Dolphin Solves Access. Here's the Part It Was Never Built to Solve.

Ask ten media buyers what Dolphin campaign management looks like in practice and you get two answers that barely overlap. One group describes profiles, fingerprints, cloud sync and who on the team is allowed to open which account. The other describes budgets, rules, creative rotation and what a week of spend actually earned. Both groups are right. They are describing two different jobs that happen to meet at the same login screen.

This article is deliberately split in half. The first half is what the browser layer solves, described the way its own documentation describes it. The second half is what is still on the desk afterwards — and that half has nothing to do with browsers. If you want the product tour first, our review of Dolphin Cloud covers the plans and the profile model in detail.

What Dolphin Was Built to Solve, and Does Solve

The Dolphin help centre is specific about what the product is for, and it is worth quoting the mechanics rather than paraphrasing them.

Cloud synchronisation stores profile data on Dolphin's servers so a profile can be used from more than one device and shared with a team. The sync happens at a defined moment: profile data is written to the cloud when the profile is closed, while everything that changes during a running session — new cookies, tabs, cache — stays local until then. Turn sync off and the documentation is blunt about the consequence: the data lives only on that machine, and a copy opened on a second device loses its current state. Cloud data is removed on request or automatically after 45 days, and sync can only be switched back on from the same PC where it was switched off. (Dolphin help centre, Profile Cloud Sync, checked 29 August 2026.)

Team handling is equally concrete. There are three roles — Admin, Teamlead and User — and two different gestures for moving a profile: share, where owner and colleague work on the same profile together, and transfer, where the profile leaves the sender's account entirely and the recipient has to accept it. An Admin can cap how many profiles a given user is allowed to create, and when a user is deleted their cloud profiles can be pulled back into the admin account, though local ones cannot. (Dolphin help centre, Teamwork, checked 29 August 2026.)

If you want that mechanism in full — what travels to the cloud, what stays on the laptop, and what a second person actually sees — we wrote it up separately in how Dolphin Anty cloud profiles work.

Read that list again and notice what it is a list of. Identity, isolation, storage location, and who holds the keys. It is a well-built answer to a real problem, and it is the problem an ad platform's API will never touch: an OAuth token knows which permissions you granted, not which machine you granted them from.

The concessions, stated before the argument

Three of them, because an argument that skips the concessions is not worth reading.

• Creative uniqueisation at launch is not something we do. Anti-detect tooling in this category makes each upload distinct at the file level; that capability does not exist on our side, and pretending otherwise would be found out in one demo.
• Identity management is not our layer and we do not want it to be. No fingerprints, no proxies, no profile store. That is the whole point of calling the two products complementary rather than competing.
• Launch rollback is Meta only. If a bulk launch needs to be unwound in one gesture, that gesture exists on Meta and nowhere else in our stack today.

What Is Still Open at Nine on Monday Morning

Here is the part the browser layer was never in the business of solving.

You open the profile. Good — you are in the account. Now you are in one account, on one network, looking at one interface, and the questions that decide the week are all somewhere else:

• What did the other five platforms spend since Friday?
• Which ad set is below break-even once refunds and cost of goods are taken out?
• Which creative started decaying on Saturday, while nobody was looking?
• Where should tomorrow's budget go, and how much of it?

None of those four questions can be answered from inside a tab. Not because the browser is bad at its job, but because the data needed to answer them is held by six different APIs, in six different shapes, and none of it is on screen. This is the ordinary shape of the week for anyone running multiple ad accounts across networks, and it is not a tooling failure — it is simply a second job.

The detail that tells you who built your cross-platform number

Google's own field reference lists metrics.cost_micros as an INT64: cost expressed in micros, a millionth of the account currency unit. (Google Ads API v22 field reference, metrics, checked 29 August 2026.) The other networks do not agree on that convention, and several report money as a decimal in the account currency instead.

If whoever assembled your "cross-platform ROAS" did not know that, you will never see the error. You will see a number — clean, aligned, confident, and wrong by six orders of magnitude on one column. That single detail separates a dashboard built by someone who has bought media from one built from a spec sheet, and it is the kind of thing that only shows up after the profile has done its job.

What Campaign Operations Actually Means, With the Counts

The word "everywhere" is doing a lot of unpaid work in this market, so here are the counts instead. Every row below is the count as built, not as marketed.

(Wevion internal data, August 2026, read from the platform registry and the rule worker.)

Saying "six" where it is six and "four" where it is four is not modesty. It is the only version of the sentence that survives the demo, and telling a prospect the limit before they find it is worth more than the extra platform in the headline.

Wevion is an ad operations platform that connects Meta, Google, TikTok, Taboola, Snapchat and Outbrain through their official APIs with OAuth, syncing insights roughly every fifteen minutes, so bulk launch, rules, budget allocation and contribution margin all sit in one panel instead of six.

Margin Is the Number No Browser Layer Can See

A profile can show you a payment method. It cannot show you a margin, because margin is assembled from three sources that live outside the ad account: platform spend, tracker revenue, and store economics.

Nine of the trackers this market actually runs on have an adapter — Keitaro, RedTrack, Voluum, Binom, BeMob, ClickFlare, Everflow, ExoClick and TrafficManager. They connect; they are not competitors, and a tracker's postback is a source of truth we read, not a rival dashboard. On the store side, Shopify and WooCommerce feed cost of goods, shipping, processing fees and refunds, with refunds attributed back down to the ad that produced the order.

The result is a contribution margin computed per campaign, per ad set and per single ad — spend, gross revenue, net revenue, refunds, cost of goods, shipping, fees, contribution profit, contribution margin, true ROAS and break-even ROAS. Two honest caveats: there is no tax column, so this is contribution margin and not net profit; and the numbers are only as good as the tracker and store you connected. If you want to see what that granularity actually catches, the per-SKU profit leak teardown walks through one.

The payoff is that a rule can then read those columns. profit, profit_margin, true_roas and break_even_roas sit in the same condition list as clicks and CPA, alongside 33 metrics in total — 19 that work on every platform and 14 that are Meta-specific. A rule that pauses on margin rather than on a revenue proxy is a different instrument from a rule that pauses on ROAS, and it is the one thing in this article that has nothing to do with anti-detect browsers at all.

One Daily Budget, Six Platforms, Every Eight Hours

Most automation in this category moves budget one campaign at a time, by percentage, on a schedule you set. That is useful, and it is also not the same thing as allocation.

A budget pool takes a single daily figure, holds N campaigns from N platforms inside it, and redistributes the money toward whatever is performing on your chosen optimisation metric — every eight hours by default, inside per-member floors and ceilings, and with a minimum change threshold so it does not fidget. It runs a simulation before it moves anything and writes an allocation log after, so the question "why is this campaign on a different number than yesterday" has an answer with a timestamp.

That is the mechanism behind the phrase "budget follows performance", and it is worth separating from the rule engine, which is a different tool doing a different job — see what one cross-network rule caught over a weekend for the contrast.

The Brake That Says Someone Has Run This Before

There is a small design decision in the rule engine that a media buyer recognises in about three seconds.

When you stop autonomy — the panic switch — the engine stops activate, stops budget increases and stops relaunch. It keeps running pause and budget decreases. The asymmetry is deliberate: the actions that can lose you money keep working while the actions that can spend money stand down. A symmetric kill switch would freeze the brakes along with the accelerator, which is exactly the wrong behaviour at two in the morning.

Two more details in the same spirit. Budget mutations are idempotent, so a redelivered message does not apply the same increase twice. And the assistant that can act on your account carries spend guardrails — 50 EUR per single action and 200 EUR per day per team by default, with human confirmation on the actions that mutate anything, and the parameters re-read from the database rather than trusted from the model's message.

Two modules — Traffic Filter and Social Media Management — are in early access and switched on per account, on request, so they are worth asking about rather than assuming.

Where the Line Falls

Read the two columns and the conclusion writes itself: there is no row where one replaces the other. Access and identity are one trade; campaign operations across networks are another. The teams that look organised from the outside are usually the ones that stopped trying to make a single tool do both.

Key Takeaways

• An anti-detect browser answers who is opening this account. It was never designed to answer what six platforms add up to, and holding that against it is a category error.
• The second question needs API-level data: launch, sync and measurement across six platforms, budget rules on five, cross-network conditions on four, relaunch and rollback on Meta.
• Margin, not ROAS, is the number that decides the week — and it can only be assembled where tracker revenue, store costs and platform spend meet.
• The details that tell you a tool was built by practitioners are small: micros versus decimals, an asymmetric stop switch, idempotent budget writes.
• Nothing here asks you to give up the profile layer. It asks you to stop expecting it to do a job it never claimed.

Ready to see the second half on your own accounts? Start on the free plan, connect one ad account, and leave Dolphin exactly where it is — the walkthrough is in how the two fit together.