E-commerce Marketing and Growth · Affiliate and creator programmes · lesson 17 of 20 · 10 min

Measuring and managing partner programmes

Beyond last-click commissions

Affiliate and creator programmes are often measured by last-click attribution: whoever's link or code was used last gets credit. This is simple for paying commissions, but it can reward partners who intercept sales rather than create them. Good programme management looks at incrementality and quality.

Key metrics

| Metric | Why it matters | |---|---| | Revenue and orders by partner | Basic performance | | New-customer share | New customers are usually more valuable than existing ones | | AOV and return rate by partner | Quality of customers referred | | Contribution margin after commission | Profitability | | Conversion path position | Did the partner introduce the customer or only close the sale? | | Customer lifetime value by partner | Long-term value of referred customers | | Content performance | For creators: engagement and ad performance of licensed content |

Assessing incrementality

Methods, from simple to rigorous:

Path analysis: how often is the partner the first touchpoint versus the last? Partners that appear mostly at the final step (for example, coupon sites) may be less incremental.
New-customer analysis: share of referred customers who had never purchased before.
Pause tests: temporarily pause a partner type (for example, coupon sites) and observe total sales and conversion.
Geo or time-based holdouts for large creator campaigns.
Post-purchase surveys: "Where did you first hear about us?"

Commission optimisation

Use insights to reward value:

Higher commission for new customers.
Lower commission (or none) for existing customers or for coupon-driven sales.
Tiered commissions for top-performing partners.
Bonuses for launching new products or content creation.
Adjust commission by product margin.

Partner relationship management

Partners are people and businesses. Strong programmes:

Communicate regularly (newsletters, product updates, upcoming campaigns).
Provide assets early before seasonal peaks.
Pay on time and transparently.
Share performance data so partners can optimise.
Enforce terms consistently (disclosure, brand bidding, claims).

Programme governance

Monthly:    Review partner performance, new-customer share, fraud signals, disclosure compliance
Quarterly:  Re-evaluate commission structure; recruit new partners; remove poor fits
Annually:   Review programme strategy against overall marketing mix and MER

Worked example: restructuring an affiliate programme

A UK fashion retailer finds that coupon sites generate a large share of affiliate revenue but mostly from returning customers already in checkout, while content partners bring most new customers. It:

Reduces coupon-site commission and restricts codes to exclusive offers.
Raises commission on new-customer sales for content partners and creators.
Runs a pause test on one cashback partner for four weeks; total sales barely change, confirming low incrementality.
Reinvests savings into creator partnerships with content licensing.

Total affiliate cost falls while new-customer acquisition from partners rises (illustrative).

Hands-on: a partner quality report

Commission reports rank partners by revenue. A better report ranks them by incremental, profitable value. Export orders with partner attribution, customer type and returns:

import csv
from collections import defaultdict

stats = defaultdict(lambda: {"orders": 0, "revenue": 0.0, "new": 0, "returned": 0, "commission": 0.0, "code_only": 0})
with open("partner_orders.csv", newline="", encoding="utf-8") as f:
    # columns: partner, order_value, is_new_customer (1/0), returned (1/0), commission, via_code_only (1/0)
    for r in csv.DictReader(f):
        s = stats[r["partner"]]
        s["orders"] += 1
        s["revenue"] += float(r["order_value"])
        s["new"] += int(r["is_new_customer"])
        s["returned"] += int(r["returned"])
        s["commission"] += float(r["commission"])
        s["code_only"] += int(r["via_code_only"])

print(f"{'partner':18s} {'orders':>6s} {'new%':>6s} {'ret%':>6s} {'code%':>6s} {'comm/new':>9s}")
for p, s in sorted(stats.items(), key=lambda kv: -kv[1]["revenue"]):
    new_pct = s["new"] / s["orders"]
    cost_per_new = s["commission"] / s["new"] if s["new"] else float("inf")
    print(f"{p:18s} {s['orders']:6d} {new_pct:6.0%} {s['returned'] / s['orders']:6.0%} "
          f"{s['code_only'] / s['orders']:6.0%} {cost_per_new:9.2f}")

Look for partners with low new-customer share, high code-only attribution (often coupon sites intercepting checkout) and high return rates, and compare commission per new customer with your other acquisition channels.

Incrementality options

| Method | How | Good for | |---|---|---| | Geo or time holdout | Pause a partner type in some regions or weeks; compare with matched controls | Coupon/cashback sites, large partners | | Code-leak test | Replace a leaked code with a new unique code; watch whether sales follow the partner or the code site | Creator codes on coupon sites | | New-customer-only commission | Pay only (or more) for new customers | Reducing paid-for existing demand | | Post-purchase survey | "Where did you first hear about us?" | Directional view of creator/content influence |

Common mistakes

Judging partners only by last-click revenue.
Paying full commission on returned orders.
Neglecting partner communication.
Inconsistent enforcement of terms.
Treating the partner channel in isolation from overall marketing.
Changing commission terms suddenly without notice, which damages trust with good partners.
Relying on a single large partner for most partner revenue, creating concentration risk.
Ignoring partner feedback about assets, landing pages or tracking problems that reduce their conversion rates.

A partner performance report template

| Partner | Type | Orders | Revenue | New-customer share | AOV | Return rate | Commission | Contribution after commission | First-touch share | Notes |
|---------|------|--------|---------|--------------------|-----|-------------|------------|-------------------------------|-------------------|-------|

Review the report monthly. Sort by contribution after commission and new-customer share, not revenue alone. Flag partners whose return rates or disclosure compliance cause concern, and agree actions with them.

When to end a partnership

Partnerships should end when a partner repeatedly breaches terms (undisclosed promotion, brand bidding, misleading claims), when incrementality tests show little added value, or when their audience no longer fits your customers. End them professionally, pay commissions owed under the agreement, and document the reasons.

Partner programme checklist

[ ] New-customer share and margin tracked by partner
[ ] Incrementality assessed (paths, pauses, holdouts, surveys)
[ ] Commission structure rewards incremental value
[ ] Regular partner communication and assets
[ ] Terms and disclosure enforced consistently

Video lecture: Measuring and managing partner programmes

Lecture coming soon · 15 chapters · about 8 minutes. Read the full transcript below.

Measuring and managing partner programmes
Why it matters
Metrics beyond commission
Incrementality
Commission optimisation
Relationship management
Attribution models
Governance
Simple example: UK beauty code leak
Realistic example: Gulf electronics cashback (illustrative)
Watch me do it: partner quality report
A partner report template
When to end a partnership
Common mistakes
Recap

Lecture transcript

Measuring and managing partner programmes

Your affiliate dashboard says partners drove twenty percent of revenue last quarter. Here's the question that matters: how much of that would have happened anyway? Partner programmes are often measured by last-click commission reports, which reward whoever touched the sale last, not whoever created it. In this lecture you'll learn the metrics that matter beyond commissions, how to assess incrementality, how to optimise commissions, how to manage partner relationships and governance, and when to end a partnership. Then you'll watch me build a partner quality report.

Why it matters

Why does this matter? Because commission is real money, and last-click attribution systematically overpays some partners and underpays others. Coupon and cashback sites often appear at checkout, when the customer was already buying, and collect commission on sales you'd have made. Content partners and creators often introduce new customers early in the journey and get less credit. If you don't measure properly, you end up funding the wrong partners and starving the right ones.

Metrics beyond commission

Metrics that matter. Revenue and orders, yes, but also the new-customer share for each partner. Return and cancellation rates. The share of orders attributed only through a discount code, which often signals coupon leakage. Commission per new customer, compared with your other acquisition channels. Contribution margin after commission and returns. And the repeat purchase rate of customers each partner brings. Together, they show which partners create profitable growth.

Incrementality

Assessing incrementality. The gold standard is a holdout: pause a partner type in some regions or weeks and compare with matched controls. A code-leak test: replace a leaked creator code with a new unique one and see whether sales follow the creator or the coupon site. Paying commission only, or more, for new customers reduces paying for existing demand. And post-purchase surveys, asking where did you first hear about us, give a directional view of which creators and content partners introduce people to your brand.

Commission optimisation

Commission optimisation. Use tiers: higher rates for partners who bring new customers, lower rates for coupon and cashback sites. Pay on net sales after returns. Reward content partners with fixed fees or bonuses for strong content, not just last-click commission. Offer performance bonuses for hitting new-customer targets. And review rates quarterly with data, rather than raising them whenever a partner asks.

Relationship management

Partner relationship management. Treat top partners like real business relationships. Give them a named contact, early access to launches, exclusive content and fast payments. Share useful data, like which products their audience buys. Ask for feedback on your offers and assets. And segment your partners: nurture the top performers, develop promising ones, and review or remove those who don't add value or break your rules.

Attribution models

Let's talk about attribution models, briefly. Most affiliate networks use last click: the partner whose link was clicked last before purchase gets the commission. Some offer rules that split credit, or give priority to a content partner who introduced the customer earlier within a window. None of these models measures true incrementality, but they can make payouts fairer. A practical approach: keep last click for simplicity, but add rules. Pay more for new customers, pay less when a coupon or cashback site is the final touch after a content partner, and deduplicate against your own paid search and email, so you don't pay twice for the same order.

Governance

Programme governance. Keep clear terms and enforce them: disclosure requirements, banned practices like brand bidding and cookie stuffing, approved claims and return-based reversals. Monitor partner content and paid search for trademark misuse. Audit a sample of partner orders for fraud every month. And keep a written record of any rule breaches and the action taken, so decisions are consistent and defensible.

Simple example: UK beauty code leak

A simple example. A UK beauty brand's creator code, meant for one creator's followers, starts appearing on coupon sites. Commission for that code soars, but the creator's content views haven't changed. The brand runs a code-leak test: it gives the creator a new unique code and deactivates the old one. Sales through the new code fall back to a level that matches the creator's real audience, and coupon-site traffic drops away. The brand keeps the creator, and tightens code rules for everyone.

Realistic example: Gulf electronics cashback (illustrative)

Now a realistic scenario with illustrative details. A Gulf electronics retailer pays about a third of its affiliate commission to two cashback sites. The partner quality report shows those sites bring few new customers, most orders are attributed by code only, and many customers were already in the checkout. The retailer runs a four-week geo holdout, pausing cashback partners in one emirate and comparing with a matched one. Total sales barely change. So they cut cashback commission rates for existing customers, keep them for new customers, and move the budget to content partners and creators who introduce new buyers.

Watch me do it: partner quality report

Watch me build a partner quality report. I export last quarter's partner orders with the partner name, order value, whether it was a new customer, whether it was returned, the commission, and whether attribution came only through a code. A short Python script groups by partner and prints orders, new-customer percentage, return percentage, code-only percentage, and commission per new customer. Sorted by revenue, the top partner looks great. But its new-customer share is low, code-only share is high, and commission per new customer is far higher than our paid social cost. Meanwhile, a smaller content partner brings mostly new customers at a low cost. The report changes the conversation.

A partner report template

Here's a partner performance report template worth adopting, reviewed monthly and discussed quarterly. For each partner: orders, revenue, new-customer share, return rate, code-only share, commission paid, commission per new customer, and repeat purchase rate at ninety days. Then a notes column with context, like a creator's big video or a leaked code. At the top, a programme summary: total commission as a share of revenue, the share going to new customers, and the results of any incrementality tests. Present it next to your other acquisition channels, so leadership can see whether partners are a cheaper or more expensive way to find new customers.

When to end a partnership

When should you end a partnership? When a partner breaks disclosure or brand-bidding rules repeatedly. When fraud is confirmed. When a holdout shows little incremental value and commission can't be restructured. When content or claims conflict with your brand or the law. Or when the audience fit has changed. End it professionally, with notice according to your terms, final payment for legitimate sales, and a clear reason.

Common mistakes

Common mistakes. Ranking partners by last-click revenue alone. Paying commission on returned orders. Ignoring code leakage. Never testing incrementality. Flat rates for every partner type. No governance, so brand bidding and weak disclosure go unnoticed. And rewarding partners who intercept rather than introduce.

Recap

Recap. Measure partners by the profitable, incremental growth they create: new customers, returns, code-only share, commission per new customer and repeat rates. Test incrementality with holdouts, code-leak tests and new-customer-only commissions. Tier commissions, pay on net sales, manage relationships, enforce governance, and end partnerships professionally when needed. Try this now: run the partner quality report from the lesson text on last quarter's data, and pick one partner type for a holdout test.

Key takeaways

Last-click commissions can reward partners who intercept rather than create sales.
Track new-customer share, margin, returns and lifetime value by partner.
Assess incrementality with path analysis, pause tests, holdouts and surveys.
Adjust commissions to reward incremental value and enforce terms consistently.

Try it

Analyse (or design) a partner performance report with new-customer share, margin after commission and incrementality indicators, and propose one commission change.