Xoxo, everyone — it’s Reznik, and today we’re going to split CPA networks into a neat dichotomy, so get your Occam’s razor ready.
From the average person’s point of view, pretty much all affiliate networks look the same: you pour traffic and get paid for conversions … or for leads. And that’s exactly where an interesting aspect comes in — one that rarely interests affiliates, advertisers, or network owners, but is a real pain in the ass for technical staff.
Important! Everything described in this article, at the time of publication, is not some established, textbook truth of the CPA community. It’s nothing more than the author’s personal take on an important technical aspect of how affiliate networks function.
LD;DR: Summary
All affiliate networks can be split into two groups based on which entity underlies how they work: the conversion or the lead. The American (Western) model works with conversions, the European (Eastern) model works with leads.
What Is a Conversion?
A conversion is a lightweight entity that reflects some action subject to payment. It really is a very light entity — as a rule, it only carries this set of data:
- Date and time. No going without these — the entire stats are built on them. That said, the exact time a conversion was generated might not even be tracked if the statistics are built on a day-by-day basis.
- Offer. A conversion can’t exist on its own — if it came into being, it’s definitely tied to some offer. Even if there’s only a single offer on the whole network, it’s still tied to that one.
- Price. The third required parameter a conversion can’t exist without. After all, a conversion is first and foremost a payment for a completed action. Sometimes a zero one …
- Status. Quite often a conversion doesn’t just exist, it sits in one of a few basic statuses: pending, accepted, or rejected. And only an accepted conversion actually gets paid.
- Goal. If an offer has several payable actions available, each one is a goal. The payout is calculated based on which goal the conversion was earned for.
All a network really needs to know about conversions is how many there are and their total amount. No single specific conversion matters much on its own. They’re tracked purely by quantity, never by quality — and that’s the single most important thing about them.
What Is a Lead?
Compared to a conversion, a lead is a much, much fatter entity. You get one by putting a conversion on a diet of donuts and burgers. And no diet coke, either!
The key, most important difference from a conversion: a lead isn’t just a digit in a column of some stats table, it’s the full reflection of a real order being carried out by a buyer, or an action being performed by a user. It always contains the complete set of information needed to process it.
A lead inherits the conversion’s entire data set, and can additionally include:
- Customer data. Usually the most important piece is the phone number, which the call center uses when processing the lead. This also includes the buyer’s full name and email. If it’s a service registration, a login and password can end up here too.
- Delivery address. An essential part of the commodity business. When calculating statistics, at the very least the buyer’s country gets pulled from the address.
- Delivery status. Also directly tied to the commodity business. Package pickup-rate statistics are built from the delivery status. In some cases, the lead’s payout itself depends on exactly this status.
- Order contents. This can take all sorts of forms. It might be just a total sum, if we’re talking about a simple service or, say, a casino payment. Or it might be a full shopping cart — an expanded order listing specific items, their unit price, and quantity.
Looking over this whole set of fields, we can immediately spot another important aspect: unlike a conversion, a lead is a dynamic entity — it can change over time. It gets some of its fields right away on creation, and picks up the rest over the course of its life. A conversion, on the other hand, can only ever change its status, and even that not always.
Speaking of statuses. A lead can boast a whole bunch of interesting, none-too-obvious states that can easily confuse the uninitiated. So here’s what’s new:
- Trash. A special cancellation status that separates “cancelled” real leads from “cancelled” fake ones. And yes, trash only applies to leads — a conversion can’t be trash by definition, because it has no concept of validity in the first place, for lack of any actual data that would need validating.
- Cancellation reasons. These can either be their own separate statuses, or just extend “Cancelled” and “Trash.” You’ll find both technical reasons in here, like an incorrect phone number, a wrong country, or a duplicate order, as well as everyday human quirks along the lines of “Changed my mind” or “Too expensive!”
- Call center. This is a whole group of statuses covering all sorts of notes about calls to customers — missed calls, requests to call back, or scheduling a confirmation call right before shipping.
- Warehouse. Similarly, this forms a group of statuses covering the process of packing parcels, sourcing items, prepping for shipment, and so on.
- Delivery. This group typically holds the statuses for the actual delivery and parcel-tracing process, plus the final statuses for processing commodity traffic — “Paid” and “Returned.”
- Deals. This group of statuses reflects work with info-products and services — running negotiations, preparing documents, verifying payment for franchises, courses, real estate, or some other serious piece of property.
This whole variety of statuses is what makes different payment models possible. This is exactly where the branches of classic CPA come from — things like CPS, CPL, or CoD — paying not just for a confirmed action (a conversion), but for a completed deal, a confirmed order, simply a valid order coming in, and so on. Conversions, on the other hand, only ever have one model to work with — CPA, no exceptions. Again, simply because there are no statuses to speak of besides a confirmed and a rejected conversion.
The American, or Western, Model
Obvious, even without Captain Obvious: since these two entities are so clearly different, they must each have their own turf where they feel right at home.
Conversions became the foundation for offers where no single specific action matters all that much. First and foremost, that’s the whole vast field of games and apps, mobile or desktop alike. Paid actions in apps fit this concept perfectly — registering in a game, reaching some level, killing a monster, or simply installing the app doesn’t call for any extra data.
Actions on serious third-party resources, like major online stores and web services, fit this model the same way. Here too, it’s enough to record the plain fact that some user action took place, and slap a price tag on that action.
Under these conditions, the entire real-world process behind a conversion stays hidden from the affiliate network. The network simply sends the visitor to the target resource carrying its own unique ID, and in return asks that resource to send back notifications about the results. In the end, all that’s left for the network to do is count up all those conversions and add them together.
I call this way of working the American, or Western, model, since it’s most widespread to the west of Greenwich. That’s probably also where it was born. It’s used in CPA platforms like HasOffers, Affise, and Cake, and in most American affiliate networks.
The European, or Eastern, Model
Leads found their place in the sun in the commodity business. With this model, the affiliate network carries far more responsibility for handling each individual lead. It needs to be cherished and nurtured, grown all the way up into an order or a deal.
Most often, a lead is born on separate sites that belong directly to the network — landing pages built to sell one specific product. Sometimes it comes into the network from outside via an API. Either way, it’s always a full-fledged entity containing everything needed to process it. And process it does — that’s exactly where it heads next. That processing can happen right where the lead landed, or in a third-party CRM system. A chain like that, passing the lead along, can be set up indefinitely.
I call this model European, or Eastern, for the exact same reason — it took root to the east of Greenwich. It’s implemented in my own AlterCPA, in various CRM solutions (LeadVertex, for instance), and in most commodity affiliate networks across the CIS. Some of them, AlterCPA included, even build serious CRMs right into their own platform — MonsterLeads, for one, has a seriously slick dashboard for managing leads.
Pros and Cons
Let’s keep playing Captain Obvious for a bit. Since both models are used so actively, and neither one has won out over the other, it becomes pretty darn obvious that each has its pros and cons, and that’s exactly what carves out these areas of use.
Conversions are perfect precisely because they’re lightweight and simple. You couldn’t invent a better solution for working with apps, third-party services, and websites.
In terms of data structure, conversions slot easily into very fast, static database tables. In terms of performance, without careful, clever engineering, leads will lag behind conversions.
To work with conversions, all you need is to receive minimal notifications from the source resource — a visit number and a conversion price — and those are easy to implement on any resource.
And that’s where the advantages of conversions over leads run out. Put a lead on a strict enough diet, and it eventually degenerates into a conversion. So with the right engineering, all those upsides transfer over to the lead model relatively easily.
One clear and hugely important upside of the lead model turned out to be full-fledged API integrations, which open up huge possibilities for reselling. As described above, leads can travel freely between networks, completely hidden from prying eyes.
Integrations aren’t just about reselling, either. They’re also essential for cloaking. However hard you try, there’s no way to reliably bolt a cloaker onto an external resource, whereas it fits a landing page perfectly. On top of that, a landing page is very easy to swipe, or simply get ready-made from the advertiser.
Because of how fat they are, leads demand far more technical resources — hardware for storing the data, and brainpower for building a data model that won’t crawl like molasses.
Which Approach Is Better?
There’s no clear-cut answer — every approach has its pros, its cons, and its own use cases. Which one you actually work with is entirely up to you, after a careful, thoughtful look at your own subject area.
For AlterCPA I chose to work with leads specifically, because it lets you implement absolutely any interaction scheme you like. At the same time, I put in serious effort to optimize the technical side as much as possible — that’s how leads end up running at the speed of plain conversions.
That’s it from Reznik — xoxo, everyone!