How to Track Influencer Sales: Codes, Links and Attribution

Give every creator a unique discount code and a unique tracking link, and record which one each sale came through. That is the whole mechanism — but both methods systematically undercount, and the gap between what they capture and what a creator actually drove is the reason so many programs get cut on numbers that were never right.

The useful goal is not perfect attribution. It is a number you understand well enough to defend when someone asks whether the £40k went anywhere.

The four methods and what each one misses

Method	Catches	Misses	Best for
Unique discount code	Anyone who types the code at checkout	Buyers who didn't need a discount; people who found a better code	Every program, always
Unique tracking link	Clicks that convert within the cookie window	Link-in-bio friction, cross-device, in-app browsers, iOS restrictions	Blog, YouTube description, newsletter
Post-purchase survey	Self-reported influence, including dark social	Recall error, "saw it somewhere" answers	Sanity-checking the other two
Geo or time-based holdout	True incremental lift	Needs scale and patience; noisy below a few hundred orders	Deciding whether the channel works at all

None of these is the truth. Codes over-credit the discount-hunters and under-credit the people who were persuaded but bought at full price. Links look precise and are the least reliable of the four in practice, because a viewer watching a TikTok in-app, then opening Safari two days later and searching your brand name, leaves no trace. Surveys catch that person and nothing else reliably.

Run codes as the primary system, links as a secondary, and a post-purchase survey continuously. When the survey says 30% of buyers heard about you from a creator and your code data says 8%, the answer is not that one is lying — it's that your codes are capturing roughly a quarter of the effect, and you should say that out loud in the deck.

Codes beat links for social, links beat codes for long-form

Pick one primary method per placement rather than running both and hoping.

Use a code when the placement is TikTok, Reels or Stories. The viewer is in-app, the link is a bio tap away at best, and a memorable code survives the journey to a desktop purchase three days later. Set the discount at a level that motivates without training people to wait for it — 10–15% is the usual band; 20%+ starts cannibalising full-price demand.

Use a link when the placement is YouTube, a newsletter, a blog or a Pinterest pin. These carry a clickable destination in a browser context, the cookie window has a fair chance, and creators there expect an affiliate structure.

Two practical rules that save more money than either choice:

• Make codes creator-specific and non-guessable. SARAH20 will be scraped into a coupon aggregator within a week and every sale it makes afterwards is misattributed to Sarah. SARAHK-AW26 is uglier and holds up.
• Cap or date-limit them. A code with no expiry is a permanent discount you forgot you were funding, still redeeming eighteen months after the creator stopped posting.

Name things once, properly

Attribution decays because nobody agreed on naming. Six months in you have sarahk, Sarah_K, sarah-k-tiktok and sarahktt in four systems and no way to join them.

Fix the format on day one and never deviate:

• Code: HANDLE-CAMPAIGN, uppercase, e.g. SARAHK-AW26.
• UTM: utm_source=creator, utm_medium=influencer, utm_campaign=aw26, utm_content=sarahk. Source stays constant so the whole channel rolls up; the handle lives in utm_content so you can break it out per creator.
• Handle: whatever the creator's actual username is, lowercase, no @. If they're on three platforms, pick one as canonical and record the others against it.

Write the code and the link into the brief itself so the creator receives them in the same document as the deliverable. The most common attribution failure is not technical: it's a creator posting without the code because it arrived in a different email three weeks earlier. Our guide to writing a creator brief covers where these belong on the page.

Do not report a zero you did not measure

The most expensive mistake in this area is reporting 0.0x ROI for a creator on a flat-fee deal with no code and no link. That number is not the finding "this creator didn't work" — it's the fact "we never set up a way to know". Those two lead to opposite decisions, and the first one gets a good creator dropped.

Any deal with a fee and no attribution method attached should be excluded from the ROI calculation entirely and reported separately as unmeasured spend. If that pile is most of your budget, that's the finding worth escalating.

This is a distinction CreatorCast enforces rather than leaves to discipline: a discount code or an affiliate deal type is what makes revenue recordable against a creator at all, deals with neither are held out of the ROI figure instead of being averaged in as zeroes, and the code stays editable after the deal is approved — because sales driven by a code arrive after the content goes live, which is always after approval.

Two more reporting habits worth having:

• Report a window, not a total. "Sales attributed in the 30 days after posting" beats a lifetime figure that keeps drifting upward and can't be compared between creators.
• Separate new customers from repeats. A creator whose code is redeemed mostly by existing customers is giving your own buyers a discount, not acquiring anyone.

Frequently asked questions

Should I pay creators a commission or a flat fee? Flat fees for anyone under roughly 50k followers, because the volume is too low for commission to be worth their time and they will decline. Hybrid — a reduced flat fee plus 10–20% commission — works for creators who have sold before and can see the upside. Pure commission only attracts creators with no other options, which tells you something.

What discount should the code carry? 10–15% for most consumer categories. High enough to be worth typing, low enough that it doesn't become the reason people buy. If your margins won't take 10%, use free shipping or a gift-with-purchase as the incentive instead.

How long should the attribution window be? Thirty days after the post goes live is the practical standard, and seven days catches most of the real effect. Anything longer starts crediting creators for demand they didn't create.

Why don't my tracking link clicks match my code redemptions? They never will. Clicks measure browser sessions; codes measure checkout behaviour, often days later on a different device. Expect codes to outnumber attributable link conversions on short-form video and the reverse on YouTube and newsletters.