Campaigns and flows

A campaign is one decision, made per click: where does this visitor go? Everything else on it exists to answer that question, or to record what happened after.

#What kind of campaign

Three kinds, chosen when the campaign is created. They share everything that decides a click — alias, tracking domain, traffic source, cost — and differ in two things: where the click goes, and how the win comes back.

Pick the wrong one and nothing breaks — it changes which screens make sense and what the funnel is measured between. The kind can be changed later; switching a campaign to a store replaces its paths with the one path to the store, and asks first.

#Affiliate offer

You are paid by somebody else, for a conversion on their site.

1. The ad points at the campaign's tracking URL. The click is recorded and your rules and paths decide where it goes.
2. If the path has a lander, the visitor sees it first, and its call to action sends them on to the offer — which is how the lander's click-through is measured.
3. The offer is the network's page. When the visitor converts there, the network calls your postback URL with the click id, and the conversion lands on the click.

The Tracking tab hands you three links: the tracking URL for the ad platform, the click-out for your lander, and the postback URL for the network.

#Lead capture page

The win is a form on a page you own — a quote request, a sign-up — and the lead is usually sold on to a buyer.

1. The ad points at the tracking URL. The paths on the campaign's Funnel tab send the click to one of your own landers — rotate two or three form pages against each other and the reports say which one people fill in. Add them as landing pages rather than as direct URLs: a direct URL has no record behind it, so it never appears in Where it fires or in the lander reports.
2. The lander carries the tag. When the form is submitted, it fires the Lead goal: recorded straight away as pending.
3. Whoever buys the lead says later whether it was any good. Their postback approves it with the real payout — that gap is your acceptance rate, beside the Lead goal on the Measurement page.

There is no site to set up. A lead campaign's goals live on a lead page on the Measurement page's Lead capture tab — Lead and Registration to start with — and its pages are simply the landers in its paths. Where it fires there lists every lander your lead campaigns sent clicks to, and says which ones the tag was never seen on: a lander with clicks and no tag loses every lead on it. If different offers need different goals or values, add another lead page and choose it on the campaign's Measurement tab.

#Online store

The win is an order on your own store. There is no offer and no lander — the visitor buys on the page they land on.

1. Choose the store on the campaign's Measurement tab. The store's domain is where every click goes; stores are set up on the Measurement page's Stores tab.
2. The ad points at the campaign's tracking link, on the tracking domain chosen on the Main tab. The click is recorded and the visitor is sent to the store with the click id on the address as cid=.
3. The tag on the store reads cid, so the click and the visit are one. Your goals — product viewed, add to cart, checkout, purchase — fire on the store, and a purchase takes its value from the order. For an amount nobody can edit, post the order from your server, signed; the Measurement page's Install tab sets that up.

Where a platform refuses redirects, the Tracking tab also gives a direct link: the store's own address with the tracking parameters on it, which the tag reads as utm_id. The tracking link takes the store's domain when the campaign is saved, so after changing a store's domain, save its campaigns again.

#On Google Ads and Microsoft Ads

Both run parallel tracking: the visitor goes straight to the Final URL and nothing redirects in between, whichever kind the campaign is. So the Tracking tab gives a Final URL carrying the parameters instead of a tracking URL, and the tag on the page opens the visit. A tracking template is offered as well when the tracking domain is on the same site as the page — it records the click a beat earlier, and the campaign works without it.

#The flow

The routing structure nests one level deeper than a flat list, and the extra level is what lets you split traffic differently for different visitors:

Rules carry conditions. Paths under a rule split the traffic that matched it. Landers and offers under a path are what the visitor is actually sent to.

Two ordering facts do most of the work:

• The first rule that matches wins outright. Not every matching rule — the first. Order them from most specific to least, or the general one at the top swallows everything below it.
• Default paths carry no conditions and take whatever reaches them. Every flow needs one, or a visitor matching nothing has nowhere to go.

#Weights

A weight is a share, not a percentage, and does not need to total 100. Two paths at 1 and 3 send a quarter and three quarters.

The split is random per click, not a round-robin. Over a hundred clicks a 50/50 will not be exactly fifty — that is what random means, and a lopsided split on small numbers is usually nothing. Check it on the clicks log with the Sent to column before concluding a rotation is broken.

Random rather than a counter is deliberate: a counter needs shared state on every request, which is cheap on one server and expensive across an edge network. Random converges to the same distribution and costs nothing, so routing behaves identically wherever it runs.

A weight of zero takes no traffic, which is how you park a lander without deleting it and losing its history.

#Rules

A rule is a set of conditions. These are what a condition can test:

Operators: equals not_equals contains not_contains in not_in greater_than less_than between regex exists not_exists.

in with a list is almost always what you want over several equals rules — one rule saying country in [US, CA, GB] is one thing to read and one thing to edit.

#Saved rules and flows

A rule or a whole flow can be saved to a library and copied into a campaign, so "US mobile" is written once rather than rebuilt by hand for the tenth campaign.

Copied, never referenced. Editing a saved rule does not reach into campaigns already taking traffic on a copy of it. That is the point: the blast radius of a mistyped country list is this rule, not every campaign that ever used it.

The consequence is worth knowing — fixing a saved rule does not fix the campaigns built from it. Those are edited one at a time.

To see routing across campaigns — where does mobile traffic go, what happens to a visitor from Germany — use the In use view under Tools → Rules, which lists every routing rule on every campaign, read-only. A campaign's own routing is written in one place only, the campaign's flow editor: two writers for one structure is how a rule ends up saying different things depending on which screen you opened.

#What a campaign hands you

Three links, each for a different party. Take them from the campaign's Tracking tab rather than assembling one.

The tracking URL is the one worth taking rather than editing: only the version on that tab carries the chosen source's macros, and a macro spelled almost right substitutes nothing at all.

#The funnel

Each campaign has a funnel view: how many arrived, how many reached a lander, how many pressed through, how many converted — each step shown both as a share of the step above it and as a share of the first, so you are not dividing in your head.

It can be split by lander or by source, which turns "the campaign converts at 2%" into which lander is carrying it.

#Deleting, and the bin

Deleting a campaign moves it to the bin, and for 30 days it can be put back as it was.

• Its link stops at once. A deleted campaign is not a paused one: it must not go on spending your money for a month. Pause it instead if you only want it quiet for a while.
• Nothing recorded is lost. Its clicks, sales and costs stay in every report, under the campaign's name.
• Restoring brings the same link back. The path is still yours, so an ad that carries it starts working again.
• After 30 days it is removed for good by a nightly job. You can also empty it yourself sooner.

To open the bin, use the trash icon on the list's toolbar with no row ticked. With rows ticked the same icon deletes them; with none it shows what was deleted, and a number on it says how many are waiting.

A path is never given back. Even after a campaign is removed for good, no other campaign can take its path. A link that was once in an ad must never lead somewhere else.

Offers, landing pages, traffic sources and flows go to a bin too. Nothing empties it on a timer: they stay until you empty it yourself, and reports keep showing their names either way.