Graph On Chain: A Real-life Deepdive Activity Analysis

• In this article, we explore the option of using publicly available on-chain transaction data to evaluate startups. We share the results of our experiment, exploring The Graph's indexing service.
• We explain what The Graph is, and provide detailed steps of accessing event logs. After we compile Cohorts, we look at them in two ways: the number of customers by Cohort and Revenue by Cohort.
• To make sure we're in the ballpark with respect to the cost of analysis, we provide our cost estimates and a pricing comparison looking at other indexers who provide similar service.
• While The Graph's user base and revenue are growing, they may not be keeping pace with market demand. However, continued analysis of The Graph's components will reveal more insights into the value proposition of its decentralized indexing.

At Headline VC, we heavily prioritize data-driven decision-making. We regularly ask startups to provide us with access to their user transaction logs, which our Investment Team then runs through our proprietary analytics platform - Deepdive. Deepdive generates insightful metrics and nearly one hundred visualizations per case, enabling us to quantify key aspects of the business and product.

In working with blockchain startups, we began contemplating an intriguing possibility - could we bypass the startup and pull the necessary transaction data directly from the blockchain itself? This would alleviate the burden on startups to prepare and share their data. It would also allow us to freely publish our analyses as the data would be public. Currently, when startups provide their data, we sign NDAs limiting our ability to share findings.

We decided to test how far we could stretch our analytics using just on-chain transaction data, without private logs from the startup itself. We chose The Graph as our initial experiment.

Maintaining our own on-chain database is impractical given the immense and ever-growing amount of blockchain data. This is where The Graph’’s decentralized indexing service delivers value.

The Graph ingests blockchain data, processes and indexes it, then serves it in an accessible format. Importantly, this entire pipeline is decentralized and transparent.

The Graph ecosystem relies on four key roles collaborating:

1. Indexers
2. Delegators
3. Curators
4. Developers

In theory, these participants can provide a highly robust, scalable indexing infrastructure when working together. We set out to explore this by analyzing how The Graph's paying customers utilize the indexing service.

Outlined below are the detailed steps we followed to collect the data for our experiment: 

• Their billing contract, according to its official billing page, is hosted on the Polygon network.
• We began by importing event logs and decode them using Yifei Huang's fantastic blog post.
• We got The Graph's Token (GRT ) USD history pricing from the CoinGecko API. This allowed us to value each transaction in USD.
• We created two versions of the analysis: one in GRT and one in USD.

To make this more intuitive, we refer only to the USD version below. 

We found 4,056 event logs of four types:

• tokenspulled (3244 times) – this is triggered when the service charges the address based on their query fee
• tokensadded (767 times) – this occurs when users store GRT into the contract
• tokensremoved (44 times)
• gatewayupdated (1 time)

We validated our findings by comparing our calculated user balances on the events log with the contract's userBalance function.

Using cohort analysis, we segmented users based on their initial transaction month. This reveals insightful patterns into user lifecycles over time.

Figure 1 below shows monthly active paying users stacked by cohort.

3 conclusions can be deduced from above graph (Figure 1):

1. The number of monthly paying users is still small, ranging from 75 to 150.
2. With the exception of the Oct 21 Cohort, users increased steadily month after month. For example, the number of users on March 22 surpassed the previous peak on October 21.
3. User Retention is close to 100% with the exception of the Oct 21 Cohort.

Figure 2 below segments users by monthly cohort in the same way, except the y-axis value represents Revenue Contribution from each Cohort.

By comparing these graphs, we can see that the number of new users consistently joining the service has been increasing. However, they're spending significantly less than the users who joined in the October 21 Cohort.

The Average Revenue Per User (ARPU) Cohort Chart illustrates this situation more clearly. Customers who joined after Oct 21 and Nov 21 spent less than $10 USD per month.

Cost Comparison of Indexers

According to this reddit post, it cost 0.00005 USD per API call. Because the price per query is determined by the indexer, there's no unified pricing. However, if we use the estimated .0.00005 per query cost, this would mean the newer Cohort uses 200k API calls per month ($10/.00005).

To help in comparing costs, below are pricing structures for two alternative indexing services:

• Coin Gecko charges a $129 USD monthly subscription fee for 500k calls per month.
• Coin Market Cap offers $35 USD per month for 40K call credits.

Although they aren't identical businesses, they're close enough for us to conclude that if The Graph is less expensive, it's worth the effort to determine whether they have the data we need to perform analysis.

Looking at this data holistically, we can make some key observations about The Graph's indexing product:

• The user base remains small but is steadily climbing
• Revenue growth has not kept pace with the rise in users

Our estimates suggest 100k API calls may sufficiently serve some smaller dApps, especially those with caching. As such, these new users may have already tapped The Graph's full capacity without reservations.

Two potential explanations emerge:

• This early market has limited total demand
• There are friction points preventing The Graph from meeting full demand

We have reason to believe the latter is more likely. For example, the most signaled Subgraph in The Graph's Explorer has received minimal fees, implying non-economic signaling motivations. We'll dive deeper into Signals and Curators in future posts.

While decentralized solutions have advantages, they may lack the robustness needed for enterprise use cases presently. We look forward to sharing more insights into the components of Indexers, Developers, Curators, and Delegators as we continue exploring The Graph, and using Deepdive for comprehensive analysis.