Allocators check five things before they deposit: fees earned per dollar of TVL, how much of the pool they could withdraw today, how stable the TVL is, how concentrated it is on one chain, and who can change the contracts. Our screen of 60 lending protocols on 14 October 2026 shows the typical values, so a protocol team can see where it stands.
How did we build the screen?
We pulled DefiLlama's protocols and fees endpoints on 14 October 2026 and kept every protocol in the Lending category with $10M or more in TVL (DefiLlama protocols, DefiLlama fees). That left 60 protocols holding $51.73B, or 99.4% of the category's TVL. Every figure below is our arithmetic on that data (our data, October 2026). Audit and oracle fields come from what each protocol submits to DefiLlama, so they show disclosure on DefiLlama, which can lag the protocol's own documents.
Which metrics do allocators check, and what does the market look like?
| Metric | What it tells the allocator | How to compute it | Screen result, 14 Oct 2026 |
|---|---|---|---|
| Annualised fees / TVL | Whether users pay for the product | Last 30 days of fees × 365 / 30, divided by TVL | Median 2.42%; middle half 1.22% to 4.13% (54 protocols with fee data) |
| Utilisation | How much supplied capital is lent out, so how much an allocator can withdraw now | Borrowed / (TVL + borrowed) | Median 33.6%; middle half 23.8% to 40.9% (57 with borrow data) |
| 7-day TVL change | Whether deposits are leaving | DefiLlama change_7d field | 6 of 60 lost more than 10%; 1 lost more than 20% |
| Chain concentration | Exposure to one chain's outages and liquidity | Largest chain's TVL / total TVL | 48 of 60 hold 90% or more on one chain |
| Audits disclosed | Whether reviews are public | DefiLlama audits field and audit links | 20 of 60 list no audit on DefiLlama |
| Oracles disclosed | Which price feeds the protocol trusts | DefiLlama oracles field | 46 of 60 name their oracles |
Run the same columns on your own protocol and place yourself against the medians. An allocator's analyst will do it before the call.
How do allocators tell real TVL from rented TVL?
They compare TVL with the fees users pay. A protocol whose TVL comes from token rewards shows a high TVL and low fees, and the deposits leave when rewards end. In our screen, the middle half of lenders earned between 1.22% and 4.13% of TVL in annualised fees. A protocol far below 1.22% will need to explain where its deposits come from.
Then they check direction. Six of the 60 lost more than a tenth of their TVL in the seven days to 14 October 2026. A drop like that leads to one question: did a few large depositors leave? Answer it before the allocator asks, with the depositor count and the share held by the ten largest addresses, computed from your deposit and withdraw events.
How do allocators test whether they can exit?
They look at what is left to withdraw. In a lending market, supplied capital that is lent out cannot leave until borrowers repay or are liquidated. At the median utilisation of 33.6% in our screen, about two thirds of supplied capital sat idle and withdrawable. Higher utilisation pays suppliers more and leaves less room to exit.
An allocator models its own position against that. Suppose a fund plans to deposit $20M into a market with $100M supplied and 80% utilisation. Before the deposit, $20M is free to withdraw. After it, $40M is free, and half of that is the fund's own money. If other suppliers take out $20M first, the fund waits for borrowers to repay. Publish your utilisation history and your interest rate curve, so the allocator can see how fast rates rise to pull borrowers back.
What do allocators check about control?
They read the contract permissions before the yield. Twenty of the 60 lenders in our screen listed no audit on DefiLlama. Some of those will have audits on their own sites, and an analyst who finds none on the first page they check will mark it as a gap anyway. Add the links.
Allocators also ask:
- Who holds the admin keys, and how many signers must approve a change?
- How long is the timelock on upgrades and parameter changes?
- Which oracle prices each asset, and what happens when it stops updating?
- Who can pause the protocol, and who can unpause it?
- Has the protocol had an incident, and what did it repay?
Each answer should link to a contract address, a governance post or a report. Allocators reuse those links in their own approval memos.
How much does chain concentration matter?
It matters more for funds with mandates across several chains. Forty-eight of the 60 lenders held 90% or more of their TVL on one chain. That is normal, and allocators accept it when they can size it: what share of the fund's exposure to that chain would this position add? State your chain split and the bridges your assets depend on.
For staking products the questions shift toward validators and exit queues. We cover that case in why institutional capital would stake on a new layer 1.
What should a protocol publish before allocator calls?
Publish each of these, dated, on one public page:
- Annualised fees / TVL for the last 30 and 90 days, with token rewards shown separately.
- Utilisation history per market and the interest rate curve.
- Depositor count and the share of the ten largest depositors.
- TVL by chain and the bridges each asset depends on.
- Audit links, admin keys, signer threshold and timelock length.
- Oracle per asset and the fallback when it stalls.
- The query or script behind every number.
Decision rule: if your fees / TVL sits below the screen's lower quartile or your utilisation sits above its upper quartile, prepare the explanation before the first allocator call.
To benchmark your protocol against these metrics and fix the gaps an allocator would find, start with The Diagnostic at $2,500.