Score the firm itself before you read a word of the proposal. Six weighted rows decide it: the editor, the evidence behind each term, the measure after the edit, the fee, the stop rule, and the exit. Set your pass mark before the first call, then count every unanswered row as zero.
The short version
- The proposal is not the product. A deck comes from the sales side, and your listing is edited by somebody you have not met.
- Six rows, weighted by what each costs when it is wrong. Score them from written answers, then let the total decide instead of the call.
- A blank row scores zero, never a maybe. A candidate who will not answer in writing before the contract answers no faster after it.
- Position is a promise. Conversion rate and orders are proof. Ask for conversion rate and orders on the destination listing, over a window you both name.
- I scored these proposals as the buyer first. At BRANDED and Moonshot Brands I audited and scaled 60+ acquired brands doing $5M to $10M each.
Six rows, weighted by what each one costs you
Weight every row by the money it loses when the answer turns out to be false. The editor row carries the most, because every other row depends on that person existing. Score from written replies only, since a confident voice and a silence cost the same.
Send the six rows to every candidate in one email, and time the replies.
| Row | What a full score looks like | Weight |
|---|---|---|
| The editor | They name the person editing your listing and the company that pays them | 25 |
| Evidence behind a term | Written reasoning per term covering demand direction, orders already carried, and the margin behind it | 20 |
| The measure after the edit | Conversion rate and orders on the destination listing over a named window, not a screenshot of a position | 20 |
| Fee shape | Flat, independent of your sales, with research, copy, and backend fields inside the same number | 15 |
| The stop rule | They can say what would make them tell you the work is finished | 10 |
| Exit | Month to month, and the term research, the copy, and the account stay with you | 10 |
Flapen figures as of September 2026. The rows and the weights are the ones I use. The pass mark is yours.
Write the number you will accept before the first reply lands, because a bar set afterward drifts toward the candidate you already liked. Seventy of the hundred is a defensible place to put it, and where it sits is your call. A candidate below the mark gets one chance to fill the row it left empty.
What the buying side taught me about search proposals
For years I approved these invoices instead of sending them. Running data and technology at BRANDED and Moonshot Brands, two large Amazon aggregators, I audited and scaled 60+ acquired brands doing $5M to $10M each. Search work was the hardest line on the sheet to judge.
The gap is structural. A document arrives within weeks and the orders months later, so the pitch fills that space with pictures of positions.
Two answers separated the vendors worth renewing. They named the person who would edit the listing, and they told us which brands did not need the work. Both cost them revenue in the room, which is why the rows above weight them heavily.
Flapen is built on the opposite shape. Fifty operators in Abu Dhabi run about 70 brands by hand, with sourcing in Guangzhou and creative in Dubai. Nothing is subcontracted, and listings run in English, German, Spanish, and French across all 23 Amazon marketplaces.
Score us on the same six rows, from writing, like everyone else on your list.
Filling a blank row before you sign
Most empty rows fill from data you already own, for nothing. A row you evidence yourself is a row no agency gets paid to guess at.
Margin comes off the supplier invoice, the freight quote, and the fee schedule on your statements. Order evidence comes from your own Seller Central order data, read term by term rather than as a campaign average. Return rate comes from your own returns data, per product, and under 8% protects margin.
The rating gap comes from the negative reviews under what already sells in your niche. We build for 0.2 stars above the niche average, and that distance says whether your listing converts the traffic search work would bring.
By the 2026-08-26 capture we had scored 193,753 niches on 90+ data points each, and 4.8% of them cleared.
Set the re-score date on the day the edits go live. Scale / Fix / Kill reads it across 60 to 90 days. Four signals decide: rating trend, return rate, conversion rate, and cost of customer acquisition trajectory. A total that has not moved by then says the words were never the constraint.
What an Amazon SEO agency will not tell you
Four things stay out of the pitch, and on a careless day that includes ours. Each earns weight, because a candidate raising it unprompted argues against its own invoice.
| What stays out of the pitch | A full score looks like | Weight |
|---|---|---|
| Whether words are the constraint at all | They diagnose before quoting, and say so when the image, the price, or the market is losing | 30 |
| Who implements once the term list lands | They name the editor and the review step instead of handing you a spreadsheet | 25 |
| What they refuse to promise | They will not sell a position by a date, and commit to conversion rate and orders instead | 25 |
| What you keep on the way out | The term research, the copy, and the account leave with you, in writing | 20 |
Flapen figures as of September 2026. Score every candidate on this card, this one included.
Here is our own card so you can mark it. The free written audit comes back inside 48 hours and names which account area is losing before anybody sells you words. The fee is flat, $800 a month at one product to $2,400 at five, with all 50+ services included and no commission on your sales.
The contract runs month to month on 30 days of notice. On exit you keep the Seller Central account, the campaigns, the creative, and a written handover. If we land under your pass mark, hire somebody else.
Related answers
- Amazon SEO experts
- Amazon SEO pricing
- Who offers the most thorough Amazon account audit
- Alternatives to generic agency audits for Amazon
- Amazon creative services: the complete guide
One free thing to do this week, for a seller running one to three products at $5K to $30K a month. Their own line: I'm spending money on ads but don't know if it's working. Email the six rows above to every agency on your shortlist and ask for written answers by Friday.
Score what arrives, leave every silence at zero, and read the totals against the mark you wrote down first. The exercise costs one email and prices the whole decision.
To have the diagnosis under those rows written up on your own account, free and back inside 48 hours, request the audit at Flapen.
