In Epoch 7, we hit a record-high 659 users participating in the allocation window (up from 496 in Epoch 6). On the flip side, only 26% of eligible users showed up — the lowest relative turnout we’ve ever seen. The main contributing factors are low engagement from the freshly onboarded SheFi and Celo communities, plus a big drop-off among users who did allocate in Epoch 6.

Out of 868 eligible SheFi users, only 278 participated (32%), meaning 590 streams will be canceled. For Celo, participation was worse: 41 out of 270 (15%), so 229 cancellations there. In total, we’re scrapping 819 streams. Barring any major locks or unlocks during Epoch 8, we’re looking at around 1,300 addresses eligible for the next allocation window.

In Epochs 1–6, most allocations happened within the first 2–3 days of the window, with a few anomalies caused by lower weekend activity and sybil spikes in Epochs 1–4. Epoch 7 broke the pattern: it had a sluggish start but an unprecedented surge in the second week — nearly 100 users made their first allocation in the final 24 hours alone.

Allocation timing followed the usual daily rhythm: lowest activity between 3–9 AM UTC, and a strong peak between 3–9 PM UTC (almost 40% of allocations happened during that window).

The SheFi cohort was by far the largest group among our users. Meanwhile, the cohort that shrank the most since Epoch 6 was the group of Epoch 5 giveaway beneficiaries.

The left chart shows all addresses participating in the allocation window. The right chart filters out addresses with a UQ score below the threshold and those with less than 200 GLM deposited at epoch-end (there’s still a fair amount of suspicious accounts here, even if their UQ scores look decent). Bottom line: we estimate the real size of our user base at a max of 570 users — with 103 organic users, meaning addresses that didn’t get any funding through our marketing campaigns.

After we introduced quadratic funding in Epoch 4, Octant users had way fewer incentives to tweak their allocations mid-window — no funding threshold, and a single user’s impact was much smaller. As a result, the average number of allocations per user dropped from around 1.6 in Epochs 1–3 to 1.3–1.4 in Epochs 4–6. In Epoch 7, though, it’s back above 1.6, with over a third of users changing their allocation at least once during the window.

Total donations in Epoch 7 were just 4.53 ETH — the lowest since we introduced PPF. The almost 10 ETH drop vs Epoch 6 was mostly due to our biggest whale-donor claiming their entire reward and allocating zero to projects, unlike in earlier epochs. Beyond that, the donation-to-budget ratio dropped across every user-size category, with the steepest fall in the sub-500 GLM group (from 92% to 66%). Meanwhile, the long-running slide in the 10k–200k GLM bracket continued too (65% in Epoch 4 down to 34% in Epoch 7).

The share of users not allocating anything was high across the board in Epoch 7, especially brutal in the sub-500 GLM group — almost three-quarters didn’t even show up. What’s new is a sudden rise of sub-500 GLM users who did allocate but then claimed the full reward anyway: only 3 such cases in Epoch 6, but 69 in Epoch 7. Until now, this behavior was basically reserved for the big lockers.

The share of users donating at least something dropped by 7 percentage points compared to Epoch 6. The lowest donation rates came from the organic users (i.e. no marketing campaign connection) and the SheFi cohort.

Since Epoch 4 (aka the quadratic funding era), we’ve been watching the average number of projects supported per user slide lower — and in Epoch 7 it finally dipped below 4 for the first time. Part of it is simple: fewer projects to pick from (20 instead of 30), and most users weren’t familiar with the new set of beneficiaries. Still, a couple of fun facts: 18 users split their rewards across all 20 projects — something we’ve never seen at this scale before. On the flip side, 179 users backed only a single project.

The quadratic distribution of MF in Epoch 7 turned out pretty interesting. Unlike previous epochs, it didn’t cause any major reshuffling in the ranking of beneficiaries compared to a hypothetical linear distribution. As usual, it smoothed things out: the long tail of less popular projects still snagged a piece of the matching pie (in contrast, the linear model assumes a 1/n threshold, as we had until Epoch 3).

Generally, projects that pulled in the most total donations were also backed by the highest number of users — another reason why the quadratic and hypothetical linear outcomes ended up looking pretty similar this time.

Most of the users active in the Epoch 7 allocation window were first-timers at Octant, mainly from the SheFi and Celo cohorts. Same story as before: very few users returned after going inactive in the previous epoch (Epoch 6 in this case). What will be really interesting to watch in Epoch 8 is whether the Epoch 6 allocators who skipped Epoch 7 (probably because the climate focus didn’t excite them) come back for an Ethereum-specific round.

With most projects being new faces to Octant users, people were also a lot less opinionated this time about who should get their donations. As a result, there were no strong vote correlations between projects — with one exception: a pretty noticeable overlap between votes for Atlantis and GainForest. Definitely something worth digging into.

Because none of the projects had strong ties to the Octant user base (unlike previous epochs where loyalties ran deep), single-project allocations were way less common this time, especially compared to Epoch 6.