The State of Web3 Marketing — 2026 Q2

The numbers, with their sources

The most common “state of” piece in this category cites a different set of numbers in every paragraph and never sources them. We list every number used in this report with its primary source, snapshot date, and the qualification that matters.

The numbers above are the only ones used in this piece. Where a follow-up paragraph references a percentage or dollar figure, it derives from one of these rows. We don't cite numbers we can't source.

Channel ROAS in 2026: the 13× spread nobody talks about

Most web3 marketing pieces describe channels qualitatively: “Twitter is good for awareness, KOLs are good for activation, Discord is good for community.” That's useless when you're deciding where to put the next $10K. The only useful version is a real protocol's real numbers.

We've published a full 6-channel ROAS breakdown from a Base lending protocol's first $1M in attributed TVL on $25K of total marketing spend. The summary:

The 13× spread between Discord and SEO is not the headline. The headline is that most protocols don't have the attribution to see this spread until after a quarter of spend. Without channel-level on-chain attribution, the natural decision is “split the budget across channels and hope for the best.” That's functionally equivalent to losing 30–40% of marketing spend to channels you would have cut if you could measure them.

The caveat: this is one protocol's mix, with one product, one geography, one team. The absolute ROAS numbers aren't universal. The 13× spread between best and worst channel, for this protocol type, is consistent with the small sample of similar protocols we've seen.

The transparency gap: 4 of 8 publish pricing

Of the eight attribution tools we surveyed in our pricing report and feature matrix, only four publish pricing publicly. The other four hide pricing behind a sales call or expose only partial information. The split is not random.

The opacity is a deliberate choice, not an oversight. Sales-led pricing protects margin against price-sensitive buyers and enables variable pricing by account size — both rational from a vendor standpoint. The cost is buyer-side: every evaluator without enterprise budget has to either (a) sit through a sales call to learn pricing or (b) self-eliminate.

The structural problem: AI search citation correlates with public information. Tools that publish pricing get cited more often in AI Overview answers, Claude/ChatGPT responses, and Perplexity research because language models can quote concrete numbers but cannot quote “contact sales for pricing.” In 2026, hidden pricing means hidden citations.

Agent traffic: the channel nobody's measuring

The x402 payment standard went live on Base in late 2025 and now powers a small but growing share of API and protocol traffic. AI agents — long-running LLM-based systems that pay for service on the fly — settle x402 payments without ever loading a browser page. They don't click a UTM link. They don't connect a wallet via a website. They call a contract.

Every attribution stack we surveyed is built on the assumption that the click happens in a browser. Of eight tools, zero publicly address agent traffic as a distinct attribution surface. The current treatment is implicit: agent transactions are bucketed under “unknown channel” or “direct,” depending on the tool.

For most protocols today, agent traffic is <5% of conversions and the measurement gap is small. But the trajectory matters: every quarter, more protocol revenue comes from automated systems rather than browser sessions. The protocols that build attribution for this surface in 2026 — not 2028 — will have a year of data when their competitors don't.

The practical 2026 Q3 task for any DeFi or AI-adjacent protocol: instrument agent transactions as a distinct funnel. Tag agent-originated x402 calls with a campaign identifier (e.g., a partner-specific endpoint or signature pattern), and measure agent-originated conversions separately from human ones. See our AI agent ROI tracking guide for the methodology.

Q1 → Q2: what changed

The category moved in three ways between Q1 and Q2 2026, all of them visible from outside.

• Publishing cadence accelerated. Formo.so shipped a 4,000-word “Onchain attribution in 2026” piece on May 4 that immediately ranked #2 organic for “defi marketing attribution” — a query that one of our own guides held position 4 for the week prior. The lesson is not that formo wins by content alone; it's that the SERP for category queries re-orders weekly based on freshness × DA, and tools that don't publish at that cadence drop. The defensive content move is now a quarterly minimum.
• AI Overview firing got more volatile. Across four weekly checks on our three standard category queries (best web3 attribution tools, alternatives to Spindl, defi marketing attribution), Google AI Overview rendered 4 of 12 times — a 33% firing rate. Two consecutive weeks saw 0/3. The implication: AIO is not yet a reliable channel to optimize for in this category. Treat AIO citation as a bonus, not a target.
• The case-study lever became the default. Three of the eight category vendors published case studies or data drops in Q2 that hadn't existed in Q1. Spindl pushed a benchmark report; Formo's onchain piece functions as a long-form case for their methodology; we published the first $1M attributed TVL breakdown. Generic explanation pages are losing to specific, sourced numbers.

Practical recommendations by team stage

What to do in 2026 Q3 depends on where you are. Three sets of teams, three different priority orders.

What to watch in Q3

Four signals worth tracking through the third quarter:

• Whether AIO firing rate stabilizes >50%. The current 33% firing rate across category queries means it's not a reliable channel. If Google rolls AIO out more broadly to attribution queries in Q3, the ranking lever shifts and content optimization changes shape.
• x402 transaction volume on Base + Solana. If agent-originated transactions cross 10% of any major protocol's total transactions before year-end, the attribution stacks that don't address agent traffic become materially under-counting.
• Sales-led pricing transparency. Two of the four hidden-pricing vendors (Spindl, Addressable) have product pages that don't mention pricing tiers at all. Watch for a shift: if either publishes any public price point in Q3, it signals the category is consolidating toward self-serve.
• Tier-2 KOL pricing inflation. The 4.4× ROAS for Tier 2 KOLs in our case study is the strongest paid-channel return in the sample. As more protocols discover this, Tier 2 KOL rates will rise. Track whether your protocol's Tier 2 KOL price quotes are >1.5× their Q1 levels by Q3.

Methodology + open data

All numbers in this piece are sourced. The full feature/pricing dataset for the eight tools surveyed is open-source on GitHub under CC-BY-4.0:

For the underlying case-study data (channel-by-channel ROAS, cohort retention, wallet quality), see the first $1M attributed TVL case study. For the 2026 DeFi marketing benchmarks (CAC, LTV, wallet quality across cohorts), see the benchmarks page.