Free affiliate leakage audit
Your network reports what it attributed, not what it caused. Send the export and we will separate the two as far as the data honestly allows.
Send a transaction export from Commission Factory, Impact or whoever runs your programme. We read it and tell you what share of your commission went to partners who met the customer at the checkout rather than bringing them to you. No access to anything, no cost, and nothing to buy at the end of it.
What the report answers
- Where the commission goesEvery partner sorted into coupon, cashback, comparison, paid search or content, with the word in their name that decided it so you can overrule us.
- Click to conversionThe gap between the click we paid for and the sale. Under two minutes is a customer who was already at your checkout and went looking for a code.
- The share worth questioningWhat proportion of commission went to the kinds where the sale may have happened anyway. Industry reporting puts 15 to 25 per cent of affiliate spend on invalid or non-incremental activity.
- What the export could not tell usNamed, not guessed around. A missing click-time column means the strongest signal is unavailable, and we say so rather than substituting a weaker one.
- What to do about itRun a holdout, not a cull. Pause the questionable partners, hold the rest steady, and compare TOTAL revenue against attributed revenue. Cutting a partner who was working is the expensive mistake.
A worked example
Everything below was computed by the same code that would read your file, from an invented export of 632 transactions. No brand’s data appears here and none ever will: publishing a real one without permission would end the business the audit is supposed to start.
87.8%
of commission went to coupon, cashback or paid-search partners
A$6,220.09
across 538 transactions in the period
7
partners, 607 transactions, A$7,088.20 in commission
Where the commission goes
| Kind | Share | Commission | What it means |
|---|---|---|---|
| coupon | 59.1% | A$4,189.77 | Publishes discount codes. Usually meets a customer at the checkout who was already buying. |
| cashback | 19.0% | A$1,344.25 | Pays the customer part of the commission back. The customer seeks it out at the point of purchase. |
| content | 10.4% | A$739.70 | Reviews, guides and editorial. Most likely to introduce a customer who did not know you. |
| search | 9.7% | A$686.07 | Bids on search terms. On your own brand name, it is usually buying a customer you already had. |
| unknown | 1.8% | A$128.41 | Not classified. Name alone was not enough, and guessing would be worse than saying so. |
Click to conversion
The gap between the click you paid for and the sale. A click 120 seconds or less before the purchase is someone who was already at your checkout and went looking for a code. Days earlier is someone who read something and came back. Both times are in every network export, and almost nobody looks at the gap.
| Partner | Kind | Median gap | At the checkout | Commission |
|---|---|---|---|---|
| CouponBirds AU← | coupon | 41s | 201 of 201 | A$2,464.66 |
| Discount Codes Australia← | coupon | 63s | 153 of 153 | A$1,725.11 |
| Cashrewards← | cashback | 87s | 92 of 123 | A$1,344.25 |
| Brand PPC Partners← | search | 88s | 48 of 61 | A$686.07 |
| The Weekend Review | content | 2 days | 0 of 35 | A$493.03 |
| Aussie Money Blog | content | 31 hrs | 0 of 22 | A$246.67 |
| Kelly M | unknown | 17 hrs | 0 of 12 | A$128.41 |
What we will not tell you to do
Cut a partner. Dropping one that was genuinely driving sales is the expensive mistake, and it is invisible until revenue falls. Nothing in the report is called fraud either: that is a legal conclusion and this is a spreadsheet.
Classification is by partner name, which is usually all an export gives. It is a heuristic, it will be wrong sometimes, and every row shows the word that triggered it so you can overrule us.
Run a holdout instead
- Pause the questionable partners for a fixed period. A fortnight is usually enough.
- Hold everything else steady. Change nothing else in that window.
- Compare TOTAL revenue, not attributed revenue.
If total revenue is unchanged, that commission was buying sales you already had. If it falls, those partners were working and the attribution was right.
What to send, and what happens to it
A transaction export
From Commission Factory, Impact, Awin, Rakuten or whoever runs your programme. Any period; three months is plenty.
With click date and time
The one column that matters most. Without it we can still read the file, and the report says plainly that the strongest signal was unavailable.
Nothing else
No access to your site, your network account, your analytics or your customers. A file, once.
We read it and delete it
It is not added to anything, not shared, and not kept. The report is yours; the export goes.
What it costs
Nothing, and there is nothing to buy at the end of it. We would rather show you something true about your own programme than describe what we might do.
If you want the work afterwards, that is Partner OS, and it is priced on its own page. If you do not, you keep the report.