Skip to content

Metricbrief · How it works

The details behind your weekly review.

How we check the numbers, how your data is handled, what each stage involves, and what is included or optional. The examples use fictional data. They show how a review can look, not features promised for every scope.

Full example

One week, from the review to the follow-ups.

The review shows what changed and which products contributed. Follow-up questions run in Claude or ChatGPT, on the same data and definitions. The standard scope configures one of the two.

  1. The review arrivesOn your schedule, in Slack or email.
  2. See what changedThe totals, and the products behind the change.
  3. Ask a follow-upIn Claude or ChatGPT, on the same data.
  4. Get tables, charts, summariesSame metrics, same definitions.

Weekly Amazon Business Review

Amazon US · Oct 12 to 18 vs Oct 5 to 11

Tue, Oct 20, 7:00

Sales fell 7.1% ($13,120). Two products account for $12,540 of the fall (96%).

Sales
$171,200
−7.1% was $184,320
Units
2,837
−4.4% was 2,968
Sessions
33,280
−2.8% was 34,230
Conversion
8.52%
−0.15 pts was 8.67%
Ad spend
$25,310
−4.9% was $26,620

Sales by week

What changed, by product

Change in sales by product, Oct 12 to 18 vs Oct 5 to 11
Walnut Monitor Stand−$8,370
Leather Desk Mat−$4,170
Brass Desk Organizer+$460
Other 21 products−$1,040
Total−$13,120

This shows where the change came from, not why it happened.

Source: Sales and Traffic by child ASIN; Sponsored Products advertised product. Oct 12 to 18 vs Oct 5 to 11, Pacific Time.

# amazon-reporting Email to 3 people

Weekly Amazon review

Show traffic and conversion for these two products.

Walnut Monitor Stand kept almost the same traffic (−1.9% sessions) and converted less: 7.57% to 6.34%. Leather Desk Mat converted at about the same rate but had 11.4% fewer sessions.

Sessions and conversion, Oct 12 to 18 vs Oct 5 to 11
ProductSessionsConversion
Walnut Monitor Stand4,040−1.9%, was 4,1206.34%−1.23 pts, was 7.57%
Leather Desk Mat5,190−11.4%, was 5,8608.90%−0.01 pts, was 8.91%
Observed
Walnut Monitor Stand: conversion fell, traffic held. Leather Desk Mat: traffic fell, conversion held.
Possible explanations, not confirmed
For Walnut Monitor Stand, a change in price, stock, reviews or competing offers could lower conversion. This data doesn't include those. For Leather Desk Mat, fewer sessions can come from less ad visibility or a lower search rank.

Source: Sales and Traffic by child ASIN, Oct 12 to 18 vs Oct 5 to 11. Conversion = units ordered / sessions.

Query details
SELECT product, week_start, sessions, units_ordered,
       SAFE_DIVIDE(units_ordered, sessions) AS conversion
FROM reporting.weekly_product_performance
WHERE product IN (@top_two) AND week_start IN (@last_two_weeks)

Compare their ad spend over the last four weeks.

Walnut Monitor Stand was steady, between $9,640 and $9,910 a week. Leather Desk Mat spent $6,180 to $6,240 for three weeks, then $4,870 last week (−21.6%).

Ad spend by week, Sep 21 to Oct 18
4k6k8k10k
  • Walnut Monitor Stand
  • Leather Desk Mat
Ad spend by week, Sep 21 to Oct 18
Week ofWalnut Monitor StandLeather Desk Mat
Sep 21$9,640$6,180
Sep 28$9,720$6,240
Oct 5$9,880$6,210
Oct 12$9,910$4,870
Observed
Leather Desk Mat: ad spend and sessions both fell in the same week.
Possible explanations, not confirmed
The timing matches, but this doesn't show that lower spend caused the lower traffic. Worth confirming whether the budget change was planned.

Source: Sponsored Products advertised product, Sep 21 to Oct 18.

Query details
SELECT advertised_product, week_start, SUM(spend) AS ad_spend
FROM reporting.weekly_ad_spend
WHERE advertised_product IN (@top_two) AND week_start >= @four_weeks_ago
GROUP BY 1, 2 ORDER BY 2

Summarise this for our weekly meeting.

Weekly Amazon review, Oct 12 to 18

What changed

  • Sales $171,200, −7.1% (−$13,120) on the week before.
  • Walnut Monitor Stand (−$8,370) and Leather Desk Mat (−$4,170) account for 96% of the fall.

What the data shows

  • Walnut Monitor Stand: sessions −1.9%, conversion 7.57% to 6.34%.
  • Leather Desk Mat: sessions −11.4%, conversion 8.91% to 8.90%; ad spend −21.6% the same week.

Open questions for the team

  • Did anything change on the Walnut Monitor Stand listing, price or stock?
  • Was the Leather Desk Mat budget change planned?

Source: the review and the two answers above, same definitions.

Weekly Amazon review

Show traffic and conversion for these two products.

Walnut Monitor Stand kept almost the same traffic (−1.9% sessions) and converted less: 7.57% to 6.34%. Leather Desk Mat converted at about the same rate but had 11.4% fewer sessions.

Sessions and conversion, Oct 12 to 18 vs Oct 5 to 11
ProductSessionsConversion
Walnut Monitor Stand4,040−1.9%, was 4,1206.34%−1.23 pts, was 7.57%
Leather Desk Mat5,190−11.4%, was 5,8608.90%−0.01 pts, was 8.91%
Observed
Walnut Monitor Stand: conversion fell, traffic held. Leather Desk Mat: traffic fell, conversion held.
Possible explanations, not confirmed
For Walnut Monitor Stand, a change in price, stock, reviews or competing offers could lower conversion. This data doesn't include those. For Leather Desk Mat, fewer sessions can come from less ad visibility or a lower search rank.

Source: Sales and Traffic by child ASIN, Oct 12 to 18 vs Oct 5 to 11. Conversion = units ordered / sessions.

Query details
SELECT product, week_start, sessions, units_ordered,
       SAFE_DIVIDE(units_ordered, sessions) AS conversion
FROM reporting.weekly_product_performance
WHERE product IN (@top_two) AND week_start IN (@last_two_weeks)

Compare their ad spend over the last four weeks.

Walnut Monitor Stand was steady, between $9,640 and $9,910 a week. Leather Desk Mat spent $6,180 to $6,240 for three weeks, then $4,870 last week (−21.6%).

Ad spend by week, Sep 21 to Oct 18
4k6k8k10k
  • Walnut Monitor Stand
  • Leather Desk Mat
Ad spend by week, Sep 21 to Oct 18
Week ofWalnut Monitor StandLeather Desk Mat
Sep 21$9,640$6,180
Sep 28$9,720$6,240
Oct 5$9,880$6,210
Oct 12$9,910$4,870
Observed
Leather Desk Mat: ad spend and sessions both fell in the same week.
Possible explanations, not confirmed
The timing matches, but this doesn't show that lower spend caused the lower traffic. Worth confirming whether the budget change was planned.

Source: Sponsored Products advertised product, Sep 21 to Oct 18.

Query details
SELECT advertised_product, week_start, SUM(spend) AS ad_spend
FROM reporting.weekly_ad_spend
WHERE advertised_product IN (@top_two) AND week_start >= @four_weeks_ago
GROUP BY 1, 2 ORDER BY 2

Summarise this for our weekly meeting.

Weekly Amazon review, Oct 12 to 18

What changed

  • Sales $171,200, −7.1% (−$13,120) on the week before.
  • Walnut Monitor Stand (−$8,370) and Leather Desk Mat (−$4,170) account for 96% of the fall.

What the data shows

  • Walnut Monitor Stand: sessions −1.9%, conversion 7.57% to 6.34%.
  • Leather Desk Mat: sessions −11.4%, conversion 8.91% to 8.90%; ad spend −21.6% the same week.

Open questions for the team

  • Did anything change on the Walnut Monitor Stand listing, price or stock?
  • Was the Leather Desk Mat budget change planned?

Source: the review and the two answers above, same definitions.

Illustrative example using fictional data. Answers show what the data shows. They don't prove why something changed.

Checking the numbers

Checked against the reports you use today.

During the trial we compare each review with your current report, or with Seller Central if you don't have one, week by week. Where a figure differs, we find out why and write it down.

Each metric has a written definition, a period and a time zone, agreed with your team.

The figures are calculated from your Amazon reports by fixed rules, not by AI. An AI model writes the text around them, from templates we agree with you.

Not every figure has to match to the dollar. Ad spend, for example, keeps updating for a few days as late clicks are counted. A difference is fine once it is explained, documented and accepted by your team.

Reconciliation
Week of Oct 5 to 11, Amazon US, Pacific Time
MetricYour reportOur extractionStatus
Sales$184,320$184,320Matches
Units2,9682,968Matches
Ad spend$26,575$26,620Explained

Ad spend: why the $45 difference

Your report was exported a day before our extraction. Amazon Ads keeps updating spend for a few days as late clicks are counted, so the later figure is $45 higher. Documented, and accepted by your team.

Definitions agreed with your team

Sales
Ordered product sales, before refunds, excluding tax. Same figure as your report.
Units
Units ordered, all products and fulfilment channels.
Week
Monday 00:00 to Sunday 23:59, Pacific Time, the Amazon US marketplace's time zone.
Ad spend
Sponsored Products spend, as Amazon Ads reports it on the day of extraction.

Illustrative example using fictional data.

More definitions

Sessions
Visits to the product's detail page, as Seller Central reports them.
Conversion
Units ordered divided by sessions, for the same product and week.
Products contributing
Each product's share of the change in sales, by arithmetic. Not a cause.
Product groups
Your own groupings of products, agreed with your team and kept in writing.

Delivery

On the day and time your team agrees.

Weekly by default, or on the schedule your team uses, in Slack or email. Each review states the period it covers.

The example arrives on Tuesday at 7:00. Amazon's sales and traffic figures for a day arrive about a day later, so on Monday morning Sunday is often still missing.

Weekly Amazon Business Review, delivery

Schedule

Every Tuesday, 7:00

Slack

# amazon-reporting

Email

3 people

Next review: Tue, Oct 27, 7:00, covering Oct 19 to 25.

When it arrivesEach covers the week before, Monday to Sunday
SentCoversStatus
Tue, Oct 20, 7:00Oct 12 to 18Sent
Tue, Oct 27, 7:00Oct 19 to 25Scheduled
Tue, Nov 3, 7:00Oct 26 to Nov 1Scheduled
Tue, Nov 10, 7:00Nov 2 to 8Scheduled

Illustrative example using fictional data. Change the schedule to see when each review would arrive.

Stages

What happens at each stage.

  1. Agree the review

    • The questions the review answers, the metrics and their definitions, your product groups.
    • Who receives it, where (Slack or email) and when.
    • The scope in writing: marketplace, sources, the assistant for follow-ups, and the price after the trial.

  2. Free 30-day trial

    • Your team grants read-only access to your Seller Central and Amazon Ads reports, and shares your current report if you have one.
    • We prepare the weekly review from those reports and check it against your numbers until they match, or any differences are explained and accepted by your team.
    • The 30 days start when your first checked review is delivered. Your team gives feedback on each review.
    • During the trial the reports are prepared from the access you grant. The automated connection comes after, if you continue.

  3. Continue if it's useful

    • We set up the automated read-only connection through your Amazon account. Someone on your team with admin rights approves it in Seller Central and in Amazon Ads.
    • The review keeps running every week. We host, deliver, check and maintain it.
    • We configure follow-up access in one assistant, Claude or ChatGPT, on your team's own subscription.
    • Month to month. No minimum term, no cancellation fee.

  4. If you stop

    • You get an export of your data and reports.
    • We then delete them from our systems.
    • You can remove our access in Seller Central and Amazon Ads at any time.

Access and data

Read-only access. Hosted by us, separately for each client.

Your Amazon reports

  • Seller Central sales and traffic
  • Amazon Ads

Read-only. Through Amazon's own permissions.

Metricbrief, hosted for you

  • Your definitions, products and groups
  • Sales and ads combined by product
  • Checks against your numbers
  • Your data kept separate from other clients'

Your team

  • The review in Slack or email, on schedule
  • Follow-ups in Claude or ChatGPT

Your team keeps the decisions.

  • Read-only, through Amazon's own permissions

    You grant access in Seller Central and the Amazon Ads console, and you can remove it there at any time.

  • We read, we never change

    We don't change your listings, prices, bids or ads.

  • Hosted by Metricbrief, separately for each client

    We store and process your data for your review only, kept apart from every other client's.

  • What the review uses

    Sales and traffic, and advertising. Adding inventory, search terms or other reports to the review is a separately agreed extension.

  • Follow-ups on the same data

    The figures in your review are calculated from your Amazon reports without AI. An AI model, used through Metricbrief's account, writes the review's text from those figures. Follow-up questions run in your team's own Claude or ChatGPT account, which we connect only to the data included in your review. Your company's plan with Anthropic or OpenAI sets how they handle what the assistant reads.

  • If you stop

    You get an export of your data and reports, and we delete them from our systems. We sign your NDA.

Included and optional

$1,000/month for the standard scope. Anything else, only with your approval.

Your scope is agreed in writing before the trial starts. Additional sources, marketplaces or custom work are quoted and approved by you first. Your team's own Claude or ChatGPT subscription is required for follow-up questions.

Included

The standard scope

  • One Amazon marketplace.
  • Seller Central sales and traffic, and Amazon Ads.
  • One recurring review, delivered in Slack or email.
  • Setup, hosting, scheduled delivery, checks, maintenance and support.
  • Follow-up access configured in one assistant, Claude or ChatGPT.
  • An export of your data and reports if you stop.
Optional, quoted separately

Extensions

  • Search Query Performance: search terms, impressions, clicks and purchases for your brand.
  • Inventory figures next to sales.
  • More Amazon marketplaces in the same review.
  • More reports we build for you, such as a monthly pack or product deep dives.
  • Competitor tracking.
  • A second assistant for follow-ups.
  • Installing an independent system in your company's own infrastructure, as a separately quoted project.

Dashboards and conversational analytics in Data Studio

Once your data is connected, your team can build its own dashboards and ask questions in Data Studio, on the same checked numbers. We show you how to set them up. A dashboard we build for you is quoted separately.

Weekly sales

Oct 12, 2026 to Oct 18, 2026

Sales

$171,200

−7.1% on the week before
160k170k180k190k
Sales by week
Week ofSales
Aug 24$166,700
Aug 31$170,300
Sep 7$171,800
Sep 14$176,450
Sep 21$179,980
Sep 28$181,060
Oct 5$184,320
Oct 12$171,200
Units
2,837
−4.4%
Sessions
33,280
−2.8%
Conversion
8.52%
−0.15 pts
Ad spend
$25,310
−4.9%

Sales by product, Oct 12 to 18

ProductSalesShareChange
Walnut Monitor Stand$38,12022%−18.0%
Leather Desk Mat$31,87019%−11.6%
Brass Desk Organizer$24,61014%+1.9%
Other 21 products$76,60045%−1.3%

The same agreed numbers in a Data Studio dashboard your team can filter as it needs.

Compare last week with the week before

Last week is Oct 12 to 18, 2026 (ISO week W42) in the available data.

In W42, sales were $171,200, −7.1% on $184,320 in W41. Sessions went from 34,230 to 33,280, and the conversion rate from 8.67% to 8.52%.

Here's the result for Weekly comparison.

RowPeriod startPeriod endWeekTotal sales (USD)
12026-10-122026-10-182026-W42171,200
22026-10-052026-10-112026-W41184,320
32026-09-282026-10-042026-W40181,060
42026-09-212026-09-272026-W39179,980

A Data Studio agent answers questions from the same data, with the result under the answer.

Illustrative example using fictional data.

FAQ

Detailed answers.

Anything else: email hello@metricbrief.net.

Apply for a free 30-day trial
What's in the review?
The period against the one before, on the metrics we agree with your team (for example sales, units, sessions, conversion and ad spend), the products that contribute most to each change, and the source and period of every figure. Observations are kept apart from suggested checks, and suggested checks are not confirmed causes.
What do follow-up questions cover?
Questions about the data included in your review, in your team's own Claude or ChatGPT: comparisons between products and periods, tables and summaries, on the same metrics and definitions. The standard scope configures one of the two. New sources or metrics that need building are quoted separately.
What do we need to provide during the trial?
Read-only access to your Seller Central and Amazon Ads reports, your current report if you have one, and feedback on each weekly review.
When does the 30-day trial start?
When your first checked review is delivered, not when access is granted.
What does it cost after the trial?
$1,000 a month for the standard scope, with no setup fee. The price after the trial is agreed in writing before the trial starts. Follow-up questions need your team's own Claude or ChatGPT subscription.
What counts as additional work?
Anything that has to be built: a new source, marketplace, metric, or a report or dashboard we build for you. We scope and quote it, and start only after your approval. A new question about data already in your review isn't additional work.
Where is our data kept?
Metricbrief hosts it, separately for each client, and uses it only for your review.
Does our data go to Claude or ChatGPT?
The figures in your review are calculated from your Amazon reports without AI. An AI model, used through Metricbrief's account, writes the review's text from those figures. Follow-up questions run in your team's own Claude or ChatGPT account, which we connect only to the data included in your review. Your company's plan with Anthropic or OpenAI sets how they handle what the assistant reads.
How fresh is the data?
Some Amazon reports are available the next day, and ad figures can keep changing for a few days as late clicks are counted. Every review states the period it covers.
Can we run it in our own infrastructure?
Yes, as a separately quoted project: we install an independent system in your company's infrastructure. It is not part of the monthly service.
What happens if we stop?
Month to month, no minimum term, no cancellation fee. You get an export of your data and reports, and we delete them from our systems.
Which marketplaces and sources?
One Amazon marketplace in the standard scope, with Seller Central sales and traffic and Amazon Ads. Other marketplaces and reports are extensions. For Vendor Central or another source, tell us and we'll say if we can include it.
Are you an agency?
No. We don't run or advise on ads, listings or pricing. We prepare, check and run the recurring review. Your team, and your agency if you have one, make the decisions. We don't provide a dedicated analyst.