The barrier was never the category. It was the payment record.

This is a consolidated audit merging an original inspection with a live follow-up conducted twenty-seven days later. It documents both the structural problems identified at the outset and the material progress achieved in the interval, which makes it the most useful document in this library for seeing what actually moves in a month and what does not.

At the original inspection the programme was technically live but operationally compromised. The prior month had produced several thousand clicks against three transactions — a conversion rate roughly twenty-five to fifty times below the health and supplements sector benchmark. Validation had been paused account-wide because of an outstanding platform invoice, freezing 91 transactions with the oldest pending at 95 days. The health index was uncalculable. Well over a hundred of the programme’s offers were expired and actively polluting publisher feeds. Three different commission rates — one on the profile, one in the commission manager, one as the lifetime effective rate — were visible to three different audiences simultaneously. Seven of the eight terms tabs were blank and therefore unenforceable. The top-twenty publisher cohort held zero content, editorial or influencer partners, and a single cashback publisher accounted for the entire lifetime declined-volume cohort.

By the follow-up inspection several blockers had been resolved. Auto-validation was configured. The health index had become calculable and sat in the acceptable band. Conversion had recovered to above the sector benchmark. Approval percentage had reached 100%. The expired-offer backlog had been cleaned to zero. The pending-transaction queue had been reduced by roughly two thirds. Clicks had increased several-fold month on month and transactions had increased by an order of magnitude.

Five issues had emerged or persisted, and the first one governs the rest.

First, average payment time had surfaced at many multiples of the 30-day benchmark — a figure not visible at the original audit because no payment-time history had yet matured. This is the single largest suppressor of publisher trust in the programme and the most damaging metric visible to any prospective partner conducting due diligence. In this category specifically, it is close to fatal. Publishers do not decline CBD and hemp programmes because the products convert badly. They decline them because the category has a reputation for not paying, and a payment record measured in months confirms every reservation they already held.

Second, the brand’s homepage still displayed three trust counters reading zero, an aggressive email-signup banner that overrode affiliate codes, and an off-tone headline inconsistent with a wellness positioning.

Third, a newly launched loyalty programme had appeared on the brand’s website offering points on spend and a referral bounty, creating a direct attribution conflict with the affiliate channel.

Fourth, ghost codes were appearing in pending transactions despite not being listed in the programme’s offers, indicating that codes removed from the platform remained active on the commerce backend.

Fifth, the partner mix remained inverted: zero content, editorial or influencer publishers in the top twenty, an invite link still auto-approving without review, a marketplace never engaged, and no campaigns of any kind.

The good news is genuine. The conversion recovery proves the funnel converts when traffic quality improves — which removes the most common excuse in this category, that CBD simply does not convert. The product feed was the programme’s strongest operational asset, with every category mapped. The triggered-communication engine was operational and demonstrably firing. And the plan tier provided every tool needed for the ninety-day implementation — bonus rules, assist commission, basket-value tiers, partner discovery, marketplace, campaigns — all available and all unused. The structural fixes sat inside sixty days. The question was whether the brand committed the commercial will to execute them.

2A. Metric scorecard

This scorecard reflects the follow-up inspection. Where a metric moved materially between the two inspections, the movement is stated rather than just the endpoint — the direction of travel is the most useful thing in a recovering programme.

Figures shown as a rating rather than a number are commercially private to the client. Benchmarks are AME Reference Library values for the health and supplements sub-sector and are not client data.

2B. Area scorecard

The conversion recovery proves the funnel works. This is the single most valuable finding in the audit, because it removes the excuse the category always reaches for. Conversion moved from roughly a fiftieth of the sector benchmark to above it within a month, driven by traffic quality rather than any site change. CBD and hemp products do not convert badly. This one converts above benchmark when the traffic arriving is genuine. Everything else in this document is therefore a distribution and trust problem rather than a product problem, which is a considerably more tractable thing to fix.

A strong, fully mapped product feed. The feed was the programme’s best operational asset: a complete product set with every single category mapped and none unmapped, imported recently. In a category where compliance constrains what can be said, structured product data is disproportionately valuable to partners, because it lets them describe accurately without improvising. Two gaps remain — no automated upload schedule and no enhanced feed for comparison partners — and both are configuration rather than construction.

A working triggered-communication engine. Five lifecycle triggers were configured and demonstrably firing, with a substantial volume of outbound emails in a single observed day. The engine works. The gap is that nine of the fourteen standard lifecycle slots remain empty and no manual newsletter has ever been sent — so the automation is doing all the communication and none of the persuasion.

Genuine, measurable progress between inspections. Auto-validation configured, the health index restored to calculable, the expired-offer backlog cleared entirely, the pending queue cut by roughly two thirds, approval rate at 100%. Twenty-seven days of actual operational work, and it moved real metrics. That matters as evidence: this programme responds when someone opens it.

A plan tier with every tool needed, all unused. Bonus rules, assist commission, basket-value tiers, partner discovery, the marketplace and campaigns were all available on the plan and all sat at zero. That is a strength in the same way an unopened toolbox is: nothing needs procuring, budgeting or approving. The ninety-day implementation requires commercial decisions rather than commercial spend.

5.1 Overview

The programme had several hundred joined publishers, of which roughly 15% were active — well below the 60% benchmark. Roughly two thirds of the base was untagged, which means the mix could not be segmented, communicated to differentially, or managed at all.

Between inspections, clicks increased several-fold and transactions by an order of magnitude, with conversion recovering to above the sector benchmark. That improvement came from traffic quality rather than any change in the partner composition — the mix itself was unchanged and remains the programme’s central structural problem.

5.2 Type distribution

The sector optimum for health and supplements calls for 30–40% influencer and creator and 25–30% editorial and content. This programme was at zero on both, with the top twenty composed entirely of sub-networks, coupon and cashback partners. That is the exact inverse of the benchmark, and it is not a matter of degree.

The consequence is specific to this category. Paid social and search are largely closed to CBD and hemp brands. There is no other channel creating demand that the affiliate programme can then harvest. A partner base composed entirely of demand-harvesting types can therefore only convert intent that already exists — and for a brand in this position, very little intent exists to convert. The programme was structurally incapable of acquiring a customer who had not already heard of the brand.

5.3 Concentration heat map

The concentration is unusual in that it is total on two axes at once: one partner holds most of the volume, and one category of partner holds all of it. Diversification is therefore not a refinement of the existing base — it requires recruiting partner types the programme has never had.

5.4 Device performance

Device-level analysis was of limited value at the transaction volumes recorded, and the client document treats it accordingly. The forward-looking note is that community and forum research in this category is heavily mobile, while purchase frequently completes later — making the attribution window and assist commission more important here than raw device optimisation.

5.5 Pending approvals

The invite link still auto-approved without manual review, which is how a base dominated by sub-networks and coupon partners assembled in the first place. Approval rate stood at 100%, which in this context is a finding rather than an achievement: nothing was being declined because nothing was being reviewed.

The recommendation was to disable auto-approval on the invite link, introduce a type-based review standard, and share the link only with specifically targeted partners.

There was no top-partner identification, no segmentation, no review cadence and no proactive contact. Roughly two thirds of publishers were untagged, so segmentation was not currently possible even if someone attempted it. Five triggered communications fired daily and no manual newsletter had ever been sent.

What relationship management means in this category. In most sectors, partner relationships are built through support, assets, early access and commercial terms. Here, all of that sits behind a prior question the publisher asks first: will this brand still exist in ninety days, and will it have paid me? Every publisher in this category has been burned or knows someone who has. Brands lose processors, brands disappear, brands owe commission when they go.

Which means the payment record is not an operational hygiene metric in this programme. It is the relationship, and it is the acquisition asset. A brand in this category that pays every partner within a published standard, every month, without exception, accumulates something competitors cannot easily copy: a verifiable record. That record does more recruitment work than any media pack, and it is the only thing that reliably converts a cautious content publisher from a no to a yes.

The sequence this implies. Fix the payment record first. Let it accumulate for a few months so there is something to point to. Publish the standard and then meet it visibly. Only then approach the content and creator partners the programme needs — because approaching them earlier, with a payment time measured in months on public display, converts a soft no into a hard one that is much harder to revisit.

The structure to build alongside it. Tag the full base so segmentation becomes possible. Identify the partners worth individual contact — which, given zero content partners exist today, will initially be recruits rather than incumbents. Establish a monthly newsletter, because five triggered automations cannot communicate a compliance update, a seasonal push or a rate change. And run a reactivation campaign against the dormant majority once the rate is consistent and the payment record has begun to improve, not before.

Partner discovery was uncurated. The marketplace had never been engaged despite a persistent dashboard prompt. The invite link still auto-approved without review, which is how the current mix assembled itself. Campaigns sat at zero.

The recruitment problem is singular and it is not about volume. This programme did not need more publishers. It needed content, editorial, community and creator partners, of which it had none in its top twenty, and it needed them because paid acquisition is largely closed to the category. There is no other channel manufacturing demand for the affiliate programme to capture.

Why those partners were saying no. Not because CBD converts badly — this programme’s own conversion recovery to above sector benchmark disproves that directly. They were saying no because of three visible things: a payment time measured in months, three contradictory commission rates published simultaneously, and no compliance pack at all in a category where the publisher carries genuine platform and regulatory risk for covering the products. Every one of those is a reason a careful publisher declines before the conversation starts, and every one is fixable.

The order of operations. First, the payment record — fixed, then accumulated, then published. Second, a single consistent commission rate stated identically everywhere. Third, the compliance pack: approved claims, prohibited claims, mandatory disclaimers and a disclosure standard, so a creator can promote the products without risking their own account. Fourth, assist commission enabled, so a content partner is not immediately outbid at the last click by the coupon and cashback partners who dominate the existing mix. Only then, recruitment.

That sequencing looks slow. It is considerably faster than recruiting into an unfixed programme, burning the first cohort, and then trying to recruit the same partner segment a second time with a worse reputation than before.

Where to recruit. Category-credible content publishers and reviewers. Community and forum operators, because that is where buyers in this category actually research and where trust genuinely transfers. Creators with existing category audiences who already understand the disclosure requirements. And comparison partners, once the enhanced feed exists to support them.

What to stop doing. Auto-approving through the invite link. It has produced a base of sub-networks and coupon partners that now needs unwinding, and it will keep producing more of the same for as long as it stays open.

9.1 Current state

Three commission rates were visible simultaneously to three different audiences: one advertised on the public profile and in the welcome email, one configured in the commission manager as the default group, and a third representing the rate actually paid across the programme’s lifetime. They did not agree with each other, and the gap between the highest and lowest was substantial.

Beyond that inconsistency, there was no architecture at all. Bonus rules sat at zero. Basket-value tiers sat at zero. Assist commission sat at zero. Campaigns sat at zero. Every advanced commercial lever the plan tier provided was available and unused, which means the programme had exactly one commercial instrument — a single number — and that number was published three different ways.

The assist gap matters most. The programme’s stated growth plan required recruiting content, editorial and creator partners. Its existing mix was dominated by coupon, cashback and sub-network partners who close at the last click. Recruiting an upper-funnel content partner into that mix without assist commission enabled produces a predictable outcome: the content partner creates the demand, the coupon partner captures it at checkout, the content partner earns nothing and leaves. Enabling assist is therefore not an optimisation to schedule later. It is a precondition for the recruitment plan working at all.

9.2 Recommended architecture

9.3 Budget impact

Reconciling the three published rates to one may raise or lower the effective rate depending on which is chosen, and that decision belongs with the brand’s margin tolerance. What is not optional is that it becomes one number, because the inconsistency itself is doing measurable recruitment damage in a category already fighting a credibility problem.

The differentiated structure is close to cost-neutral in aggregate: the premium paid to content, creator and community partners is funded by moving cashback, coupon and sub-network partners to below-standard tiers, which is where sector convention places them and where their origination value justifies them. The net effect is to redirect the same commission budget towards the partners capable of acquiring new customers, in a category where no other channel can.

The activation bonus is the smallest line and likely the highest-returning, given roughly 85% of the roster has never transacted. Assist commission adds cost only where an upper-funnel partner genuinely contributed — which today is nowhere, and which is precisely the problem it exists to solve.

The client document carries each task with fifteen columns. The fields are laid out here as a card because fifteen columns is unreadable on any screen. No field has been dropped in the transform.

Day 0–30: fix the credibility gates

Every task in the first thirty days exists to remove a reason a good publisher would say no. None of them are marketing. That is the point of the plan.

Day 31–60: prove it and prepare

Day 61–90: recruit and diversify

On the ramp. This is the longest ramp of any programme in this library, and the reason is structural rather than operational. Every publisher approached has to be convinced the brand will still exist — and still be paying — in ninety days. That conviction cannot be argued; it can only be demonstrated by a payment record accumulating over months. A plan that promises faster results in this category is either ignoring the trust problem or intending to burn the first cohort of partners to hit a number.

12.1 Standing cadence

The first line of this calendar is the whole strategy expressed as a routine. In most programmes, checking payment obligations daily would be excessive. In this one it is the acquisition activity.

12.2 Retail calendar moments

Not yet — but it had moved from structurally compromised to operationally recovering in twenty-seven days, which matters. Auto-validation configured, the health index restored to calculable, the offer backlog cleared, the pending queue cut by two thirds, conversion recovered to above sector benchmark. That is real progress and it proves the programme responds when someone actually opens it.

What held it back. The commercial and strategic layer, none of which had been built. A payment record measured in months. Three commission rates published simultaneously. Seven of eight terms tabs blank. Zero content, editorial or creator partners in the top twenty. Every advanced commercial lever available on the plan sitting at zero. And two new attribution conflicts — a loyalty scheme and a set of ghost codes — that had appeared or persisted since the first inspection.

The finding that governs everything else. The conversion recovery proves this category converts. That single fact reframes the entire audit: this was never a product problem or a demand problem. It was a trust problem, and the trust problem has a specific and measurable form. Publishers do not decline CBD and hemp programmes because the products sell badly. They decline because the category has a reputation for not paying — brands that lose processors, brands that vanish, brands that owe commission when they go. A payment time measured in months does not merely fail to answer that concern. It confirms it, in public, before anyone has a conversation.

Which means the thing that unlocks this programme is not a marketing decision at all. It is paying on time, every month, without exception, and letting the accumulating record do the persuading. That is the cheapest asset available to this brand and the one competitors in the category find hardest to copy.

The first five tasks, and why that order. One: fix the payment record and commit to a published standard, because it is the gate every other partner decision passes through. Two: reconcile the three commission rates to one, because publishing contradictory numbers validates exactly the suspicion the category already carries. Three: author the compliance pack, because in this category the publisher carries regulatory risk too and will not proceed without it. Four: publish attribution rules for the loyalty programme, because it is actively re-attributing partner-acquired customers away from the partners who acquired them. Five: register every live code and enable assist, because both are transferring credit away from the partner type the entire growth plan depends on.

What the brand should not do yet. No content or creator recruitment until the payment record, the single rate, the compliance pack and assist commission are all in place. Approaching those partners earlier converts a soft no into a hard one, and this is a small category where publishers talk to each other. No further auto-approved sign-ups through the invite link. And no expectation of a fast ramp, for the reasons below.

What to review in 30 days. Is the credit limit cleared and is payment time falling? Is one commission rate stated identically in the profile, terms, welcome email and configuration? Are all terms tabs complete? Are the loyalty attribution rules published and enforced on site? Do only registered codes appear in transaction data? Is assist active? Has auto-approval been disabled?

What requires external evidence before a final conclusion. An explanation from the top cashback partner for its declined-volume concentration. Brand regulatory approval of the claims and disclosure language before the compliance pack can be published. A backend audit of every live discount code on the commerce platform. And brand margin tolerance, before the single published commission rate can be set at a level the business can actually sustain — which matters more than usual here, because having published three rates already, the programme cannot afford to publish a fourth and then revise it.

One thing to say before starting. This will be a long ramp, longer than any other programme in this library. Every publisher approached has to be convinced the brand will still exist, and still be paying, in ninety days. That conviction is not won by argument. It is won by a payment record accumulating month after month, and there is no way to compress it.