The TikTok Shop Affiliate Program Audit: Eight Areas Most Sellers Never Check

A seller running a mid-sized home goods program asked me to look at why her affiliate GMV had been flat for five months despite recruiting steadily. Her creator count had gone from thirty-one to seventy-four. Her commission rates had gone up. Nothing moved.

The audit took about three hours and found the answer in the first forty minutes: she had forty-one creators in her “active” roster. Not seventy-four. Thirty-three of them had not posted anything in ninety days, and eleven of those had never posted at all. Her program was not flat — it was growing a phantom roster while her real program stayed the same size.

The second finding was worse. Her open plan was set at 19%, established eighteen months earlier when she was desperate for pickup, and never revisited. Her recruited creators, the ones she had relationships with, sat at 13% to 15%. She was paying anonymous creators more than people she had spent months developing, and every creator could see the open rate.

Neither problem was visible in her dashboard. Both were obvious in fifteen minutes of looking at the right things. That is what an audit does — it checks the things dashboards do not show you.

This article is a complete audit structure: eight areas, what to check in each, and how to score the result. Run it quarterly. Most sellers find something worth real money the first time.

Why Your Dashboard Will Never Show You This

Understanding what standard reporting omits is the prerequisite for auditing anything.

Dashboards Show Events, Not Absences

Your dashboard reports what happened: orders, clicks, views, commission paid. It does not report what did not happen. A creator who stopped posting produces no events, so they produce no signal. They simply remain in your roster, counted.

This is why the phantom roster problem is so common and so persistent. Nothing in standard reporting flags non-events, so nobody notices them accumulating.

Dashboards Show Aggregates, Not Structure

You can see total commission paid. You cannot see whether that total reflects a sane rate structure or a legacy rate nobody revisited. You can see attributed GMV. You cannot see how much of it comes from creators you have relationships with versus anonymous pickup.

Structural problems — mispriced plans, inverted tiers, expiry lapses — produce numbers that look normal in aggregate. Only disaggregation reveals them.

Reporting Lags Reality

Data reconciles over one to three days. Decisions made on fresh data are decisions made on incomplete data, which is why audit reviews should always use settled figures from at least seventy-two hours prior.

What You Actually Need to Check

Area	Dashboard Shows	Audit Reveals
Creator roster	Total count	Actually active count
Rates	Total commission	Structure sanity
Plans	Attributed GMV	Expiry and sequencing
Attribution	Attributed orders	Undercount magnitude
Samples	Shipments sent	Cost per activated creator

Area 1: Roster Reality Check

Start here, because it is fast and often the most alarming.

The Ninety-Day Activity Test

Export your creator list. For each one, record the date of their most recent video. Anyone over ninety days is inactive for planning purposes, regardless of what your roster count says.

Do this manually for the first audit to understand your data quality. After that, activity monitoring handles it — DAMI tracks posting across your roster precisely so this number is always current rather than something you reconstruct quarterly.

Continuous monitoring is the difference between catching a creator going dormant in week three versus month four. DAMI’s activity monitoring flags declining activity automatically, which is what makes the ninety-day test a live number instead of an annual exercise.

Categorise What You Find

• Active: posted within ninety days
• Dormant: posted sometime, nothing in ninety days
• Never activated: accepted, never posted
• Declining: active but velocity falling

Healthy programs typically run sixty to seventy percent active. Below fifty percent, you have a retention or onboarding problem rather than a recruitment problem, and adding more creators makes it worse.

The Concentration Check

Sort by attributed GMV descending. Calculate what percentage of total comes from your top three, and from your top ten.

If three creators exceed forty percent of GMV, you have dangerous concentration. This does not mean remove them — it means diversification is your most urgent project, and it reframes every other decision you make this quarter.

What to Do With Each Category

Active: evaluate for tier movement. Dormant: reactivation attempt, then reclassify. Never activated: write off and fix onboarding. Declining: diagnose before judging — the patterns in our creator dropout analysis cover most causes.

Area 2: Plan Structure Review

The area where most sellers find money, because legacy decisions accumulate invisibly.

The Rate Ladder Check

List every rate you pay, grouped by plan type. Then check one thing: is any open-plan rate higher than any targeted rate?

If yes, you have an inversion, and it is expensive in two ways. You are overpaying anonymous creators, and every recruited creator can see they are paid less than someone who never spoke to you.

Our plan architecture guide covers correct sequencing in detail — open first, targeted above it, exclusivity highest, launched in that order rather than simultaneously.

Expiry Sweep

Every targeted plan has an expiry date. List them all with dates. Anything within thirty days needs a decision now; anything already expired is silently costing you attribution.

An expired targeted plan is particularly nasty because the creator keeps promoting, keeps driving sales, and attribution goes nowhere. It looks like organic traffic. Nobody notices.

SKU Mapping Verification

For every promoted SKU, confirm it sits under the correct plan in the correct shop. Multi-store operations get this wrong constantly, and the failure is invisible — clicks work, orders happen, credit goes nowhere.

Legacy Rate Identification

Any rate unchanged for over twelve months is a legacy rate. List them with the date set and the creator’s current performance. You will usually find creators still on launch rates despite substantially changed performance, in both directions.

Area 3: Rate Leakage Analysis

Where money goes without producing anything.

Effective vs Headline Rate

Calculate effective commission: total commission paid divided by total attributed GMV. Compare it to what you believe you pay. Sellers routinely discover two to four points of gap, created by rate structure drift across many small decisions.

Commission on Unattributed-but-Caused Sales

You pay on attributed orders only, but creator content drives sales attribution misses. This is not leakage in the strict sense, but understanding the magnitude tells you how much brand-building your creators do that you are not paying for — and therefore how much retention risk you carry.

Our attribution window analysis explains why this undercount exists and how large it typically is.

Sample Waste Calculation

Total sample spend over the period divided by creators who actually posted. This number shocks people. Commonly one hundred fifty to four hundred dollars per activated creator once you count everyone who received product and produced nothing.

Every point of activation improvement reduces this directly. It is usually the fastest cost reduction available in an affiliate program, and it requires no rate changes.

Tooling and Labor

Add up subscriptions and estimate team hours on affiliate work. Divide by orders. Sellers who have never done this consistently underestimate, usually by a factor of two, because the work is spread across many people doing small amounts.

The Leakage Summary

Leak Type	How to Calculate	Typical Finding
Rate inversion	Open rate above targeted	2-6 points overpaid
Expired plans	Plans past expiry date	Silent attribution loss
Sample waste	Sample spend / activated	$150-400 per creator
Legacy rates	Rates unchanged 12mo+	Both over and under pay
Uncounted labor	Team hours x loaded rate	Usually 2x estimate

Area 4: Attribution Integrity

If attribution is broken, every other number in your audit is wrong.

Baseline Checks

• Zero-click outliers: creators with strong views and near-zero click-through. Below roughly 0.5% almost always means a broken link path rather than weak persuasion.
• Plan status: every promoted SKU under an active plan
• Stock status: no promoted SKU stocked out during the period
• Multi-store mapping: SKU, shop and plan all aligned

Reporting Lag vs Real Failure

Check whether low numbers persist past seventy-two hours. Transient dips are lag; persistent ones are problems. Our tracking troubleshooting guide covers the diagnostic sequence.

Undercount Estimation

Compare organic GMV during heavy creator posting windows against your trailing baseline. The delta is unattributed creator impact. You cannot fix it, but knowing it prevents you from demoting creators who are actually your best brand-builders.

What Good Looks Like

No zero-click outliers among your top twenty creators, no expired plans, no stockouts on promoted SKUs, and multi-store mapping verified. If all four hold, your attribution is probably working and your numbers are trustworthy enough to audit against.

Area 5: Content Quality and Volume

Count is not quality, and neither is views.

Volume Per Active Creator

Total videos divided by active creators, monthly. Below one per month means your program is producing very little content per relationship, and the constraint is usually briefing or nudging rather than creator willingness.

Revenue Per Video

Attributed GMV divided by total videos. Track by creator. This is the single most useful efficiency metric in the program and the one almost nobody calculates.

The spread will be much wider than you expect — often tenfold between best and worst in the same category. That spread is where your optimization opportunity lives.

Content Quality Sampling

Manually review ten recent videos. Not for compliance — for whether they would make you buy. Sellers who do this quarterly consistently find that a meaningful share of content technically ships but does not actually sell, usually because the brief was unclear.

Approval Process Timing

If you review content, measure median time from submission to decision. Above seven days, your process is destroying content value, as covered in our content approval workflow.

Area 6: Operational Health

The process layer that determines whether everything above works.

Onboarding Step Coverage

Check your last twenty recruited creators against your onboarding checklist. What percentage received every step — confirmation, tracking, delivery check, brief, three nudges, first-post acknowledgment?

Partial execution is the norm, and coverage correlates directly with activation rate. The full sequence is in our onboarding workflow.

Response Time

How fast does your team reply to creator messages? Pick twenty messages from the last month and measure. Slow responses kill activation and are entirely within your control.

Documentation Coverage

What percentage of your active creators have documented terms — rate, content commitment, usage rights, review date? Below fifty percent means you are vulnerable in any dispute and your renewals restart from zero.

Team Load

Creators per person managing them. Above roughly thirty-five and quality degrades measurably. This is the hidden constraint behind most underperforming programs, and it is invisible in every dashboard.

Area 7: Rate and Deal Structure

Structure Distribution

What share of creators sit on each deal structure — commission only, flat plus commission, tiered escalator? Mature programs use a mix. Programs using one structure for everyone have foreclosed negotiating options.

Negotiation History

Review rate changes over the period. How many were increases? How many were accompanied by something in return — content commitment, exclusivity, rights?

Concessions given freely establish precedent. Our rate negotiation guide covers the trading discipline.

Competitive Position

What are comparable brands offering right now? If you have not checked in six months, you are probably behind and do not know it. Ask trusted creators what they are being offered elsewhere.

Escalator Utilisation

If you have performance escalators, how many creators have triggered them? If almost nobody has, thresholds are set too high and the incentive is decorative rather than real.

Area 8: Strategic Alignment

The final area checks whether the program serves the business.

New Customer Rate

What share of affiliate orders are new customers? This is usually the strongest argument for affiliate over other channels, and if you are not measuring it you are leaving the argument unmade.

Product Coverage

Which SKUs get creator attention versus which you want promoted? Sellers frequently find their entire program concentrated on one or two products while high-margin items get nothing.

Margin Check

Contribution margin after commission on your top affiliate products. If any are negative, you are buying revenue at a loss and should know it. Our margin calculator handles this precisely.

Forecast Accuracy

If you forecast, how far off were you last quarter? Variance above thirty percent means your inputs need work. The structure in our forecasting guide narrows this over time.

Scoring and Acting on Results

The Scorecard

Score each area one to five, where five is healthy. Eight areas, forty points maximum.

• 32-40: healthy program, maintain
• 24-31: functional with specific gaps, prioritise the lowest scores
• 16-23: structural problems, expect a quarter of focused work
• Below 16: consider rebuilding rather than repairing

Do not average and move on. The value is in the lowest individual scores, because those are where money is leaking.

Prioritisation

Rank findings by monthly cost times months they have persisted. A rate inversion costing two thousand monthly for eighteen months deserves attention before a documentation gap with no current cost, even if the documentation gap scores worse.

The Ninety-Day Fix Plan

Month one: fix structural leaks — expiries, inversions, mapping. These are mechanical and fast.

Month two: fix process gaps — onboarding coverage, response times, documentation. These need habit change rather than decisions.

Month three: fix strategic alignment — rate renegotiation, product coverage, tier moves. These need data from months one and two to do well.

Changing everything simultaneously is the most common audit failure. You will not know what worked, and you will annoy every creator at once.

Sequencing also gives your team something achievable rather than an overwhelming list. Three fixes per month, each complete, beats twelve started and none finished — which is what happens when an audit produces thirty findings and nobody prioritises them. Put the tracking in place so the next audit takes ninety minutes instead of five hours.

Re-Audit Cadence

Quarterly. Put it in the calendar with a named owner. Programs that audit once and never again drift back within two quarters, because the drift forces are continuous.

Auditing Across Multiple Stores

If you operate more than one store, the audit has an additional dimension that single-store sellers never encounter, and skipping it produces misleading conclusions.

Audit Each Store Separately First

Run all eight areas per store, independently. Market maturity differs, rate benchmarks differ, and buyer behaviour differs — blending them produces averages that describe no actual market.

This matters particularly for attribution. As covered in our multi-store guide, attributed share varies systematically by market for behavioural reasons rather than quality reasons. A market with lower attributed share is not necessarily performing worse.

Then Check Cross-Store Issues

After per-store audits, check three things only visible in combination:

• Duplicate creators: the same person appearing in multiple stores with separate records
• Rate inconsistency: the same creator paid materially different rates for equivalent work
• Competitor migration: a creator dropping you in one market while still active in another

That last one is invisible without unified monitoring and is often the earliest warning of a competitive problem. Sellers discover it months late, if at all.

Consolidated Scoring

Score each store separately, then identify whether low scores cluster in specific areas across all stores — which indicates a systemic problem — or appear in one store only, which indicates a local execution issue. Different diagnoses, different fixes.

Frequently Asked Questions

How long does a first audit take?

Three to five hours depending on roster size and data quality, with most of the time spent assembling data rather than analysing it. Subsequent audits take ninety minutes because you are refreshing an existing structure rather than building one. The time is almost always recovered in the first finding.

What is the most common finding?

Phantom roster — creators counted as active who have not posted in ninety days or never posted at all. Programs routinely overstate active rosters by thirty to fifty percent, which means every per-creator metric they calculate is wrong by the same proportion.

Should I audit monthly instead of quarterly?

Do a lightweight monthly version covering roster activity, expiries and attribution outliers — about twenty minutes. Keep the full eight-area audit quarterly, because several areas do not change fast enough to justify monthly attention and the full version is too expensive to run twelve times a year.

What if the audit says my program is beyond repair?

Rebuilding is sometimes correct — particularly if plan structure is inverted, rates are legacy across the board, and active roster is below forty percent. But rebuilding usually means fixing structure and cleaning the roster rather than starting from zero. Your creator relationships survive structural fixes; they do not survive being discarded.

Closing: Three Hours Recovers Real Money

The seller from the opening found two problems in forty minutes that together were costing her several thousand dollars monthly and had been for over a year. Neither was visible anywhere she was already looking.

That is typical. Audits find money not because programs are badly run, but because dashboards do not report absences, structure, or drift — which is where the money goes.

Block three hours this week. Export your roster, list your rates, check your expiries. You will likely find something worth fixing before you finish the first two areas.

DAMI’s plan tracking keeps roster activity, plan expiries and rate structures current continuously, which turns most of this audit from a quarterly project into a screen you check weekly.