Solana’s unique transaction format and large blockchain history can pose challenges for analysis, which has previously led to some concerns over some of its network metrics. Indeed, Solana’s differences prevent the direct application of adoption metrics we’ve previously utilized for chains compatible with the Ethereum virtual machine (EVM). That said, we find that Solana’s network metrics are broadly similar to other low fee chains according to an alternative set of activity distribution analysis.
Solana fee spending surges during US hours, mostly in line with activity patterns we’ve previously observed in both centralized and decentralized exchanges. That said, Solana’s peak activity timing is skewed later in the day towards Pacific Time Zone hours, suggesting it may have a distinct cohort of active users compared to other networks.
Separately, 26% percent of Solana transaction fees in 3Q24 are spent on failing transactions, down significantly from a peak of 55% in March. While still higher than other low cost chains like Base, an Ethereum L2 incubated within Coinbase, (which averaged 14% over 3Q24), the fees spent on failed transactions is moving closer towards its peers’ ranges.
The concentration of active users also appears in line with competitors. 0.13% of accounts on Solana are responsible for 90% of total non-vote fee spend. Much of this spend is linked to DEX activity, with 75-90% of total Solana fees spent on DEX trades. This is higher than Ethereum and Base DEX activity, which account for 55-65% and 60-70% of their fees, respectively. We think this may be driven by its larger ecosystem of memecoins, though its burgeoning gaming and DePIN ecosystem could help diversify its fee demand drivers in the long term.
Formatting Differences
Solana transactions have a major difference from Ethereum and other EVM transactions that prevents a direct comparison via our existing adoption measurement heuristics – they lack a singular clear recipient address. Among other things, Ethereum transactions contain a sender address (“from”), a recipient address (“to”), and a data payload (which is then executed by the recipient address if applicable). Although the recipient address may in turn call other addresses to execute additional actions, the consistency of behaviors across EVM systems makes metrics like the h-index or recipient distribution ratio comparable between EVM chains.
Meanwhile, Solana’s transaction format also contains a sender address (“signer”), which is effectively equivalent to the “from” field in the EVM. But the similarities largely end there. Instead of a recipient address, Solana transactions contain a list of instructions, each containing an “executing account” with an associated data payload.
A naive approach would treat the first executing account as the equivalent EVM recipient (“to”), but that immediately falls short on quick inspection. The first executing account of nearly all transactions is either the System Program (at address 11111111111111111111111111111111) or the ComputeBudget Program (at address ComputeBudget111111111111111111111111111111). The target recipient is often hidden in the data payload of one of the sub instructions, even for simple operations.
A simple SOL transfer, for example, could contain three separate instructions. The first two instructions interact with the ComputeBudget Program (as the executing account) to set the execution price and limit. Then the last instruction refers to the System Program, which triggers the transfer function and passes in the destination address (comparable with “to”) as part of the data payload. Thus, to isolate the recipient, one would not only need to iterate through the list of instructions, they would also need to appropriately parse the data payload. In contrast, interpreting a direct ETH transfer can be done by directly parsing the “from” and “to” fields.
That is not to say that Solana’s transaction format is better or worse than Ethereum’s. It is simply different. This requires careful attention when drawing cross chain comparisons, and we opt to look at an alternative set of metrics based on timing and fees for cross network comparisons (beyond well studied metrics like stablecoin volumes, total value locked, etc.). The difference in transaction formats are reminiscent of those between Ethereum and Bitcoin transactions that render certain comparisons challenging, albeit to a lesser degree.
West Coast Vibes
With that understanding in mind, we start by analyzing the timing of transaction fee patterns. In a previous report, we found that trading volumes – both on centralized exchanges (CEXs) and decentralized exchanges (DEXs) – peak during US East Coast hours. (This is still true today.) We also confirmed that transaction fee spending and overall transaction counts on Bitcoin and Ethereum follow this trend.
Solana fee spending similarly trends up during US hours, though it peaks several hours later in the day compared to Bitcoin and Ethereum. See Chart 1. Throughout 3Q24, Bitcoin and Ethereum fees reached their average peak at 1pm and 2pm UTC respectively (9am and 10am ET), while Solana fees peaked at 8pm UTC (4pm ET or 1pm PT). This is a fairly intriguing timezone to peak in, as there is no major economic zone beyond the US West Coast that enters the market at or immediately after 8pm UTC (just the Pacific Ocean).
This suggests to us that a significant portion of high value transactions could be US-based, potentially with a large user base skewed towards the Pacific region. On the flip side, its bottoming fee spend at 9am UTC indicates a comparatively weaker European/African presence (or at least a cohort that is not particularly active in the mornings). For reference, Base peaks at 10am ET (similar to Ethereum), while both Arbitrum and Polygon PoS peak at 9am HKT, suggesting they have a stronger Asia footprint.
Overcoming Failure
Separately, we think it's worth revisiting Solana transaction failure rates and their contribution to overall network fee spending. In March and April 2024, Solana was severely strained by heavy activity where upwards of 70% of all transactions were failing. This figure excludes “unlanded” transactions that were dropped without a trace suggesting that the end-user experience failure rate was likely even higher than that tracked onchain. Since then, a series of network upgrades including stake-weighted quality of service and the Solana Labs v1.18 release resulted in the transaction failure rate dropping to 37% over 3Q24 with (anecdotally) fewer occurrences of unlanded transactions. This has vastly improved the user experience for most casual (i.e. non-bot) users, but remains a unique fee driver for Solana compared to other networks.
That said, although the failure rate may still appear fairly high, we think that this primarily is a byproduct of Solana’s low fee environment and non-deterministic transaction ordering. This enables more spam-like transactions with a naturally higher failure rate. More importantly, when looking at the proportion of transaction fees spent on failed transactions (rather than the proportion of transaction counts), the reduction in failed activity becomes more apparent. Fees spent on failed transactions dropped from a March peak of 55% to a current range of 20-25%. While still higher than other low cost networks like Base and Arbitrum, fee spending is becoming increasingly efficient on the network.
Ethereum may represent a more desirable “end state” of efficient fee spending in our view, with less than 3% of its fees spent on failed transactions. That said, Ethereum’s comparatively lower failure rate and fee spend is partially driven by longer slot times (12s vs 400ms) and a more mature MEV ecosystem that eliminates the profitability of spam-like transaction retries.
All You Do is Trade
Following its rapid rise activity throughout 4Q23 and 1Q24, Solana’s weekly fee spend has surpassed that of Ethereum’s on several occasions this year. Much of this activity can be directly attributed to DEX activity. By filtering transactions based on Dune’s Solana DEX trades dataset, we can isolate the transactions involved in DEX trades. Doing so reveals that 82% of Solana transaction fees have been spent on trades in Q3, higher than the respective 62% and 65% on Ethereum or Base (using a similar dataset and methodology), and substantially higher than Arbitrum and Polygon PoS. See Chart 4.
We think this speaks to the importance of the trading ecosystem on Solana, where memecoin launchers like pump.fun have earned more protocol revenue ($63M) in 3Q24 than long standing DeFi giants like Aave ($15M). We think this may pose some risks to fee sustainability if memecoin attention subsides, though the growth of Solana’s gaming and decentralized physical infrastructure (DePIN) ecosystems may help diversify its fee generating activities in the long run.
Onchain Superusers
The near term concentration of Solana activity on DEXs is mirrored by the concentration of its top fee spenders. Of the 125.7M unique active accounts in 3Q24, just 169K (0.13%) of them contributed over 90% of the total transaction fee spend. In fact, the top 36 accounts alone contributed over 10% of all fee spend.
Upon closer comparison, however, it becomes clear that this skewed activity distribution is the norm – particularly for low cost chains. The top 0.02% and 0.6% of users contribute 90% of transaction fees on Base and Arbitrum respectively, for example. Chart 5 depicts a distribution curve of user fee spending. The larger the area under the curve, the more equally distributed fee spending is among users. Solana follows a similar distribution pattern to Polygon PoS and Arbitrum. Notably, Ethereum has the widest contribution base, with 5.2% of users contributing 90% of fees.
Conclusion
The distribution of Solana’s onchain activity is generally in line with other low cost blockchains, and may even have a relatively more distributed user base by some measurements. Furthermore, despite its historically high transaction failure rate, the proportion of Solana fees spent on failed transactions is moving to become comparable to low cost chains. The small group of power users contributing most of the transaction fees also appears in line with networks like Arbitrum and Polygon. In our view, this activity distribution suggests an overall healthy and growing ecosystem.
That said, Solana’s onchain activity drivers are still heavily concentrated on DEX activity. In our view, this could lead to an outsized impact on network fee revenue in the event of a market downturn (and an associated reduction in trading activity). Furthermore, its activity concentration towards the Pacific Time Zone suggests that there may be less coupling between Solana and traditional financial markets and hours (e.g. in New York or Hong Kong) compared to networks like Ethereum or Bitcoin.
