Review live SB and SD campaigns

On this page you can review SB (Sponsored Brands) and SD (Sponsored Display) campaigns that are already running. Building new campaigns lives in Create Sponsored Brands (SB) campaigns and Create Sponsored Display (SD) campaigns; adjusting bids, delivery state, and budgets after the build is this page.

About this page

Item	Details
What you can do here	Take stock of SB and SD delivery state and performance, review SB keyword and product-targeting bids, review SD targeting bids and ad group default bids, and shortlist underperforming campaigns to pause or scale back
Applies to	Sponsored Brands (SB), Sponsored Display (SD)
Data and connections required	A connected Amazon Ads account. A target ACoS (or target ROAS) and the reporting window you want to judge on
Scope of execution	Varies by prompt. Taking stock (Prompt 1) is read-only; changing bids, state, or budget (Prompts 2, 3, 4) modifies Amazon Ads settings after approval. Each prompt states this under “Settings changed”

Tool constraints these prompts are written around (important)

Reading SB and SD data has different constraints from SP. The prompt bodies are written with these in mind.

• The campaign list has no argument for filtering by name. You have to reconcile total_results against the number of rows actually returned and page through at 100 rows per page.
• You cannot treat the sum of the returned rows as an account-wide total. The list response itself states that summing rows only produces a partial subtotal; when you need account-wide figures, use Check KPIs on the dashboard instead.
• Target-level performance only works when the ad type is given in lowercase (sb / sd). Passing a campaign ID still returns the whole account, so filtering has to happen after the fetch. Fetching is 100 rows per page; page through until you reach total_results.
• One performance route ignores the ad type and campaign filters and stops at the first page. Ranking only the rows that route returned puts the wrong items at the top. The prompt bodies require the agent to report both the number of rows fetched and the size of the full population.
• Some performance rows come back with an empty target name or campaign name (older rows for things that were paused). Do not fill names in by guessing; the prompts ask for these to be listed separately as “rows whose name could not be retrieved”.
• The current bid and state (bid / state) and the period performance (ACoS, ROAS, CVR) come from different tools. Join them on the target ID, and separate out any target that exists on only one side.

Terms used on this page

• Bid — the maximum cost per click set on that keyword or target.
• Default bid — the ad group’s default bid, applied to targets that have no individual bid.
• State — enabled / paused / archived.
• ACoS — ad spend divided by ad-attributed sales. Lower is more efficient.
• ROAS — ad-attributed sales divided by ad spend; the inverse of ACoS.
• Product targeting — targeting specific products (ASINs), used in SB and SD.

An Amazon Ads account connection is required

The prompts in this category read and write Amazon Ads settings, so complete Prompt 3 of Connect your ads account first. Taking stock (Prompt 1) is read-only.

Every change runs in propose-then-approve mode

Prompts 2, 3, and 4 first return a list of the proposed changes (target, current value, new value, rationale). Nothing is written to Amazon Ads until you say “go ahead”.

Before you move a bid, check there are enough clicks to judge on

The ACoS of a target with only a handful of clicks swings wildly on a single order. These prompts make you specify a minimum click count and hold back anything below it as “not enough data to judge”. There is no default, so set it to match the size of your account.

When to use this

• You have lost track of how many SB and SD campaigns are live and which of them are performing
• SB keyword and product-targeting bids have been left at whatever they were when the campaign was built
• Every SD target still sits at the same bid, so performance differences are not reflected in bids
• You want to pause SB and SD campaigns that are spending without producing sales

Before you run

• Which account — if you handle several accounts, confirm the target account first.
• Target ACoS (or target ROAS) — without it there is no basis for calling something high or low.
• Reporting window — for example the last 30 days. Too short a window will not accumulate enough clicks.
• Minimum click count to judge on — anything below this is held back.
• Settle the overall stance (new-customer growth versus efficiency) in Build an advertising strategy first, so bid moves have a consistent rationale.

Prompt quick reference

What you want to do	What you get	Settings changed	Prompt
Take stock of SB and SD delivery state and performance	A list of live campaigns with performance, grouped by what needs attention	None (read-only)	Prompt 1
Review SB bids against performance	Proposed bid changes per keyword and product target	Changes ad settings after approval (bids, state)	Prompt 2
Review SD bids against performance	Proposed bid changes per target and for ad group default bids	Changes ad settings after approval (bids, state)	Prompt 3
Pause or scale back underperforming campaigns	A shortlist of campaigns to pause or cut, with rationale	Changes ad settings after approval (state, budget)	Prompt 4

Nothing reaches your ads until you approve it

Prompts 2, 3, and 4 only return a list of proposed changes. They write to Amazon Ads the moment you say “go ahead”, and not before.

---

Prompt 1: Take stock of SB and SD delivery state and performance

Fetches every live SB and SD campaign and lists them alongside performance. Read-only; nothing is changed.

When to use this — before you start reviewing, to see what is actually running and where the spend is going.

What you need first — the reporting window, your target ACoS, and which ad types to cover.

Take stock of my live SB and SD campaigns.

- Reporting window: {{reporting window (e.g. last 30 days)}}
- Target ACoS: {{target ACoS (e.g. 40%)}}
- Ad types to cover: {{scope (SB only / SD only / both SB and SD, e.g. both SB and SD)}}

Steps:
1. Fetch the campaign list for those ad types from Picaro MCP.
   The list has no argument for filtering by name, so reconcile total_results against the
   number of rows actually returned and page through at 100 rows per page until you have them all.
   If you cannot fetch them all, state how many of how many you retrieved.
2. Show the count by delivery state (ENABLED / PAUSED / ARCHIVED).
   Analyse only the ENABLED campaigns from here on, and report PAUSED and ARCHIVED as counts only.
3. For each live campaign, put the following in a table:
   - campaign name / campaign_id / ad type
   - daily budget and budget type
   - portfolio (write "not set" if there is none)
   - period performance (impressions, clicks, spend, sales, orders, ACoS, ROAS)
   For campaigns whose performance could not be joined, do not write zero: separate them out and
   state that performance could not be retrieved, and why.
4. Sort the live campaigns into these four groups, with the count and total spend for each:
   - meeting target ACoS
   - exceeding target ACoS
   - spending but with zero sales
   - zero impressions and zero clicks in the period (effectively not running)
5. Do not sum these rows and present the result as account-wide spend or sales.
   The returned rows may be only part of the population, so their sum is only a partial subtotal.
   If an account-wide figure is needed, say so and point to the separate source for it.

Do not fill in any figure, ID, or setting that you did not retrieve from Picaro MCP.
For anything you could not retrieve, state that it could not be retrieved, and why.

What you can change — the reporting window, target ACoS, and ad types covered.

What you get back — a table of live SB and SD campaigns and a four-way split by performance, with counts and total spend.

Settings changed — none (read-only).

What to do next — review bids on the over-target campaigns with Prompts 2 and 3; treat the zero-sales and zero-delivery campaigns as pause candidates in Prompt 4.

---

Prompt 2: Review SB keyword and product-targeting bids

Joins current bids with period performance at the keyword and product-target level in SB, and proposes bid changes.

When to use this — when an SB campaign’s ACoS is off target and you need to find which keywords or product targets are responsible.

What you need first — the campaign, reporting window, target ACoS, minimum click count, and a cap on how far a bid may move.

Review the keyword and product-targeting bids on the SB campaign
"{{campaign name (e.g. SB | Brand A | Generic Keyword)}}".

- Reporting window: {{reporting window (e.g. last 30 days)}}
- Target ACoS: {{target ACoS (e.g. 40%)}}
- Minimum clicks needed to judge: {{minimum clicks (e.g. 10)}}
- Cap on a single change: {{cap (e.g. plus or minus 30% of the current bid)}}

Steps:
1. Identify the SB campaign from real Picaro MCP data and show me the campaign_id.
   If the name does not match, or several campaigns share it, do not pick one by guessing:
   list the candidates and ask me to confirm.
2. Fetch the campaign's ad groups, then the keywords and product targets in each one.
   List the current bid, match type, and state for each. State how many rows you retrieved
   and whether another page remains.
3. Fetch target-level performance for the period. The ad type must be given in lowercase as sb.
   Passing a campaign ID still returns the whole account, so filter to the campaign_id after fetching.
   Fetch 100 rows per page and page through until you reach total_results, and always report
   "rows retrieved / size of the full population".
   If you stop early, say so, and state that the ranking may therefore differ from the true ranking.
4. Join the settings from step 2 to the performance from step 3 on the target ID.
   - For rows whose target name or campaign name comes back empty, do not guess the name:
     list them separately as rows whose name could not be retrieved.
   - List targets that exist on only one side separately, with a count.
5. Sort the enabled targets into these groups and propose a change for each:
   - fewer clicks than the minimum -> hold back (no change). Give only the count and total spend
   - ACoS above target -> propose a bid decrease, with the new bid and the rationale (ACoS, clicks, spend)
   - ACoS below target with orders -> propose a bid increase to widen exposure
   - clicks but no orders -> candidate for a bid decrease or a pause. Say which you recommend, and why
6. Present the proposed changes in this form:
   type (keyword / product target), target name, target ID, ad group, current bid, new bid,
   percentage change, period clicks / spend / sales / ACoS, and the reason for the change.
   Keep every change within the cap, and for anything you would have moved further,
   stop at the cap and say that the cap was reached.
7. Estimate the resulting change in spend, holding the period's click volume constant,
   and state that this is an estimate under that assumption.
8. Do not write anything to Amazon Ads until I have reviewed it and approved.

Do not fill in any figure, ID, or setting that you did not retrieve from Picaro MCP.
For anything you could not retrieve, state that it could not be retrieved, and why.

What you can change — the campaign, reporting window, target ACoS, minimum clicks, and the cap.

What you get back — proposed bid changes per keyword and product target, with current values and rationale, plus the held-back list.

Settings changed — after approval, the bids and state of SB keywords and product targets.

What to do next — for search terms that do not improve after a decrease, consider Add negative keywords. Track the effect with Log initiatives and external events to verify their effect.

---

Prompt 3: Review SD targeting bids and ad group default bids

Reviews SD bids at the target level and the ad group default bid against period performance.

When to use this — when SD targets are all sitting at the bid they were built with and performance differences are not reflected in bids.

What you need first — the campaign, reporting window, target ACoS, minimum click count, and a cap on how far a bid may move.

Review the targeting bids and ad group default bids on the SD campaign
"{{campaign name (e.g. SD | Brand A | Context_Competitor Product)}}".

- Reporting window: {{reporting window (e.g. last 30 days)}}
- Target ACoS: {{target ACoS (e.g. 40%)}}
- Minimum clicks needed to judge: {{minimum clicks (e.g. 10)}}
- Cap on a single change: {{cap (e.g. plus or minus 30% of the current bid)}}

Steps:
1. Identify the SD campaign from real Picaro MCP data and show me the campaign_id.
   If the name does not match or several campaigns share it, list the candidates and ask me
   to confirm rather than guessing.
2. Fetch the campaign's ad groups and show each one's default bid and state.
3. Fetch the targets in each ad group and list the target ID, bid, state, and targeting
   expression (type and value). State how many rows you retrieved and whether another page remains.
4. Fetch target-level performance for the period. The ad type must be given in lowercase as sd.
   Passing a campaign ID still returns the whole account, so filter to the campaign_id after fetching.
   Fetch 100 rows per page, page through until you reach total_results, and always report
   "rows retrieved / size of the full population".
5. Join the settings from step 3 to the performance from step 4 on the target ID, and give a
   separate count for targets that could not be matched.
6. Sort the enabled targets into these groups and propose a change for each:
   - fewer clicks than the minimum -> hold back (no change). Give only the count and total spend
   - ACoS above target -> propose a bid decrease
   - ACoS below target with orders -> propose a bid increase
   - spending but no orders -> candidate for a bid decrease or a pause. Say which you recommend, and why
7. If an ad group has every target sitting at the same bid, flag it as possibly still on the
   build-time default with no individual tuning, and propose setting individual bids first on
   the targets whose performance diverges most.
8. For the ad group default bid, propose raising it, lowering it, or leaving it, against the
   ad group's overall ACoS, and give your rationale.
9. Present the proposed changes in this form:
   what is being changed (target / ad group default bid), target name or targeting expression,
   target ID, current bid, new bid, percentage change, period clicks / spend / sales / ACoS,
   and the reason. Keep every change within the cap and flag anything that stopped at the cap.
10. Do not write anything to Amazon Ads until I have reviewed it and approved.

Do not fill in any figure, ID, or setting that you did not retrieve from Picaro MCP.
For anything you could not retrieve, state that it could not be retrieved, and why.

What you can change — the campaign, reporting window, target ACoS, minimum clicks, and the cap.

What you get back — proposed bid changes per SD target and for ad group default bids, with current values and rationale, plus the held-back list.

Settings changed — after approval, SD target bids and state, and ad group default bids.

What to do next — if there are products or categories you do not want to appear against, add negative targeting with Prompt 4 of Create Sponsored Display (SD) campaigns.

---

Prompt 4: Shortlist underperforming SB and SD campaigns to pause or scale back

Works through your live SB and SD campaigns and shortlists what to pause and what to cut back, with the evidence.

When to use this — when spend is over target and you need to decide what to cut first, or as part of a monthly or quarterly review.

What you need first — the reporting window, target ACoS, the conditions that make something a pause candidate, and which ad types to cover.

Shortlist pause candidates and budget-reduction candidates among my live SB and SD campaigns.

- Reporting window: {{reporting window (e.g. last 30 days)}}
- Target ACoS: {{target ACoS (e.g. 40%)}}
- Conditions for a pause candidate: {{conditions (e.g. at least 5,000 JPY spent with zero sales, or ACoS at least twice the target)}}
- Ad types to cover: {{scope (SB only / SD only / both SB and SD, e.g. both SB and SD)}}

Steps:
1. Fetch the full campaign list for those ad types.
   The list has no argument for filtering by name, so reconcile total_results against the rows
   returned and page through at 100 rows per page. If you cannot fetch them all, say so.
2. Join period performance for the ENABLED campaigns only.
   For campaigns whose performance could not be joined, do not write zero: list them separately as
   performance could not be retrieved, exclude them from the judgement, and say that you did.
3. Present the campaigns that meet the conditions in three groups:
   - pause candidates: they meet the conditions and look unlikely to improve
   - budget-reduction candidates: they are producing sales but efficiency is off target, so the
     budget should be cut back and watched
   - held back: they meet the conditions but there are not enough clicks or not enough elapsed
     time to judge
   For each, give the campaign name, campaign_id, ad type, daily budget,
   period spend / sales / orders / ACoS, and why you put it in that group.
4. For the pause and budget-reduction candidates, estimate the spend that would be saved based on
   actual period spend, and state that this is an estimate and on what assumption.
5. Exclude the following from the pause list even if they meet the conditions, and say why:
   - campaigns targeting your own brand terms or your own products (they may be defensive)
   - campaigns that started recently and may not have finished learning
   If you cannot retrieve enough information to judge this, do not exclude them: state that the
   information was not available.
6. For budget-reduction candidates, propose the new daily budget and give your rationale.
7. Do not write anything to Amazon Ads (state changes or budget changes) until I have reviewed
   and approved. If you do apply changes after approval, keep a list of the previous values.

Do not fill in any figure, ID, or setting that you did not retrieve from Picaro MCP.
For anything you could not retrieve, state that it could not be retrieved, and why.

What you can change — the reporting window, target ACoS, the pause conditions, and the ad types covered.

What you get back — a three-way shortlist (pause, reduce, hold back), an estimate of the spend saved, and proposed new daily budgets.

Settings changed — after approval, the state and daily budget of SB and SD campaigns. The previous values are kept in a list, so use those if you need to revert.

What to do next — decide where the freed budget goes with Adjust budgets and Manage budgets across several campaigns.

---

Related categories

• Create Sponsored Brands (SB) campaigns — building new SB campaigns
• Create Sponsored Display (SD) campaigns — building new SD campaigns
• Adjust bids — bid optimization, centred on SP
• Adjust budgets — reviewing daily budgets and budget caps
• Find high-performing keywords and product targeting — decide what to grow first
• Log initiatives and external events to verify their effect — measuring the effect of your changes