. The upgrade was migrated to PR 2216 and implemented in October 2024 by Aave Labs. In doing so, the feeClaimer workflow was made obsolete and the Positive Slippage was sent directly to the Treasury.

Since then, several teams have contributed to the Velora Protocol to improve swap routing, such as stataToken routing, GSM routings and liquidity routings by @BGDLabs, @AaveLabs and @TokenLogic.

In comparison, the Cowswap solver approach creates a competitive landscape too with over 11 solvers actively supporting orderflow from aave.com. The chart below highlights the distribution of orderflow between solvers demonstrating a robust solve ecosystem supporting aave.com orderflow.

Screenshot 2025-12-16 at 13.30.54
Reference: Aave Analytics | TokenLogic

With the introduction of Cowswap to the frontend, the Flashloan Fee and Partner Fees have been applied as outlined in the extract below from the Aave documentation.

telegram-cloud-photo-size-4-5906693077529725978-y
Source: Aave Protocol Developer Documentation

Users incur either a 15 bps or 25bps Partner Fee, for correlated assets and non-correlated assets, when performing any of the following actions via the aave.com frontend:

  • Swap between Debt types;
  • Swap between Collateral types;
  • Repay Debt using Collateral;
  • Swap when Withdrawing asset; and
  • Swap Tokens held in wallet.

The distribution of the orderflow via the frontend has materially changed since June 2025 when Cowswap was first introduced, with Cowswap now making up over ~92% of recent orderflow from the 4th December 2025, coinciding with this announcement.

Screenshot 2025-12-15 at 20.50.06
Reference: Aave Analytics | TokenLogic

The most dominant orderflow type is Collateral Swap, followed by Debt Swap as shown in the image below. When users perform a Debt Swap, Cowswap solvers are utilising Aave’s flashloan facility. The flashloan generates a 5bps fee that is received by the Aave DAO.

Examples:

  • Debt Swap, Flashloan Fee here, Log 338.
  • Repay with Collateral, Flashloan Fee here Log 897.
  • Swap Collateral, No Flashloan here.
  • Swap when Withdrawing, No Flashloan, here
  • Token Swap, No Flashloan, here.

Screenshot 2025-12-15 at 21.31.05
Reference: Aave Analytics | TokenLogic

The table and chart below compares the fees received by Aave DAO and Aave Labs, each fee type generated by Cowswap and Velora, and the distribution of the Orderflow originating from the aave.com frontend.

Screenshot 2025-12-16 at 13.23.35
Reference: Aave Analytics | TokenLogic

Over the last 12 months, the Aave DAO received $1.34M in revenue from the Velora. Going forward, we expect this revenue to decline to sub $50k during 2026. The introduction of the Cowswap orderflow has generated $69,287 in Flashloans Fees since November, partially offsetting $111,581 USD monthly Velora revenue over the last 12 months. Due to the short sample period, the Flashloan Fee revenue estimation provides limited insights into future revenue generation.

However, by an order of magnitude, the Partner Fee has generated substantially more revenue since the first payment to Aave Labs was made on the 3rd June 2025.

Screenshot 2025-12-16 at 11.50.03
Reference: Aave Analytics | TokenLogic

Recommendation

The economic effects described above are highly sensitive to the level of the fee. Reducing the combined Swap Fee, Flashloan Fee and Partner Fee from 22 to 32bps to a lower level, such as 5bps, would allow revenue generation without resulting in users incurring meaningful payback periods that discourage arbitrage of borrow rates. A combined 5bps fee is equivalent to 38 days payback period using the 48bps borrow rate discount mentioned earlier.

Conclusion

From a purely economic perspective, the stacked fees (5bps Flashloan, 2bps Cowswap, 15-25bps Aave Labs Partner Fee) from the Cowswap integration are significantly larger than those applied via the Velora integration. The Aave DAO <> Velora integration includes a concession whereby Flashloan Fees are waivered and a Partner Fee was not implemented, creating a 20 to 30bps fee discount relative to the Aave Labs <> Cowswap integration containing both Flashloan and Partner Fees.

The Aave DAO’s Velora derived revenue has materially decreased and whilst it is partially offset by Cowswap solvers using Aave Protocol Flashloans, the net result is a reduction in Aave DAO revenue. Since the Cowswap integration went live on 3rd June 2025, the Cowswap Partner Fee generated $3,792,697 in revenue relative to the lifetime revenue of $2,560,193 from the Velora integration by the Aave DAO.

Users are adversely affected by reduced ability to arbitrage borrow rates across markets. A 48bps borrow rate discount between stablecoins requires a 115-day payback period to recover the additional fees incurred using the Cowswap integration relative to the Velora integration via aave.com.

27 Likes

Thanks for this, I was waiting to see this. Lowering the fees to around 5 bps will definitely encourage healthier user activity and of course support better arbitrage. The Payback period will also drop to roughly less than 30 days.

Revenue would likely be around ~$2M per year, compared to ~$10M at 25 bps (with $3.79 M already earned since June).

Putting Aave.com ownership discussions aside, this seems like a reasonable compromise that benefits both users and the DAO, while still allowing Aave Labs to earn some sort of revenue from swap fees.

For the long-term health of Aave, including adoption, liquidity usage, and DAO growth, the full 25 bps simply isn’t worth it.

2 Likes

@JosueMpia @TokenLogic I agree.

The current high fee structure (up to 25 bps) makes rational user actions like debt rebalancing and arbitrage economically unattractive, which hurts market efficiency and GHO peg dynamics. At the same time, DAO revenue declines while a large share of fees accrues outside the DAO.

Lowering the total fee to around 5 bps would materially improve user behavior. At that level, the payback period drops to roughly under 30 days, making debt migration and arbitrage economically viable again for most users.

From a revenue perspective, this still looks reasonable. Even at 5 bps, the annual revenue would likely remain around 2 Mio. USD, versus chasing 10 Mio. USD at 25 bps while suppressing activity.

–> The current setup prioritizes fee extraction over volume and efficiency.

Putting ownership discussions aside, this feels like a fair compromise: Users regain rational incentives, the DAO benefits from better market efficiency and peg dynamics, and Aave Labs still captures meaningful interface revenue.

For long-term adoption and DAO growth, 25 bps is simply too high.

2 Likes

Tremendous work here with this report. Very important to have this data. Many thanks to @TokenLogic !

Again Tokenlogic is delivering high quality data for the DAO and another great addition to the existing dashboards.
Most interesting for me is the aspect around GHO.
From what I understand the implementation is risking that GHO depegs because of lower demand, which is something we cannot allow, especially after GHOs launch problems.

Thanks for flagging this

4 Likes

Thanks to TokenLogic for such a precise and detailed report!

In addition to the economic arguments presented there, I would like to add that the justification used for the integration with CowSwap was mainly based on MEV protection features.

However, Velora also includes MEV protection as a core part of its value proposition.

The new integration therefore provides essentially the same MEV protection characteristics that were used as the rationale for the migration, but with significantly higher fees.

If the goal is truly to respect and prioritize the interests of $AAVE tokenholders and the protocol as a whole, Aave should reconsider its decision and reintegrate Velora integration.

1 Like

Super helpful analysis from the @TokenLogic team, thanks for sharing.
We’ve already put up the change to reduce the swap fee on the Balancer GHO pool to reduce friction and help routing/best execution.
Balancer is always happy to coordinate with @TokenLogic and the wider Aave ecosystem on whatever helps GHO perform best.

4 Likes

I would argue that no rational user would use the aave labs frontend for a swap. The revenue is solely from users who lack understanding of the fees and payback period. It’s acceptable to charge them 25bps because they don’t know they’re being charged 25bps and don’t notice. I doubt we’ll see any benefits from a reduction to 5bps because the rational users who know they are being charged a fee still have no reason to pay it. This only benefits users who don’t notice even after the fact that the fee was charged.

1 Like

Hello @myriad,

With all due respect, a model that relies on extracting value from users because “they don’t know” they are being extracted is not aligned with the Aave brand, history, or culture.

Aave earned its reputation by being a disciplined steward of user interests through prudent risk parameters, careful onboarding, and strong market curation.

The core issue here is simple: the DAO owns the “factory” (the protocol) that delivers the product to users, and the protocol’s margins are comparatively thin. Even with a 10–20% reserve factor on borrowing, on an average 5% stablecoin borrow rate that’s roughly 50–100 bps per year per dollar of borrow volume.

The DAO also funded and financed the “storefront” users rely on, notably aave.com. If, on that storefront, a DAO service provider can unilaterally implement a monetization change overnight, keep the proceeds without a revenue share to the DAO, and do so at the expense of users, then we have a clear misalignment.

This is simply bad business terms for everyone involved except the party capturing the upside, and it deserves reconsideration.

11 Likes

I do not believe this should be the responsibility of the frontend but rather the responsibility of the aggregator and arbitrageurs. If the GSM is over used and favored over other sources of liquidity, it would actually weaken the peg by disincentivizing non-GSM liquidity and placing more stress on the GSM during periods where the peg is being tested (e.g. massive redemptions). It would also result in worse UX for the user, since they may be receiving less favorable quotes.

In general I’m against all of these fees as they seem short sighted and focused on maximizing revenue extraction from the users. I think AAVE should be focusing on ensuring their survival against competition, whether that be Morpho, Spark, Euler, or Fluid. Focusing on revenue that is detrimental to the users, driving them away from AAVE and towards its competitors, is not in the long term interests of AAVE or AAVE Labs.

EDIT: I would like to further state some much necessary FUD. AAVE is not guaranteed to remain king of DeFi in the future, and may lose to its competition if it cannibalizes on itself whether through internal politics or by eating its userbase alive. History is littered with examples of seeming monopolies growing stagnant and extractive and being usurped by a competitor it believed to be too weak to do so. Even atop the mountain, you can still fall, hence why my opinion (as an AAVE holder) is to set these fees as low as possible, if not 0 bps.

1 Like

Thanks @TokenLogic for the detailed analysis of the CoW Swap integration. We have also been collecting similar data points since the collateral/debt swap adapters went live, and we reached similar conclusions. We will therefore enact the following changes:

1. Partner fees for stablecoin to stablecoin debt swaps will be reduced to 0 basis points. This change will eliminate the friction identified in the analysis for users managing their debt positions through the Aave interface, especially when GHO is involved.

2. We will look for a way to integrate the GHO GSM in the swap route whenever a user swaps USDC/USDT for GHO. As you correctly identified, routing GHO swaps through the GSM when GHO is at peg provides a superior user experience by eliminating unnecessary swap fees. This integration will improve the efficiency of GHO debt management and help strengthen the peg.

3. We will continue to monitor swap integrations and adapt as needed. This analysis highlights the importance of ongoing evaluation of how interface-level features impact user behavior and protocol efficiency.

We value this type of substantive, data-driven analysis and appreciate the time you spent on it.

We can’t provide an ETA for these changes for now. While we’re available and online, implementing these changes requires coordination with the CoW Swap team, who may not be available during the holidays. We will provide updates on the implementation timeline as we work through the technical integration.

7 Likes

Hmmmmm. good way to start the new year. More communication and more DAO Benefits. Happy Holidays everyone.

4 Likes

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.

Following up on the items from our last update, please find below the actions we have taken in collaboration with the CoW team.

  1. On January 15th, partner fees for debt swaps were reduced to 0 basis points across all pairs.

  2. GHO GSM routes were already integrated in the swap route for USDC/GHO and USDT/GHO, and CoW solvers were already using them. We will keep monitoring routing behavior and execution quality as volumes evolve.

  3. On February 3rd, CoW’s volume fee for correlated asset pairs (including stable-to-stable) was reduced to 0.3 bps. For reference, see CIP-74 Retrospective: Aligning Rewards with Revenue.

With partner fees now minimized and CoW’s fee model moving towards lower rates for correlated pairs, user-facing friction for these flows are meaningfully reduced going forward.

Aave Labs will continue monitoring swap integrations, with a focus on improving the Aave Interface experience and the underlying protocol mechanics around swaps. We invite @TokenLogic to refresh their analysis using the updated fee parameters to quantify the impact and surface additional areas to improve routing, pricing, and reliability.

2 Likes

Hello,
could you please say why you only reduced it on debt swaps to zero and not on the other actions like “repay with collateral or swap collateral”?

3 Likes