How attribution works

Creator Jam credits revenue to your creators in three independent ways: a post-purchase survey, discount codes and affiliate orders. This page explains each one, how a creator’s survey revenue is split between their videos, and what the numbers can and can’t tell you.

At a glance

Survey revenue is an estimate. Shoppers name a creator in your Fairing post-purchase survey. Creator Jam matches those answers to your creators and scales them up for the shoppers who skipped the survey.
Code uses and affiliate revenue are counts. Shopify records each use of a creator’s discount code, and GoAffPro records each order an affiliate referred. Nothing is scaled or guessed.
The three are never added together, because one order can show up in more than one of them.
A creator’s survey revenue is split between their videos. For YouTube creators the split favours the videos people were actually watching right before they bought. For everyone else, it’s an equal split.

Three sources, three kinds of answer

Source	Where it comes from	What kind of number
Fairing	Post-purchase survey answers that name a creator	An estimate: surveyed revenue, scaled up for shoppers who didn’t answer
Code uses	Shopify’s record of each creator’s discount code	A count of orders that used the code
Affiliates	GoAffPro’s record of referred orders	A count of orders GoAffPro credited to an affiliate

The ROI page has a tab for each, and Home shows one tile for each source you choose.

Survey revenue (Fairing)

Which answers count

After checkout, your Fairing survey asks shoppers how they heard about you. Creator Jam finds this main question automatically by its wording, such as “How did you hear about us?”, “How did you find us?”, “How did you discover us?” or “Where did you hear about us?”. If more than one question matches, Creator Jam uses the one with the most answer options.

Creator Jam reads two kinds of answer:

Follow-up questions. When a shopper picks YouTube, Podcast, Instagram, TikTok or Influencer, Fairing can ask which one. Creator Jam reads the follow-up behind any option whose label contains one of those words. Follow-ups behind other options aren’t read. If one follow-up sits behind two options, such as “TikTok Influencer”, its answers are read once, never twice.
“Other” on the main question. Answers of “Other”, or answers with typed-in text, are matched against your creators too.

Each answer brings the value of the order it came with, and counts on the date the shopper answered, which is right after checkout.

Scaling up for shoppers who skipped the survey

Only some shoppers answer, and fewer still get through a follow-up. Creator Jam divides each answer’s order value by an assumed response rate to estimate the full total:

Answer came from	Scaled by	One surveyed order stands for about
Main question (“Other” or typed text)	33%	3 orders
YouTube follow-up	33% × 55% ≈ 18%	5.5 orders
Podcast, Instagram, TikTok or Influencer follow-up	33% × 80% ≈ 26%	4 orders

A follow-up answer is scaled by both rates, because the shopper had to answer the main question first and then the follow-up.

For example, a Ridgeline Coffee shopper spends $60, picks YouTube, then picks Sam Builds Stuff in the follow-up. That adds one survey mention to Sam’s row and about $330 ($60 ÷ 18%) to Sam’s estimated revenue.

Matching answers to your creators

Creator Jam checks each answer against your whole creator list at once. The first rule that fits decides who gets the credit:

#	The answer looks like	Who gets credit
1	A pick from a Fairing list, such as “Sam Builds Stuff (@sambuilds)”	Exactly the creator with that name or handle
2	Exactly one creator’s name or handle, such as “Hannah Brews” or “@kaioutdoors”	Only that creator
3	Free text naming one or more creators, such as “Sam Builds Stuff and Hannah Brews”	Every creator named
4	Free text naming nobody, but containing a handle, such as “the sambuilds channel”	The creator with that handle

A few details make free text more reliable:

Case, spaces and punctuation are ignored, so “hannah-brews” and “HannahBrews” both match Hannah Brews.
Short names must be whole words. A name of five letters or fewer only matches as a whole word, so it can’t hide inside a longer one.
A name inside another creator’s name is dropped. Suppose you also work with a creator named “Sam”. The answer “Loved the Sam Builds Stuff video” credits Sam Builds Stuff only, and “Samantha recommended you” credits neither.
Handles are a fallback. Creator Jam reads a creator’s handle from their YouTube, Instagram or TikTok profile URL. In free text, a handle only counts when no creator is named and the handle is longer than five characters.

Matching against the whole list is what lets Creator Jam tell “Sam” apart from “Sam Builds Stuff”: you can only do that if you know both exist. The ROI page, Home and the per-creator views all credit answers the same way.

Splitting a creator’s credit between videos

The ROI page and Home credit each creator as a whole. When you look at individual videos, in a creator’s Channel Performance panel or a deal’s ROI summary, Creator Jam also has to decide which video earned each answer, because shoppers name the creator, not the video.

The videos that compete for an answer are the creator’s videos that were already out when the shopper answered, within the period you’re viewing.

YouTube creators: recent views win

Splitting by lifetime views over-credits old videos that nobody watches any more. Creator Jam weights each video by the views it gained shortly before the purchase instead. Views gained after the purchase never count, since they can’t have caused it.

For example, a shopper buys today and credits Sam Builds Stuff, who has two videos:

Video	Published	Lifetime views	Views in the last month
Video A	8 months ago	500,000	about 2,000
Video B	3 weeks ago	80,000	about 80,000

By lifetime views, Video A would get about 86% of the credit, even though hardly anyone is watching it. By recent views, Video B gets about 98%. The video people are actually watching gets the revenue.

Other creators: equal shares

For creators on other platforms, such as The Slow Mornings Podcast or Hannah Brews, each episode or post that was out at the time gets an equal share. That’s true even if you’ve recorded view counts for them by hand.

Until a video has view history

Creator Jam records each YouTube video’s views once a day from the moment it starts tracking it. Recency weighting needs at least two of those daily counts from before the purchase for every competing video. If any competing video doesn’t have them yet, that answer is split by lifetime views instead.

View history can’t be filled in for dates before tracking started, so purchases from before then keep the lifetime-views split.

The purchase consideration window

“Shortly before the purchase” depends on what you sell. Someone buying a $15 impulse product behaves differently from someone researching a $400 purchase for a month. The Purchase consideration window is your estimate of how long a typical shopper takes to buy after watching a video.

Creator Jam uses it as a smooth taper, not a cutoff:

A view from right before the purchase counts fully.
A view as old as your window counts about half as much.
A view twice as old counts about a quarter, and so on.

Set it under Settings → Integrations → Attribution Settings. Pick 7, 14, 21, 30, 45 or 60 days, then click Save. The default is 21 days, and the setting applies to your whole team.

Shorter (7 to 14 days) suits impulse or low-priced products that people buy soon after watching.
Longer (30 to 60 days) suits considered or higher-priced products that people research first.

It doesn’t need to be exact, because it sets the timescale and the taper is forgiving. It only changes how a creator’s revenue is split between their YouTube videos. It never changes a creator’s total.

Code uses and affiliates: counted, not estimated

Code uses come straight from Shopify: how many times each creator’s discount code has been used, and the sales on those orders. There’s no scaling and no name matching, because the code belongs to the deal. On the ROI page, the date range picks which deals are included (those published in the range), and each code shows its totals so far.

Affiliates come from GoAffPro, which credits each referred order to an affiliate itself. Creator Jam adds up that revenue as it is, with no scaling, and leaves out orders GoAffPro has rejected. It then links each affiliate to one of your creators. GoAffPro explains how that link is made and how to correct it.

Why the sources are never added together

One order can appear in more than one source. A shopper who uses SAM15 is a code use. If SAM15 is also Sam’s GoAffPro coupon, the same order is an affiliate sale. And the shopper may name Sam in the survey too.

A blended total would have to guess at that overlap. Three honest numbers you can compare are more useful than one that quietly double-counts, so the ROI page and Home always show them side by side.

They also answer different questions. The survey estimates everyone a creator influenced, including shoppers who never touched a code or link. Codes and affiliates count only the orders that came through one. Expect a creator’s affiliate revenue to be lower than their survey revenue. That gap isn’t an error: it’s the influence that can’t be traced to a code or link.

How to read the numbers

Treat survey revenue as directional. It’s built from a sample, so it’s useful for comparing creators and spotting trends, not accurate to the dollar. Code uses and affiliate revenue should match what Shopify and GoAffPro report.
Prefer longer date ranges. Spend counts by publish date and revenue by order date, so a video published on the last day of a month carries its full cost against one day of sales. Thirty or ninety days gives a steadier ROAS than seven.
“—” isn’t zero. It means there’s no figure, for example no spend to divide by. When Fairing or GoAffPro can’t be read, Creator Jam shows an error rather than $0.

Known limitations

Survey numbers are estimates. They’re scaled up from the shoppers who answered, so treat them as close, not exact.
The response rates are assumptions. Creator Jam uses the rates in the table above rather than measuring your survey’s own. If more of your shoppers answer than that, the estimates run high, and if fewer answer, they run low.
Matching needs the name or handle as written. Case, spacing and punctuation don’t matter, but a misspelled name isn’t matched. A creator also has to exist as a channel in Creator Jam to get credit.
Views stand in for influence. Recency weighting assumes views turn into sales at a steady rate. A creator’s core audience, who watch early, may buy differently from later, more casual viewers.
It needs view history. New videos use the lifetime-views split until enough history builds up.
Recency weighting is YouTube only. Other platforms split credit equally.