Home > Questions and Answers > General Questions

---

Basic model on affiliate marketing (8)

---

Hi. I am building a basic model on affiliate marketing, using the RAND() function to show potential equity curves. What do you think of the following assumptions, please (all averages):

1. Average ROI of a winning campaign: 50%
2. Days to optimise a campaign, or decide to cut it: 10 days
3. Average ROI WHILE optimising for the number of days above: -70%
4. How long a profitable campaign lasts for before dying: 90 days
5. Very important: % of campaigns which are profitable: 20%
6. Total hours work per campaign to get it to profitability, or decide to cut it (incl creatives, LPs, optimising, offer selection, angle selection): 15 hours

Thanks!

---

Those are big assumptions.

I don't think it's possible to answer unless you're specifying a single vertical, traffic source and country even...

15 hours is a lot of time to spend on a campaign to get it profitable. That's almost two full working days.

I think most affiliates -- once they have their niche keyed in -- will spend only a fraction of that.

Likewise, don't think you'll find many affiliates willing to run at -70% avg for 10 days running.

You have to remember that 'affiliate marketing' isn't a tangible business in its own right. You could create a basic model on pushing app offers, or dating leads, or weight loss rebills. But a general affiliate marketing model? Never going to be accurate enough to carry any weight.

---

Thanks Finch. Awesome to get an answer from you. All sounds fair.

I would have thought that creating/uploading/monitoring stats on say 30 different banners and 3-5 landing pages would take a while? (but I guess one gets fast over time).

"You have to remember that 'affiliate marketing' isn't a tangible business in its own right" - you and Lorenzo and some others have produced a profit for several years in a row right? In my mind that's a business, but I understand that pushing an app offer is totally different than a dating offer on Facebook.

---

For the banners, sure, although many affiliates are likely to rip creatives rather than create their own. And if you create your own, they are usually recyclable, which makes them more of a 'one-off' cost, plus the occasional investment translating text for new countries.

Same for landing pages...

We've all seen the most common landing pages that get used in the major verticals. You normally have 3-4 major LP types, which affiliates will then make small edits to depending on their needs.

While I might test 3-5 landing pages, these changes are usually 5 minute edit jobs from my base templates.

It's a business, but I don't think it has a model.

When I look back at the biggest campaigns I've had, each of them could be dissected in to a model that calculates chances of success/failure based on certain metrics. But those models depend on the industry that I'm working in.

In terms of general guidelines like average ROI, average campaign length, days to optimise... I think the only way you get an answer is by narrowing down to:

1. What is being sold, specifically.
2. Where it's being sold.
3. Your budget

Only when you know those three variables can you even start to model a system that would give you a roadmap for what percentages/targets to aim for.

---

Awesome, thanks.

---

The benefit of creating any type of model or a business plan is NOT for its predictive capabilities.

Instead, the main benefit will be that it FORCES you to understand the key drivers of business performance ... i.e. how each metric influences your GP or net income and to what extent. For example, you can start to see what the impact is of getting a 5% higher payout is on your gross profit, the impact of increasing your CVR by 15%, the impact of cutting losers 25% faster, etc.

I have never seen a business plan or model that turned out to be correct 12 months later ... and this applies to Fortune 500 companies as well as scrappy startups. That does not mean that they should not be created.

---

^^What cmdeal said.

Your numbers are broadly correct, I'd say, but the confidence interval bars on them are HELLA wide. For example, #1 is correct for values of "correct" that are at least +100% / -40% depending on the model of AM you're using.

AM isn't a single business model - it's dozens of different ones. High-risk or low risk? Which traffic source and vertical? Direct or brokered? Optimised for growth, speed, ROI, sustainability?

---

Nicely said regarding understanding the key drivers (were you a banker/consultant before?). For financing one's business though one needs an idea of cash requirements and expected volatility - though as Caurmen and Finch said each type of camp has different characteristics so hard to generalise.

---

Home > Questions and Answers > General Questions