Crypto Marketing Agencies Sell Reach. Here Are Five On-Chain Metrics Founders Should Ask for Instead

Most agency reports I see before a TGE could be swapped between projects and nobody would notice. Impressions are up. The KOL posts went out on schedule, a quest leaderboard has thousands of entries, and somebody has a screenshot of the ticker trending on X for an afternoon.

Then the token lists.

The holder chart rarely does what the deck implied. The call after launch tends to follow a script: the agency talks about awareness while the founder watches price, and nobody in the room can say how many actual people the campaign brought in (because nobody set out to count them). On-chain metrics fix that, and they’re the reason I think founders should stop treating reach as the main thing they buy from a crypto marketing agency.

The stakes went up last year. CoinGecko counted roughly 11.6 million tokens that stopped trading in 2025, which says a lot about how many launches got attention and kept nobody. Your wallet data is public. You can check which kind of launch you’re paying for.

Why reach stopped being a useful number

Reach counts screens. Whether any of those screens belonged to someone who would hold your token, use the protocol or come back in a month is a separate question, and reach can’t answer it.

Bots are the part everyone already suspects. Cookie3 scored more than 175,000 active crypto KOLs and found that the median account produces no quality-weighted engagement at all. So a large share of any per-impression budget goes to noise that a standard dashboard won’t separate out.

The other problems come from how campaigns get reported:

• Wallet connects get counted as users, even though a connect costs the visitor nothing.

• Airdrop and points campaigns attract farmers by design, so a jump in addresses is often one person’s script running overnight.

• Social metrics and on-chain activity live in different tools, and most agency reports only include the first.

• A 30-day window flatters almost everything, since the selling usually isn’t over yet.

Agencies aren’t hiding anything here. They report reach because clients ask for it and because it’s the easiest number on the page to move.

Metric 1: Sybil-adjusted new holders

This is the number of distinct operators that started holding your token during a campaign, with each operator counted once however many addresses it runs. You get there by clustering wallets on funding source and timing, then checking how they behave after the claim.

Last year’s aPriori airdrop shows why. Bubblemaps traced about 14,000 connected wallets to a single entity that ended up with most of the allocation. On a raw holder chart, that would have looked like a community campaign going extremely well.

How to count it

Start with funding. When a batch of wallets is topped up from one source, with identical amounts, inside a short window, you’re usually looking at one operator. In the aPriori case, each wallet received the same tiny amount of BNB through Binance before claiming. After that, compare behavior: identical transaction sequences and transfers into a shared set of fresh addresses tend to settle it.

What to ask the agency

Ask which clustering method they ran and where they set the threshold. Bubblemaps and Nansen have both published sybil analyses, so a line in the report saying “we filtered bots” shouldn’t be accepted on its own.

Where founders go wrong

The mistake I see most often is filtering the airdrop list and nothing else. Paid quests and referral programs get farmed just as hard. They almost never get checked.

Metric 2: 90-day holder retention by acquisition cohort

Retention here means the share of wallets a given campaign brought in that still hold a non-trivial balance three months later. Tag every wallet with the campaign that acquired it, and check balances at day 30, 60 and 90.

Founders resist the 90-day window more than anything else on this list, mostly because it delays the verdict. Delphi Digital’s airdrop study, published in June, found that between 78% and 94% of recipient wallets had sold most of their allocation within 90 days.

It also found that selling kept climbing after day 30. So a report your agency hands over at the one-month mark is, in effect, written before most of the exits happened.

Blended retention hides nearly everything useful. A KOL push and a Galxe quest bring in very different holders. Average them together and you’ll keep funding whichever one happens to be cheaper per click.

Building cohorts like this needs someone who can write Dune queries and tie campaign links to wallets, and plenty of early teams don’t have that person yet. Blockchain App Factory, a crypto marketing agency that works with token projects, handles this by reporting each campaign as an acquisition cohort, with day-90 retention sitting next to spend.

Metric 3: Second-action rate

Second-action rate is the share of new wallets that do something more within 30 days of their first transaction: stake, add liquidity, vote, bridge in more funds, or come back to the product. A first transaction can be pure curiosity.

If a founder could only track one metric, I’d pick this one. Plenty of growth people would argue for retention instead, and I understand why. But holding is passive (sometimes it’s just a forgotten wallet), while a second action is a choice somebody made on purpose.

Settle four definitions before launch:

1. Which actions count. Tie them to revenue or protocol health, because cheap clicks are exactly what farmers script.

2. A minimum value, so a tiny swap repeated twice doesn’t qualify.

3. The window. Thirty days suits most DeFi products; games and social apps may want something shorter.

4. A baseline from organic wallets, measured the same way.

That last one is easy to skip. Without it you can’t read the number at all. A second-action rate that looks weak for a lending protocol might be excellent for an NFT marketplace.

Metric 4: Campaign sell-through to exchanges

Sell-through tracks how much of the token a campaign put into wallets later reached exchange deposit addresses or got sold on DEXs, and how quickly. It tells you whether the campaign created buyers or simply lined up future sellers.

Arrakis Finance looked at last year’s token launches and found that about 85% ended 2025 below their launch price. Its explanation leaned on launch mechanics, early sell pressure in particular, more than on market timing.

Marketing sits right inside that. Any campaign that hands out tokens has also placed a sell order that will fill at some point, whether anyone planned it or not, and somebody ought to be watching when it does.

What to pull from the data:

• The share of campaign-distributed tokens that reached a CEX deposit address within 7, 30 and 90 days.

• The median number of hops between claim and exchange, since farmers usually move in one or two.

• Whether selling bunches up around particular KOL posts or quest end dates.

When sell-through is high and retention is low, the campaign mostly paid people to leave. Book what it cost as a token emission.

Metric 5: Cost per retained wallet

Take total campaign spend, add the market value of any tokens handed out, and divide by the sybil-adjusted wallets still holding or active at day 90. That one figure puts a KOL program and a quest campaign on the same footing, even though their reports look nothing alike.

The arithmetic is easy. The inputs are where people cheat, usually without meaning to.

Count tokens as spend

Token rewards get left out of campaign cost all the time, because no cash left the treasury. Price them at their value on the day they were claimed, or a points campaign will always look cheaper than it was.

Use the filtered denominator

Divide by the operators from Metric 1 who are still holding under Metric 2. Raw wallet counts make every campaign look efficient, which is exactly why they show up in so many decks.

Compare channels as well as campaigns

Patterns show up after three or four campaigns. We’ve seen KOL weeks where wallet connects jumped and the retained count hardly moved. We’ve also seen smaller community programs that cost more per click but far less per wallet still around at day 90.

How to write on-chain metrics into a crypto marketing agency contract

Write all five metrics into the scope of work, definitions included, before the campaign goes live. If they first appear in the post-mortem, every definition gets renegotiated after the results are in.

I’d take impressions and follower growth out of the success criteria completely, and I know some agencies won’t like that. Leave them in the report as context. Just don’t pay against them.

A brief that holds up covers at least this much:

Agree on a data source at the start. Dune, Nansen, Flipside or an in-house indexer will all do the job, but switching halfway through makes the numbers impossible to compare. And give the agency read access to your wallet tagging, so attribution doesn’t run on screenshots forwarded through Telegram.

Expect pushback on the 90-day wait. It’s a fair objection, and it’s why the weekly sybil-adjusted count and the day-30 second-action rate are in the table: they’re your early read. Payment still follows the day-90 numbers.

FAQ

What on-chain metrics should I track for a crypto marketing campaign?

Track sybil-adjusted new holders, 90-day retention by campaign cohort, second-action rate, sell-through to exchanges and cost per retained wallet. Together they show whether a campaign brought in actual people and whether those people stayed.

How can I tell if my token holders are bots or sybil wallets?

Cluster wallets by funding source and timing, then compare what they do after claiming. A group funded from one source in a tight window, which then behaves identically, usually belongs to one operator.

Is 30-day retention enough to judge a token campaign?

Usually not. Selling from airdrops and quest rewards often runs well past day 30, so a 30-day figure tends to overstate how many holders a campaign kept.

Should I pay a crypto marketing agency for reach or for results?

Pay against on-chain results you can verify, such as retained wallets at day 90. Keep reach in the report as background.

Where to start this week

Export the wallets that touched your contract during your last campaign. Run them through Bubblemaps or a basic clustering query on Dune, count operators instead of addresses, and check how many still hold today. Bring those two numbers to your next agency call and ask how the next campaign is supposed to move them.