Monetizing a wallet or dapp without wrecking it

If you build a wallet, a dapp, or an explorer, the word “ads” probably triggers an immediate no. You have spent years earning the trust that keeps a user signing transactions in your app instead of someone else’s, and the mental image of a blinking banner selling that trust for pennies is genuinely offensive. That instinct is correct more often than it is wrong. This piece is not an argument that you are being precious. It is an argument that the instinct is aimed at a specific kind of ad, and that a different kind exists.

Why builders hate ads, and why they are usually right

Most in-app advertising is rented junk. The network does not know your product, does not know your user, and does not care whether the placement makes sense, because it is paid to fill a slot regardless. So you get an ad for a casino next to a swap confirmation, or a token that has nothing to do with what the user is doing, served because someone bought the impression cheaply. That ad costs you trust every time it renders, and trust is the entire asset. For a product team, the product is the balance sheet. Anything that quietly erodes it in exchange for a small revenue line is a bad trade, and you are right to refuse it.

The failure is not that ads are inherently corrupt. It is that the usual model treats your surface as inventory to be filled and your users as impressions to be counted, and it optimizes for fill rate over relevance because fill rate is what gets paid. A network that will always show something, no matter how irrelevant, has already decided your product is worth less than the impression.

An ad that behaves like a recommendation

Reframe the thing you are placing. Not an ad in the banner sense, but a recommendation: relevant to what the user is actually doing, native to the surface it appears on, and absent when nothing relevant exists. A user who just bridged to a new chain and is shown a protocol live on that chain has been helped, not interrupted. The same slot showing a random token to a user mid-transaction is the junk you already reject. The difference is discipline, and specifically the discipline to serve nothing.

That last part is the tell. Specify runs on the rule that no ad is better than a bad ad, so a slot returns empty rather than filling with something off-target. Coin98 is the concrete version of this: they run Specify placements as recommendation surfaces inside the wallet, not as banners bolted onto it, and the placement earns its keep because it fits what the user came to do. When the match is real the thing reads as a feature. When there is no match it disappears. An ad that can disappear is closer to a recommendation engine than to the tax you are picturing, and it is the only version worth putting near a product you care about. The same logic that makes wallet targeting work for advertisers is what makes the placement relevant for your users.

The bar an SDK has to clear

A recommendation model only survives if the engineering underneath it respects your product. For a product team, that is a real bar, and any network worth integrating should clear all of it. Here is the one Specify is built to, and the one you should hold anyone else to.

• Fail open, always. An ad outage can never break your app. If the network is down, the slot is empty and nothing else changes. This is non-negotiable and it is the first thing to test.
• Fast and cached. A serve around 200ms with 24h ad caching, so the placement never becomes the reason a screen feels slow.
• No fingerprinting, no page-content collection. The network does not scrape what is on your screen or build a shadow profile of your user. Identity is consent-gated, not harvested.
• Controls in a dashboard, not in code. Allow lists, block lists, and category filters live in a dashboard you can change without shipping a release, so you decide what is allowed near your users and can change your mind in a minute.
• First-look above existing monetization. The SDK sits at first-look priority above whatever you already run, with an instant no-fill fallback, so an empty response falls straight through to your current stack instead of leaving a hole.
• Integration measured in days. Most surfaces go live in under a week, and there is a coding-agent prompt that handles the integration, so this does not become a quarter-long project.

The full interface is in the SDK reference. Read it against your own architecture before you commit, and treat any network that cannot meet the fail-open and no-fingerprinting bars as disqualified regardless of the rate it quotes.

The economics of wallet-verified surfaces

The reason this is worth the engineering is that a wallet or dapp holds the most provable inventory in the market. When a user has connected a wallet, the impression is not an anonymous pageview an advertiser has to take on faith. It is a verified, onchain-legible audience, and advertisers pay for what they can measure. That provability is what turns a recommendation surface into revenue instead of noise.

Concretely, Specify guarantees a $10 CPM floor on every impression served, regardless of performance, then layers a tiered revenue share keyed to how often your users convert, reaching $80+ eCPM at the top tier. Conversion rates of 0.7% are not rare on the network, which is the kind of number that moves a surface up the tiers. Payment settles monthly within 30 days, and the revenue-share figures are onchain-auditable rather than a number you have to trust. You can put your own user counts through the revenue calculator to see where a surface like yours lands. The wider comparison of how wallet-verified pricing stacks against display networks is in the eCPM breakdown and the publisher network guide.

The users who should never see an ad

There is a discipline here that most networks will not tell you about because it costs them impressions. If a segment of your users already earns you direct revenue, do not request an ad for them. A user paying in-wallet swap fees is already monetized. Showing that user a recommendation does not add a revenue line, it competes with one you already have, and it spends attention you are already being paid for. The right move is to exclude that segment at the request layer and never call for an ad in the first place. Monetization should be additive, filling the surfaces and moments that earn you nothing today, not cannibalizing the ones that already work. A network that encourages you to draw that line is aligned with your product. A network that wants every impression it can get is not.

Who is already doing this

This is not theoretical. The live surfaces running Specify as recommendation placements include Coin98, Blockscout’s 40+ chain explorers, Turtle, Collab.Land, and Outposts, which is a spread of wallets, explorers, and community tools rather than a single category. If you want to see the terms in full, the publishers page has the floor, the tiers, and the integration path, or you can talk to the team directly. The bar to hold them to is the one above: an ad that behaves like a recommendation, an SDK that fails open, and the discipline to show nothing when nothing fits.