Mansoor · spécialiste indépendant de la pub Amazon
Votre ACOS n’est pas un mystère — c’est un problème de gestion. Je suis Mansoor. PPC, fiches et défense de marque sur comptes live : couper le gaspillage, récolter les requêtes qui convertissent, pousser les ASIN en page 1.
Pas de contrat. Pas de ferme de juniors. Un opérateur — moi. La revue gratuite est une lecture écrite de la fiche publique : fuites, fit oui/non, semaine 1.
Je gère des comptes amazon.fr (FR, BE, CA). Réponse et revue en Français. Frais, pubs et politiques se confirment dans Seller Central.
Lien boutique public seulement. Pas de Seller Central avant accord. Réponse sous 24 h ouvrées.
Fourchettes de motifs sur catalogues live, pas des totaux de clients nommés. Je les remplace dès que ces marques disent oui.
La plupart des comptes « optimisés » fuient aux mêmes quatre endroits. Si l’un colle, le compte en est un.
Les campagnes broad et auto non gérées jettent du budget sur des termes hors sujet — personne n’a négativé depuis des mois. Vous payez Amazon pour n’apprendre rien.
Les rivaux ciblent vos fiches et siphonnent l’acheteur au moment d’acheter. L’espace de marque n’est pas défendu.
L’ACOS grimpe d’un ou deux points par mois et la dernière agence a dit « le marché est compétitif ». Les CPC montent vraiment. Le problème, ce sont les enchères non gérées.
Vous dépensez fort en pubs mais l’organique ne bouge pas — les campagnes n’ont jamais été structurées pour nourrir le rang sur vos money keywords.
Enchères, fiche, défense, stock, deals et DSP. La plupart des marques au-delà de 1 M$ ont besoin de plus qu’un login PPC. Un job d’abord. Le reste quand les pièces se battent.
Contrôle quotidien des enchères, budgets et search terms sur Sponsored Products, Brands et Display.
Voir le service →Trouver les requêtes qui convertissent dans auto et broad — puis promouvoir ou tuer selon un calendrier.
Voir le service →Défendre les termes de marque et les fiches contre conquesting et hijacking.
Voir le service →Lancements structurés avec sortie — pas « agressif » pour toujours.
Voir le service →Titre, puces et backend qui portent les requêtes payantes — pas seulement le volume.
Voir le service →Programmatic display, video, audio, and Streaming TV on Amazon audiences — self-serve CPM, gated until Sponsored Products actually converts.
Voir le service →Pas d’onboarding mystère. Quatre semaines vers un rythme de desk visible.
Envoyez le lien. Je lis la fiche publique et je dis par où je commencerais. Si je ne peux pas montrer les fuites, je refuse.
Exact pour les converters, discovery pour le harvest, défense de marque. Passe fiche si la page ne convertit pas le trafic que nous allons acheter.
Négatifs, enchères et un change log écrit chaque semaine. Pas de « le marché est compétitif » comme explication.
Mois par mois. On scale ce qui contribue. On coupe le reste. Jamais coincé dans un PDF 12 mois.
Most brands read every page here before they write. They are hiring in private: who owns the account, what week one looks like, and whether the first step costs them Seller Central. Here is the answer.
Title, bullets, A+, reviews, brand defense, and obvious ad-real-estate problems. Storefront URL only — no Seller Central login until we agree to work.
If I cannot show the leaks, I decline. You are not added to a roster. A “no” is a complete answer, not a follow-up sequence.
Harvest, defense, listing pass, or “not yet.” Named in writing — not a 40-slide deck and a 12-month PDF.
Les fondateurs shoppent ce site comme une agence : qui possède le compte, à quoi ressemblent les 30 premiers jours, si le fee est un pourcentage du spend. Utilisez cette grille, puis envoyez le lien si les réponses tiennent.
| Ce que les fondateurs vérifient | Agence typique | Ce desk |
|---|---|---|
| Who works the account? | A pod. You meet a closer. A junior runs the console. | Mansoor. One operator. Same person who replies and changes bids. |
| What is the first step? | A sales call, then Seller Central access before you have a written plan. | A free review of the public listing. Login only if we agree to work. |
| What do you get back? | A pitch deck. Sometimes a 90-day PDF nobody can audit. | Leaks I can already see, fit yes/no, and what week one would be. |
| Contract | Often 6–12 months so the agency can recoup onboarding. | Month to month. If the work is not moving the number, you leave. |
| How you pay | Often 10–20% of spend — the fee rises when you spend more, including on waste. | Flat fee after the review. Compare it to 30 days of zero-order search-term spend. |
| How you know work happened | A dashboard and “the market is competitive.” | A four-minute written change log every week. |
| ACOS guarantee | Sometimes promised. It is a red flag — Amazon sets the auction. | None. I show the leaks. I do not sell a number I cannot control. |
A sample of account patterns I keep seeing — and fixing. I will swap these for named client stories as soon as those brands say yes.
Same catalog. Same budget. The waste was unmanaged broad match and a campaign structure nobody had touched in months.
Read the work → Keyword harvest $11k / mo Waste removedThree prior agencies called the category expensive. The real problem was one-campaign-does-everything structure.
Read the work → Scale without torching ROAS $80k → $210k Ad revenue / moScaling usually means buying worse traffic. Harvesting the long tail is how both volume and ROAS can move together.
Read the work →Policy changes, fee updates, and ad-platform shifts — tracked so nothing blindsides the account. Each brief says who it hits and what to do this week.
Amazon’s Holiday 2026 post: fulfillment surcharge 15 Oct 2026–14 Jan 2027. Storage is a separate Oct–Dec window. Recalculate break-even ACOS before you raise Q4 bids.
Read the brief →Two fee clocks, inbound cutoffs, event ads, then harvest. Print this before you raise a single Q4 bid.
Read the brief →About Amazon: Prime Big Deal Days 6–7 Oct 2026. Holiday inbound for that badge was posted as 2 / 9 / 16 September (AWD / minimal / optimized). Ads follow inventory, not the press page.
Read the brief →ACOS, TACOS, IPI, Account Health, fees. Answer-first, then the weekly move. Calculators stay in your browser.
Models draft harvest tables and the weekly letter from exports. They do not log into Seller Central or set bids.
Read the playbook →Xnurta, Flywheel, and Cobalt after structure and a contribution floor — never unsupervised on a blended campaign.
Read the playbook →Cerebro and Keyword Scout estimate demand. The search-term report is the only list you already paid to test.
Read the playbook →Helium 10 and Jungle Scout suggest. Cobalt, Xnurta, and Teikametrics execute inside rails. Sellerboard sets the floor. Claude, Codex, Kimi, and Cursor draft the letter. Each product is its own crawlable page.
Motifs d’exemple — pas de citations nominatives. Les vrais noms dès que ces marques disent oui.
Sample pattern
“Our ACOS went from 42% to 24% in ninety days. Same catalog, same budget. The difference was someone actually managing the account every single day.”
Marcus T.Founder, kitchen goods — $4.2M/yrSample pattern
“Three agencies told us the category was just expensive. The account was restructured into single-keyword campaigns and wasted spend dropped about $11k a month.”
Priya S.CEO, supplements — $7.8M/yrSample pattern
“The weekly reporting is the part I did not know I needed. I can see where every dollar went, and the hero ASIN has held #2 organic for four months.”
Daniel R.Co-founder, pet supplies — $2.9M/yrMansoor. ASIN Experts is a one-operator desk — not a junior farm and not a 40-person pod. You talk to the person who changes the bids.
No. Month to month. If the work is not moving ACOS, rank, or wasted spend, you should not be stuck in a 12-month PDF.
Best fit is an FBA brand already spending on ads and feeling the leak — typically past about $1M in trailing revenue, or a serious launch with budget. If I cannot show the leaks on a review, I will say so.
Yes. Send the store link and a spend range. I read the public listing and tell you what I would fix first. No deck, no bait-and-switch webinar.
No. The first pass is the public listing — storefront or ASIN URL. I can already see title, bullets, A+, reviews, and obvious brand-defense gaps. Seller Central access comes only if we both agree to work.
Because I cannot tell you if I am a fit without seeing the catalog. A call without the listing is a sales conversation. A review with the listing is a diagnosis. You get the diagnosis first — leaks I can see, fit yes or no, and what week one would be.
No. Anyone who does is selling a number Amazon still sets in the auction. I show wasted spend, structure, and a week-one plan. That is the honest version of “what you get.”
Within 24 hours on business days. I am usually in the accounts Monday–Friday. Weekend notes wait until Monday unless something is on fire.
Either. Most brands start with PPC + a listing pass. Full desk is for when ads, listing, and defense are already fighting each other.
Yes — those seats are named on the operator stack. Helium 10 and Jungle Scout (JS) suggest keywords; the search-term report proves them. Cobalt, Xnurta, and Teikametrics execute only inside contribution rails. Sellerboard sets the P&L floor. Claude, Codex, Kimi, and Cursor draft analysis — they do not bid. Seller Central and the Ads Console stay source of truth.
Most 2026 retainers land around $1,500–$5,000 a month or 10–20% of ad spend. This desk is a flat month-to-month fee after a free review — not a percent that rises when you spend more. Compare the fee to 30 days of zero-order search-term spend first.
Two clocks: peak FBA storage from 1 October through December, and peak fulfillment from 15 October 2026 through 14 January 2027. Recalculate break-even ACOS before you raise bids. Dates are on the operator calendar — confirm live rates in Seller Central.
Not ready to send a listing?
Calculators, case studies, and Amazon Updates. Work email and a consent tick. No webinar.
Two doors, same desk
I read the public listing by hand. You get three things back: the leaks I can see without logging in, whether I am a fit, and what week one would be. If I cannot help, I say so.