Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Affiliate Marketing › Programme Management

Running the Quarterly Affiliate Partner Review

How to run an affiliate programme's quarterly partner review: build one partner table, set aside thin evidence, place partners in tiers, act and record.
Tutorialintermediate16 min readUpdated

Why the review needs its own method

The parts of the review are spread across the section. The reporting and incrementality guide sets the indicators and asks for every active partner to be placed in one of four tiers each quarter: grow, maintain, fix or remove. The commission structures guide lists the questions a structure review asks. The programme management hub puts the quarter into the operating rhythm. The diagnoses of commission costs rising, coupon sites taking commission and top affiliates leaving each end by asking for the quarterly review as the habit that stops their problem coming back.

What no page does is run the meeting: which table to bring, what to do with a partner whose figures rest on a handful of orders, who is told what, and what is written down so that the next quarter can be read against it. Without a method the review shrinks to a glance at the league table, and the reporting guide is clear about what a league table of tracked revenue rewards: the partners closest to the checkout.

The review has neighbours it must not be mistaken for. The monthly close validates and approves commission, and this review reads the same data a quarter at a time. An audit checks the whole programme in one pass and suits an inherited programme or one the regular reviews cannot explain. The meetings with key partners, set out in the guide to managing key affiliate partners, take one partner at a time with shared figures and a joint plan. An incrementality test answers a different question, what a partner or partner type added, and the reporting guide asks for one or two a year. A suspected fraud case is its own investigation. This review is the recurring decision about partners and structure, made from one table.

Two things on the site are called tiers, and they are different. The review's tiers are management categories: what happens to a partner this quarter. Commission tiers are rate bands that partners reach, which the commission guide covers. The review may decide to rebuild the second kind, but it keeps them apart.

Before you start

  • Validated, deduplicated figures by partner. Take them from the month-end approvals, not from tracked totals, with the indicators the reporting guide sets, including new-customer share and total cost with fees, bonuses and placement fees included.
  • The credit rule as it stood. If the deduplication and credit rule from the attribution guide changed during the quarter, partner figures changed with it. Note the date.
  • Margin ceilings by segment. Contribution margin by partner type, product category and customer type, from finance. Either the CPA calculator or the break-even calculator can convert the margin into a ceiling for commission plus fees. Finance owns the margin figures.
  • The partner record. Type, stage and dates, contact, partner-specific rates, exclusives, promotional rates with their end dates, placements and overrides, open findings from content checks and the case records of any investigation. Add last quarter's tiers and decisions.
  • Test history. Every holdout or test run on coupons, brand terms or partner types, each with its date and result.
  • The terms and the platform's settings. The enforcement ladder, the change-notice rule and the version log, as the terms tutorial sets them.
  • Owners. Who builds the table, who places partners, who may change a rate or end a partnership, and who speaks to partners. Write it down before the meeting rather than during it.

The steps

  1. Fix the quarter, the basis and the tier rules before reading any name

    Choose the quarter's start and end dates to match the finance calendar and read validated figures, not tracked ones. Conversions still inside the validation period when the quarter ends are shown apart, as the month-end tutorial keeps pending commission apart, rather than counted as won or left out.

    Then write the tier rules while the table still has no partner names beside it. For each tier, state the indicators that place a partner there, measured against the programme's own figures and the segment ceilings rather than against another programme. Grow: brings new customers at a cost per new customer inside its segment's ceiling, with returns close to the business's own, and has room to do more, such as an audience that matches the customers the business wants and has barely featured it. Maintain: pays for itself on validated figures, with nothing to change. Fix: raises a question the partner can answer, such as a new-customer share low for its type, returns far higher than the business's own, open content findings or code use outside the coupon clause. Remove: no activity after the agreed follow-up, a breach that survived the enforcement ladder, or no value after a fix.

    Rules written first are harder to bend for the partner who argues hardest or the one everyone likes, and they are changed as a decision at the next review rather than argued partner by partner. The reporting guide names the four tiers but gives no rules for placing a partner in them; writing the rules before any name is read, and the order of this meeting, are this tutorial's own method, offered as general practice. Settle the scope too: active partners are in, partners still inside their first onboarding are out, and partners under investigation wait for the case.

  2. Build one partner table from the approved figures

    List every partner with a validated order or a live link, and give each the indicators from the reporting guide, the same ones the close already produces so that the table agrees with the commission paid and accrued. Add the partner's type, its stage and dates, any partner-specific arrangement with its review date, and open findings. Show orders credited to a partner's own code with no click on their own line, so that code-only commission is not buried in the clicked total, as the guide to crediting orders by code asks.

    Sort by new customers and contribution after commission, not by tracked revenue, as the Pareto principle page ranks affiliate partners, and add a running total down the list. Where the total climbs steeply the programme is concentrated, and the same ranking shows which partners the key partner list should hold. Keep the partners with no click this quarter in a separate roster with their stage dates, laid out as the affiliates not generating sales diagnosis lays out its count, so that the share of approved partners who ever placed a live link, the activation rate, sits beside the table.

  3. Set aside the partners the numbers cannot judge

    The small numbers guide gives the habits for thin counts, and this review applies them to partners: a minimum count below which a partner is not judged alone, and a comparison of each partner with the pooled figure for partners of its own type rather than with the whole programme.

    Partners below the minimum keep their place and are marked as too few to judge; they are read next time over a longer window. A partner whose trade is seasonal, such as a gift guide, is read against the same quarter a year earlier where the history exists. Partners still inside onboarding belong to the onboarding tutorial and its first-link and first-sale check, and partners with an open case wait for the investigation.

  4. Read each remaining partner against the ceiling and the quality signals

    Start with the segment, read against the margin ceiling finance gave. Then use the commission guide's review questions as the agenda.

    For each partner, set new-customer share, cost per new customer and order value beside the pooled figures for its type. Read the return rate before any partner is placed in grow: the reporting guide treats a return rate far above the business's own as a possible sign of misleading promotion, incentivised orders or fraud. Look at baskets too, since a partner whose customers buy only discounted lines, or only with a code, can cost more than it appears to. The new-customer flag comes from the order system, defined in the way set out by the new and returning customers guide, and not from a partner's own claim.

    Where a test exists it outranks the table. The table shows credit and quality; it cannot show what a partner added, and the sales objective page is explicit that a partner group which fails an incrementality test loses its higher rates.

  5. Place each partner that can be judged in a tier

    Apply the rules written in the first step, one partner at a time, and write the reason beside each placement in a single line that names the indicator. A partner in grow gets more support and, where it brings customers the business values, a higher commission rate through the structure with a recorded reason and review date, not as a favour. A partner in maintain is paid for results as now. A partner in fix gets a named question. A partner in remove is dealt with under the terms.

    Two checks on the result. If nearly every partner lands in one tier, look at the rules before the partners. If the grow tier fills with checkout-stage partners, the indicators are rewarding volume, the cause the coupon diagnosis finds in tiers based on volume alone. Also keep the review's remove tier apart from the dormant mark that onboarding uses. A dormant partner can stay on the books where the terms allow, which leaves a seasonal offer able to reach it, and the review decides whether it stays or goes.

  6. Check special deals, placements and the key partner list against the table

    Go through every partner-specific rate, exclusive, promotional rate, override and paid position. Has its review date come? Does the reason recorded when it was agreed still hold? The commission guide asks for that record to stop old arrangements piling up with no one able to say why they exist. Confirm that promotional rates stopped at their end date in the platform, a fault the promotions guide lists among its mistakes.

    Before ending a special arrangement, apply Chesterton's fence: find out why it exists. Read the record kept when it was agreed, then ask the person who agreed it, then the partner. If the reason is gone, end it with the notice the terms require. If it still holds, keep it and set a new review date. For paid positions, the placements guide says a rebooking needs a comparison of total orders; where the first booking had none, set a test up before rebooking.

    Then hold the key partner list against the table. Does it still name the partners who bring the most new customers and margin? Which partners are climbing towards it? The key partners guide covers what to do when one partner's share grows large. Key partners are placed by the same rules as everyone else; their own review stays separate.

  7. Decide an action for every partner in fix and remove

    Give each partner in fix one written ask: what the table or a finding shows, what the business needs instead, a date and an owner. Route it by cause. A claim, a price or a disclosure goes to the content check tutorial. Orders carrying codes the partner was not issued go under the coupon clause and the code register. Returns above average prompt a look at what the partner's pages promise. A pattern that looks like abuse goes to the fraud investigation tutorial, and the review does not decide it. After the date, check the item and take the next rung of the enforcement ladder if it has not been fixed.

    For a partner in remove, follow the terms: the clause relied on, the date it takes effect and what happens to commission still pending. Close the lines for codes issued to the partner in the register, end its special arrangements, and ask for brand assets to come down, as the asset withdrawal tutorial describes. Record the evidence beside the decision, because it has to stand up if the partner or the platform challenges it. For inactive partners, decide between keeping them dormant and removing them.

  8. Decide any change to the structure and send it through the change gate

    Change the structure only where the table supports it. Follow the commission guide's advice on how to change a structure and, as general practice, change few things at a time so that the next table can show what each did. Where the real doubt is whether a partner type adds orders at all, schedule a test, since a table of validated figures cannot settle it.

    The affiliate commission change checklist is the gate between the decision and the partners: it covers the rebuilt economics, the notice period the terms commit to, the settings in the platform and the partners themselves. Terms follow the change-notice rule and the version log. If the review keeps finding the same faults, or cannot explain what the programme costs, stop and run the audit instead.

  9. Tell each partner what applies to them, with notice

    Key partners are told first, by their named contact, ahead of any general message. Then each partner gets one message about its own placement. Grow: what the brand will provide and the date any rate change applies to tracked conversions. Fix: every finding, the ask, the date and the next rung of the ladder. Remove: the clause and the date. Partners in maintain need nothing beyond routine notices. No change reaches the platform before the notice date. A key partner who first meets a change as a statement that no longer matches what they accepted is one of the ways a programme loses the partners it most wants, as the diagnosis of top affiliates leaving shows for rates and terms changed without notice.

  10. Record the decisions and read the next quarter against them

    Save the table as it stood, the tier rules, each decision with its reason, evidence, owner and date, each change with the date it takes effect, and what the next review will check. A row of the marketing decision log template carries the evidence, reversibility, expected effect, guardrail and check date that a decision log needs, and fits these decisions.

    Open the next review with last quarter's decisions. Did the partners in grow bring more new customers at a cost inside the ceiling? Did the partners in fix fix? Did the segment's cost per new customer improve while total new customers held, the reading the coupon diagnosis gives for rebuilt tiers? Did a rate change move other partners? Judge on a longer window, since one quarter can flatter or punish a partner by chance. A partner who stays in fix quarter after quarter, or who keeps moving tier, may point to a fault in the rules rather than in the partner.

The review at a glance

TierWhat the table showsWhat the review doesWho is told, and howWhat the next review checks
GrowNew customers at a cost inside the segment ceiling, returns close to the business's own, room to do moreMore support; a rate change through the structure where a reason is recorded; a place on the key partner list consideredKey partners through their named contact first, then a notice with the date any change appliesWhether new customers and cost per new customer moved as the reason expected
MaintainPays for itself on validated figures, no open questionsNothing beyond the recordRoutine notices onlyThe same indicators over a longer window
FixA question the partner can answer: low new-customer share for its type, returns well above average, open findings, codes outside the clauseA written ask with a date, routed by cause; the next rung of the ladder if unansweredThe partner, once, with every findingWhether each item was fixed; the ladder step if not
RemoveNo activity after follow-up, a breach that survived the ladder, no value after a fixApply the terms; close code lines; ask for assets to come down; end special arrangementsThe partner, with the clause and the dateThat nothing of the partner's is still live and the record is complete
Too few to judgeBelow the programme's minimum countLeave in place; read over a longer windowNobodyWhether the volume now supports a placement

What the review can and cannot show

The table reads credit and quality from validated, deduplicated figures. It cannot show what any partner added, which only a test can, and the reporting guide's coupon, brand-term and partner-type tests are the tools for that. A tier is a decision about this quarter, not a verdict on a partner's worth, and a partner's own plans, a falling audience or a change of business model can reach the brand through conversation rather than through a table.

The review is also only as steady as the rules underneath it. A change to the credit rule, the window, the validation period or the new-customer definition changes the figures partners are read on, so the record notes when each changed. Finance owns the margin figures and their accounting; nothing here is accounting advice. The review does not set commission by itself: it decides what to put through the change gate, and the checklist and the terms decide when partners see it.

A skate seller runs its first quarterly review (illustrative)

Illustrative example — not a real result

Pennisket Skates reviews its programme

Pennisket Skates, an invented seller of ice skates and blade-care accessories through its own online shop, has 186 approved partners. In the quarter 71 placed a click and 38 had validated orders: 960 orders and about £141,000 of validated revenue after deduplication. Finance gives a margin ceiling for each partner type, and the team writes its tier rules before building the table. From last year's quarters it sets a minimum of 12 validated orders before a partner is judged alone.

GroupPartners with salesValidated ordersNew-customer shareReturnsCost of sale against ceiling
Content and review sites1141064%4%Inside
Voucher and cashback sites736015%7%Over on returning customers
Creators914048%6%Inside
Gift guide and comparison sites115030%5%Inside

The business's own return rate is 5%. Twenty-four of the 38 partners fall below the minimum, including every gift guide and comparison site, whose best quarter is the winter one; they stay where they are, flagged to be read again over two quarters. Of the 14 judged, four content sites and one creator go to grow, five go to maintain, three go to fix and one goes to remove. Two voucher sites go to fix because orders carrying codes issued to someone else sit on their line, and one creator goes to fix because 13% of its orders are returned and its boot page says a blade "never needs sharpening", a claim the brand has withdrawn. The partner for remove is a coupon page that copied the brand's product text and left the last quarter's written request unanswered, and the later rungs of the ladder, commission withheld and then suspension, brought no reply.

The review then finds four special arrangements. A content site's special rate has a recorded reason and a review date and stays. An exclusive code agreed three years ago has no reason on record; asking the founder, who agreed it, shows it was a launch exclusive that lapsed, so it ends at the next notice. A placement on a gift guide had no comparison on its first booking, so Pennisket sets up a holdout before rebooking. The fourth, a promotional rate, stopped on time. The key partner list loses one site whose figures fell and gains a creator.

On structure, Pennisket raises the new-customer rate for content sites and lowers the rate on returning customers' orders for voucher and cashback sites, in that order, and sends both through the change checklist. The two key partners with changes hear by phone before the general notice goes out. Each decision goes into the log with a check date. At the next review the team first reads those decisions against their check dates.

Common mistakes

  • Ranking on tracked revenue. The top of the list fills with the partners closest to the checkout, and the review rewards what the programme least needs.
  • Writing the tier rules after seeing the names. The rules then follow the partners, and the partner with the loudest case or the longest history wins.
  • Tiering partners on a handful of orders. A thin rate swings widely, and the partner at an extreme this quarter will look more ordinary next quarter without any action.
  • Calling a fix a gentle removal. A partner in fix gets a specific question and a date; if nothing is asked, nothing can be fixed.
  • Ending a special deal nobody can explain without asking. The reason may still be live. Find it first.
  • Deciding a fraud case in the review. A pattern that looks like abuse goes to its own investigation, with the commission held under the terms.
  • Changing rates in the platform before partners have notice. The change gate and the terms come first, and key partners hear from their contact before anyone else.
  • Reviewing without acting, or acting on everyone at once. A review that changes nothing is a report, and one that changes everything leaves the next table unreadable.
  • Keeping no record. Without last quarter's rules, reasons and check dates, the next review starts from nothing.

Frequently asked questions

How is the quarterly partner review different from an audit?

An audit is one pass over the whole programme, covering the economics, tracking, terms, partner mix, compliance, fraud controls and reporting, and it suits a programme you have just inherited or one the regular reviews cannot explain. The quarterly review is a recurring decision about partners and the commission structure, made each time from the same validated partner table, with written rules for the tiers and a record that the next quarter is read against. A review that keeps finding the same faults, or cannot explain what the programme costs, is a prompt to run the audit instead.

Should a partner be moved down a tier after one weak quarter?

Not on thin evidence. A rate resting on very few orders swings widely from one period to the next, and a partner picked out because its quarter was extreme will tend to look closer to average next time even if nothing is done. Set a minimum volume from the programme's own history, compare the partner with partners of its own type, read seasonal partners against their own calendar, and move a partner to fix only with a specific question it can answer. A move to remove rests on the terms and on evidence in the record, not on one table.

Do key partners go through the same review?

They sit in the same table and are placed by the same written rules, so the programme holds its biggest partners to the same standard as the rest. Their own meeting is separate, with shared figures and a joint plan, and their contact speaks to them before any general notice goes out. The review is also where the key partner list is tested against the figures: whether it still names the partners who bring the most new customers and margin, and whether any one partner has become too large a share of the programme.

Terms used on this page

Tools to try

Frameworks

Sources

  1. Performance Marketing Association — PMA

Related