Design label classifications

Assigning Labels (individual classification values) to Search terms, ASINs, and campaigns lets you slice sales and ad spend on dashboards and N-gram analysis along axes like “offensive KW / defensive KW / brand / use / competitor.” This page collects prompts for designing a Taxonomy (the overall structure of the classification system), auto-labeling competitor ASINs, bulk-labeling Search terms, and managing naming conventions. In the AI era, differentiating bid and creative strategies based on search intent — why a shopper searched — is increasingly critical. Picaro Labels support 3-tier hierarchy (major → mid → minor) for analysis at a granularity that Amazon’s portfolios cannot match.

When to use this category

• You want to classify Search terms along the 5 axes of offensive KW / defensive KW / brand / use / competitor
• You want to auto-classify competitor ASINs that have slipped into Search terms by price band, reviews, and Category
• You want to apply Taxonomy rules — designed once — in bulk to existing Search terms
• You want to standardize campaign naming conventions for stable aggregation in reports
• You want to classify by search intent (Branded / Discovery / Solution-based / Competitive) and optimize bids and RoAS per intent type

How Picaro Labels differ from Amazon Portfolios — 3-tier hierarchy

Amazon Portfolios group campaigns in just 1 layer. Picaro Labels can be designed in 3 tiers (major → mid → minor) and applied across Search terms, ASINs, and campaigns. The ability to break down sales and ad spend to granularity like “within offensive KW, by use-case, then by specific scenario” is the key differentiator.

1 target = 1 Label constraint

In Picaro, only one Label can be attached per Search term / ASIN / campaign. When you assign a new Label, the existing one is silently overwritten and is not automatically reverted.

For this reason, always run Label assignment prompts as a 2-step flow: run with dry_run=true to review the execution plan → if it looks good, run with dry_run=false to execute.

---

Prompt 1: Auto-label competitor ASINs on 4 axes

🎯 Goal

Extract competitor ASINs that have slipped into Search terms, and generate and assign Labels that auto-classify them on 4 axes: price band, reviews, Category relevance, and product name.

📋 When to use

• You want to classify competitor ASINs mixed into Search terms for benchmarking
• You want to organize competitors along price band / review count / Category relevance axes
• You want to keep periodically labeling newly detected competitor ASINs

📝 Prompt

Extract ASIN-format strings (B0XXXXXXXX) from Search terms,
and generate Labels that auto-classify competitor ASINs on 4 axes:
1. Price band (low / mid / high)
2. Review rating (many reviews / high rating, etc.)
3. Category relevance
4. Product name

skip_labeled={{SKIP_LABELED}} (exclude existing Label),
limit={{LIMIT}} (required, for external API cost control).

First confirm the plan with dry_run=true →
if it looks good, run for real with dry_run=false.

Placeholders:

Placeholder	Description	Default source
{{SKIP_LABELED}}	Whether to exclude ASINs that already have Labels (e.g., true)	Default true (set false only when relabeling)
{{LIMIT}}	Upper bound on ASINs handled per run (e.g., 50)	Default 50 (required, to cap external product-info API costs)

⚙️ What the AI does

1. Extract ASINs from Search terms — pull B0XXXXXXXX-format strings from Search terms
   • Why pull from Search terms: to capture cases where users search for other products inside Amazon, so the competitor benchmarking universe has no gaps
2. Fetch product info — retrieve price, reviews, Category, and product name for the extracted ASINs via an external product-info service
   • Why limit is required: external product-info retrieval is metered, so running without a cap can blow up the bill
3. Generate Labels on 4 axes — auto-create Labels along the price band, review rating, Category relevance, and product name axes
   • Why 4 axes: combining price, reputation, Category match, and per-product views connects directly to competitor strategy (choosing differentiation axes, tuning expected price bands)
4. Skip ASINs that already have a Label — with skip_labeled=true, ASINs that already carry a Label are excluded from relabeling
   • Why skip: because of the 1 target = 1 Label constraint, to avoid accidentally overwriting an existing classification
5. Return an approval table — present 5 columns: ASIN, recommended Label, price, review count, Category

Cases requiring prep:

• For ASINs from overseas marketplaces (US / EU / UK), first register amazon_domain / currency / language via update_asin_enrichment in Prompt 5
• When not configured, defaults to Japan (amazon.co.jp / JPY / ja_JP)

📊 Example Output

Competitor ASIN auto-labeling candidates: 18 (new)

ASIN	Recommended Label	Price	Reviews	Category
B0XXXXXXXX	competitor / mid-price / many reviews	¥2,480	1,243	Hair care
B0YYYYYYYY	competitor / high-price / high rating	¥4,980	412	Hair care
B0ZZZZZZZZ	competitor / low-price / new	¥1,280	38	Body care
… (15 more) …

skip_labeled=true excluded 7 ASINs that already had Labels. Reply “execute” to run. To confirm one at a time, say e.g. “execute item 1 only.”

🔒 Safety guards

• Labels are not actually written until you confirm the plan with dry_run=true
• skip_labeled=true is the default, preventing silent overwrite of existing Labels
• Execution is rejected when limit is not specified (prevents external API cost runaway)
• Each proposal has a 1-hour validity; expired proposals must be regenerated
• If an overwrite occurs, roll back via Prompt 3 (manual assignment) by reassigning the original label ID

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Daily (on new ASIN detection)	Taxonomy registered + saved as a “Saved prompt”
Phase 3 (Q4 2026)	Auto-execution (approval flow)	Weekly auto + Slack 1-click approval, within volume cap
Phase 4 (2027)	Fully automated	ML confidence ≥ 0.85, within guardrails

👉 Next steps

• Analyze N-gram by Label — decompose sales contribution per competitor Label
• Review dashboard KPIs — check Taxonomy-grouped aggregates
• Why ACoS alone is not enough — the philosophy of judging in 3D along offensive / defensive / competitor axes

❓ FAQ

Q: Does this also label the Search term side? A: This prompt assigns Labels to “competitor ASINs.” For Search-term-side labeling, use Prompt 2.

Q: Can I cover overseas marketplaces in one go? A: Yes — register amazon_domain etc. via Prompt 5. If not registered, ASINs are interpreted under Japan’s product info.

---

Prompt 2: Bulk-label Search terms with the Taxonomy

🎯 Goal

Use Label patterns already registered in the Taxonomy to bulk-assign Labels to all Search terms ingested into Picaro.

📋 When to use

• You want to classify Search terms at once using your offensive KW / defensive KW / brand / use / competitor Taxonomy
• You want to relabel existing Search terms after updating your Label design
• You want consistent, rule-based classification without one-by-one manual assignment

📝 Prompt

For all Search terms registered in Picaro,
bulk-apply Labels using the Taxonomy rules.

skip_labeled={{SKIP_LABELED}} (exclude existing Label),
confirm the plan with dry_run=true →
if it looks good, run for real with dry_run=false.

Placeholders:

Placeholder	Description	Default source
{{SKIP_LABELED}}	Whether to exclude Search terms that already have Labels (e.g., true)	Default true (set false only when relabeling)

⚙️ What the AI does

1. Load Taxonomy patterns — fetch the “Label × match pattern” rules registered via Prompt 5
   • Why via Taxonomy: changing the Label design only requires a re-run to apply consistent classification across all Search terms
2. Match Search terms — evaluate each Search term against the registered patterns
3. Skip existing Labels — with skip_labeled=true, exclude Search terms that already have a Label
   • Why skip: because of the 1 Search term = 1 Label constraint, to avoid silently overwriting classifications you previously assigned by hand
4. Return an approval table — present 4 columns: Search term, planned Label, matched pattern, count

📊 Example Output

Taxonomy bulk-label candidates: 184

Search term	Planned Label	Matched pattern
amino acid shampoo	offensive KW / use	amino acid, shampoo
picaro original shampoo	defensive KW / brand	picaro
B0XXXXXXXX	competitor	ASIN format
camp shampoo	offensive KW / use	camp
… (180 more) …

Skipped 42 Search terms that already had Labels. Reply “execute” to run.

🔒 Safety guards

• Nothing is written until you confirm the plan with dry_run=true
• skip_labeled=true is the default, preventing silent overwrite of existing Labels
• Search terms that match no pattern are left unlabeled (no forced classification)
• Each proposal has a 1-hour validity
• If bulk overwrite introduces unintended classifications, narrow the target and reassign to the original label ID via Prompt 3

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Daily-runnable	Taxonomy registered + saved as a “Saved prompt”
Phase 3 (Q4 2026)	Auto-execution (approval flow)	Weekly auto + Slack 1-click approval
Phase 4 (2027)	Semi-automated	Approval only on new pattern detection; existing patterns auto-applied

👉 Next steps

• Design / update the Taxonomy — add or modify match patterns
• Decompose sales contribution via N-gram — see sales structure by Label
• Why ACoS alone is not enough — reading along offensive / defensive / competitor axes

---

Prompt 3: Manually label Search terms (label ID direct assignment)

🎯 Goal

Without going through Taxonomy auto-matching, directly assign Search terms or ASINs to a specified label ID.

📋 When to use

• You want to manually add ASINs to a competitor Label generated by Prompt 1
• You want to register, into a specific Label directly, Search terms that didn’t hit any Taxonomy pattern
• You want to fix terms that bulk-labeling misclassified by reassigning them to the intended Label

📝 Prompt

For label_id {{LABEL_ID}},
directly assign the following Search terms / ASINs:
{{TARGETS}}

Verify with dry_run=true →
if it looks good, run for real with dry_run=false.

Placeholders:

Placeholder	Description	Default source
{{LABEL_ID}}	The destination label ID (from the Taxonomy)	User input (retrievable via show in Prompt 5)
{{TARGETS}}	List of Search terms or ASINs (e.g., ["term1", "term2", "B0XXXXXXXX"])	User input

⚙️ What the AI does

1. Validate input — confirm that the label ID exists in the Taxonomy
   • Why validate: assignment to a non-existent ID will fail, so we catch it during dry_run
2. Interpret targets — split list entries into “ASIN format (B0XXXXXXXX)” and “Search terms”
   • Why auto-split: lets you mix Search terms and ASINs in one list, reducing input effort
3. Detect existing Labels — if a target already has a Label, flag it explicitly as an overwrite
   • Why flag it: because of 1 target = 1 Label, the user can confirm before an overwrite happens
4. Return an approval table — present 4 columns: target, type (Search term / ASIN), existing Label, post-assignment Label

📊 Example Output

Manual assignment candidates: 6

Target	Type	Existing Label	Post-assignment Label
competitor brand X shampoo	Search term	— (none)	competitor / high-price
B0AAAAAAAA	ASIN	— (none)	competitor / high-price
B0BBBBBBBB	ASIN	competitor / mid-price	competitor / high-price ← overwrite
… (3 more) …

1 overwrite will occur. Please confirm this is intended. Reply “execute” to run.

🔒 Safety guards

• Nothing is written until you confirm the plan with dry_run=true
• Targets with existing Labels are flagged with an overwrite indicator so you don’t unintentionally wipe out another Label
• A non-existent label ID stops with an error (typo protection)
• Each proposal has a 1-hour validity
• If you overwrite by mistake, reassign to the original label ID to restore

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Remains one-shot	Item-level scope; not an automation target
Phase 3 (Q4 2026)	Remains one-shot	Same as above
Phase 4 (2027)	Remains one-shot	Kept positioned as a manual correction tool

👉 Next steps

• Design / update the Taxonomy — if the same correction keeps recurring, register the pattern
• Bulk-label with the Taxonomy — relabel after reflecting the new pattern

---

Prompt 4: Assign Labels to campaigns / DSP / products

🎯 Goal

Assign Labels not only to Search terms and ASINs but also to Sponsored Products (SP) campaigns, DSP line items, and product ASINs — so they can be used as slicing axes in dashboards and N-gram.

📋 When to use

• You want to group Sponsored Products (SP) campaigns by “offensive / defensive / competitor” Labels for analysis
• You want to classify DSP line items with Labels
• You want to attach product Labels to product ASINs to add product-level aggregation axes

📝 Prompt

To label_id {{LABEL_ID}}, assign the following:
- SP campaigns: campaign_ids = {{CAMPAIGN_IDS}}
- DSP line items: line_item_ids = {{LINE_ITEM_IDS}}
- Product ASINs: product_ids = {{PRODUCT_IDS}}

Verify with dry_run=true →
if it looks good, run for real with dry_run=false.

Placeholders:

Placeholder	Description	Default source
{{LABEL_ID}}	The destination label ID	User input (retrievable via show in Prompt 5)
{{CAMPAIGN_IDS}}	List of SP campaign IDs (omit if not needed)	User input
{{LINE_ITEM_IDS}}	List of DSP line item IDs (omit if not needed)	User input
{{PRODUCT_IDS}}	List of product ASINs (omit if not needed)	User input

⚙️ What the AI does

1. Inspect applyTo — verify which targets (campaign / DSP / product) the Label can be applied to
   • Why inspect: each Label has eligible apply targets; this prevents assignment to unintended targets
2. Conflict detection (where possible) — for SP campaigns and DSP line items, detect collisions with existing Labels upfront
3. Flag overwrite risk on product ASINs — for Product assignments, no conflict-detection API exists, so the approval table emphasizes the possibility of an overwrite
   • Why emphasize: in the Product category — where 1 target = 1 Label combines with the absence of a detection API — overwrites can occur silently
4. Return an approval table — present 4 columns: target, type (SP / DSP / Product), existing Label status, post-assignment Label

📊 Example Output

Per-target Label assignment candidates: 12 (SP 6 / DSP 3 / Product 3)

Target ID	Type	Existing Label status	Post-assignment Label
cmp_XXXXXX	SP campaign	none	offensive KW / use
cmp_YYYYYY	SP campaign	competitor ← overwrite	offensive KW / use
li_ZZZZZZ	DSP line item	none	offensive KW / use
B0AAAAAAAA	Product ASIN	undetectable (caution)	offensive KW / use
… (8 more) …

For 3 Products, conflict detection is unavailable, so existing Labels may be silently overwritten. Reply “execute” to run.

🔒 Safety guards

• Nothing is written until you confirm the plan with dry_run=true
• SP / DSP have pre-execution conflict detection; for Product there is no conflict-detection API, so the caution flag is always shown
• Each proposal has a 1-hour validity
• If an unintended overwrite occurs, restore via Prompt 3 (manual assignment) by reassigning the original label ID

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	One-shot	Low structural change frequency; low automation priority
Phase 3 (Q4 2026)	Remains one-shot	Notify only when naming-convention violations are detected
Phase 4 (2027)	Semi-automated	Auto recommendations tied to naming conventions + approval

👉 Next steps

• Review dashboard KPIs — check aggregates by per-campaign Label
• Register / infer campaign naming conventions — derive Label design from naming conventions

---

Prompt 5: Manage the Taxonomy (classification system)

🎯 Goal

Centralize Taxonomy management — current state, additions, updates, overseas marketplace setup, and reset — for Category hierarchy → Labels, all in a single prompt.

📋 When to use

• You want to see what Taxonomy is currently registered
• You want to add new Label hierarchies like offensive KW / defensive KW / brand / use / competitor
• You want to configure product-info retrieval for overseas marketplaces (US / EU / UK)
• You want to wipe the Taxonomy and start over from the built-in demo values

📝 Prompt

Taxonomy operation: {{ACTION}}

- show: display current state
- clone_default: copy from demo values
- upsert_category: add or update a Category hierarchy
- upsert_label: add or update a Label
    (includes match pattern, rule type, Picaro-side label ID)
- update_asin_enrichment: overseas marketplace setup
    (amazon_domain / currency / language)
- create_picaro_label: create a new Label on Picaro side
- reset: delete everything (destructive, requires confirmation)

Placeholders:

Placeholder	Description	Default source
{{ACTION}}	The operation to run (one of show / clone_default / upsert_category / upsert_label / update_asin_enrichment / create_picaro_label / reset)	User input

⚙️ What the AI does

1. Fetch current state with show — display the full Taxonomy: Category hierarchy, Label list, match patterns, overseas marketplace settings
   • Why start with show: upserting without understanding existing design tends to cause conflicts
2. Initialize (if needed) — copy Picaro’s built-in demo Taxonomy via clone_default
3. Add / update — add or update Category hierarchies via upsert_category and Labels via upsert_label
   • Why upsert: “update if exists, otherwise add” completes in a single operation
4. Overseas marketplace setup — register amazon_domain (e.g., amazon.com), currency, and language via update_asin_enrichment
   • Why this is needed: for overseas-marketplace ASINs, the default amazon.co.jp won’t return meta info, which degrades labeling accuracy in Prompt 1
5. Warning on destructive ops — reset wipes everything, so it inserts a confirmation step before execution
   • Why confirm: the Taxonomy is a design asset; rebuilding an equivalent after deletion takes time

Typical flow:

1. Confirm current state with show
2. On first setup, copy built-in demo values with clone_default
3. Customize to offensive / defensive / brand / use / competitor etc. via upsert_category / upsert_label
4. If you handle overseas marketplaces, register via update_asin_enrichment

📊 Example Output

Current Taxonomy (show)

Major	Mid	Minor (Label)	Match pattern	Picaro-side label ID
Solution-based	use	Camp	camp, outdoor	lbl_XXXXXX
Solution-based	use	Pro-use	pro-use, salon	lbl_YYYYYY
Branded	brand	—	ピカロ, picaro	lbl_ZZZZZZ
Competitive	competitor ASIN	—	ASIN format	lbl_AAAAAA

Overseas marketplace setup: amazon.co.jp / JPY / ja_JP (unchanged)

Specify the next operation you want to add or update.

🔒 Safety guards

• reset always inserts a confirmation step; nothing is deleted until you explicitly reply “execute”
• Changing match patterns via upsert_label does not immediately relabel existing Search terms (relabel separately via Prompt 2)
• Overseas marketplace settings keep using the default (amazon.co.jp) until you run update_asin_enrichment
• Past Taxonomy settings are retained as version history; you can roll back to the built-in demo with clone_default even after reset

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Remains one-shot	Low design frequency; not an automation target
Phase 3 (Q4 2026)	Remains one-shot	Same as above
Phase 4 (2027)	Semi-automated	Auto-suggest new Label candidates from history + approval

👉 Next steps

• Bulk-label with the Taxonomy — reapply to existing Search terms after a design change
• Auto-label competitor ASINs on 4 axes — run only after overseas marketplace setup is done
• Why ACoS alone is not enough — the philosophy of which Label axes to choose

❓ FAQ

Q: If I update the Taxonomy, do existing Search-term Labels rewrite automatically? A: No, they don’t auto-rewrite. To apply a match-pattern change, re-run Prompt 2.

Q: What’s the relationship between label ID (lbl_XXXXXX), “Category hierarchy,” and “Label name”? A: The “Taxonomy” is the entire hierarchical structure, a “Label” is an individual classification value at a leaf node, and a “label ID” is an internal ID uniquely identifying that leaf node. The ID is used to assign Labels to targets in Prompts 3 / 4.

---

Prompt 6: Register / infer campaign naming conventions

🎯 Goal

Register naming templates for campaigns and ad groups to enforce consistent naming at creation time. You can also infer naming patterns from existing names.

📋 When to use

• You want consistent naming when creating new campaigns
• You want to extract the team’s implicit naming rules from already-created campaign names and templatize them
• You want to update an existing naming template

📝 Prompt

Manage campaign / ad group naming conventions:

- Infer: from {{SAMPLE_SIZE}} existing campaign names, return a draft naming pattern
- Register: register the naming template {{TEMPLATE}}
    (e.g., {brand}_{product_line}_{match_type})
- Update: update the existing rule

Placeholders:

Placeholder	Description	Default source
{{SAMPLE_SIZE}}	Number of existing campaigns to sample for inference (e.g., 50)	Default 50
{{TEMPLATE}}	Naming template to register (e.g., {brand}_{product_line}_{match_type})	User input

⚙️ What the AI does

1. Sample existing names — extract common prefixes, separators, and token structure from SAMPLE_SIZE campaign names
   • Why sample: processing all entries is noisy; 30–100 names is enough to capture the representative naming style
2. Generalize tokens — generalize concrete examples like SP_picaro_shampoo_Exact to a form like {ad_type}_{brand}_{product_line}_{match_type}
3. Register / update the template — save either the inferred result or the user-specified template on the Picaro side
4. Reconcile with existing names — optionally list existing campaigns that don’t conform to the registered template as “naming violations”

Registered naming conventions are automatically applied at creation time in Create a campaign. When none is registered, the built-in default (picaro-default-v1) is used.

📊 Example Output

Inferred naming pattern (sample 50)

Inference	Match rate	Example
{ad_type}_{brand}_{product_line}_{match_type}	78%	SP_picaro_shampoo_Exact
{ad_type}_{product_line}_{target_type}	14%	SP_shampoo_Auto
Other	8%	(4 naming-violation candidates)

To register the top pattern as the naming template, reply “register option 1.”

🔒 Safety guards

• Template registration takes effect only after you reply “register” explicitly
• Existing campaign names are not modified (only listed as naming violations; renaming is a separate task)
• Registered naming conventions are kept as history and can be rolled back to a previous version
• Setting SAMPLE_SIZE extremely small risks overfitting and producing an incorrect template; the default 50 is recommended

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Remains one-shot	Low design frequency; not an automation target
Phase 3 (Q4 2026)	Naming-violation detection + notification	Slack-notify template violations on new campaign creation
Phase 4 (2027)	Auto-naming	Auto-name from the template + approval only on violations

👉 Next steps

• Create a campaign — auto-apply the registered naming convention at creation time
• Assign Labels to campaigns / DSP / products — apply Labels in a form that aligns with the naming convention

---

---

Prompt 7: Design and classify labels by search intent

🎯 Goal

Extract 4 search intent types (Branded / Discovery / Solution-based / Competitive) from existing Search-term data and design a 3-tier Taxonomy (major → mid → minor) for registration.

📋 When to use

• You want to design a Taxonomy on the 4 intent axes: Branded / Discovery / Solution-based / Competitive
• You want to extract representative terms per intent type from existing Search-term data and get a 3-tier Taxonomy draft
• You want to organize bid-strategy rationale around per-intent RoAS / CVR

📝 Prompt

Propose a Taxonomy design that classifies Search terms by 4 purchase-intent types
(Branded / Discovery / Solution-based / Competitive):

1. Extract 10 representative examples per intent type from current Search-term data
   and show the reasoning for each classification
2. Design a 3-tier Taxonomy: major (intent type) / mid (topic) / minor (specific term)
3. Present match patterns for each tier

After reviewing, reply "register" to add to the Taxonomy.

⚙️ What the AI does

1. Extract representative terms by intent type — mapping between the 4 intent types and existing Taxonomy axes:
   • Branded: terms containing the brand or product name → defensive KW / brand
   • Discovery: category names and generic terms → offensive KW / generic
   • Solution-based: terms containing use-cases or problems → offensive KW / use
   • Competitive: competitor brand names and competitor ASINs → competitor
   • Why intent classification: bids can be justified by “degree of purchase intent,” and intent signals strengthen AI optimization engines
2. Generate 3-tier Taxonomy draft — place intent types at the major tier, topic groups at the mid tier, and specific terms at the minor tier (e.g., major “Solution-based” > mid “use-case / scene” > minor “camp”)
   • Why 3 tiers: Amazon Portfolios provide only 1-layer campaign grouping. 3 tiers enable aggregation at the granularity of “RoAS for camp-use within Solution-based only”
3. Present match patterns — list keyword patterns tied to each minor Label
4. Guide to registration — show registration steps using Prompt 5’s upsert_category / upsert_label
5. Guide to bulk application — remind that Prompt 2 must be run separately after registration to apply Labels to existing Search terms

📊 Example Output

Search-intent Taxonomy proposal

Major (intent type)	Mid (topic)	Minor (specific term)	Representative examples
Branded	brand	—	picaro, picaro shampoo
Discovery	category	hair care	amino acid shampoo, silicone-free
Solution-based	use-case / scene	camp	camping wash, outdoor
Solution-based	use-case / scene	pro-use	pro-use shampoo
Competitive	competitor brand	—	Competitor X shampoo, B0XXXXXXXX

In the AI era, bid and creative strategies tailored to why a shopper searched are especially important. Reply “register” to add this Taxonomy.

🔒 Safety guards

• Nothing is written to Picaro until you explicitly reply “register”
• Registering the Taxonomy does not immediately relabel existing Search terms — run Prompt 2 separately for bulk application
• If intent-type names conflict with existing Taxonomy categories, confirm with show before running upsert_category

🚦 Execution modes

Phase	State	Conditions
Phase 1 (now)	One-shot	Available with Picaro connection only
Phase 2 (Q3 2026)	Remains one-shot	Low design frequency; not an automation target
Phase 3 (Q4 2026)	Intent-drift detection + notification	Slack-notify when new Search terms fall outside existing intent patterns
Phase 4 (2027)	Semi-automated	Auto-infer intent for new terms + approval flow

👉 Next steps

• Bulk-label Search terms with the Taxonomy — apply intent-type Labels to existing Search terms
• Decompose sales contribution via N-gram — break down sales by intent type and tier
• SQP funnel analysis — combine search intent with funnel stage for deeper insight

---

Prompt quick reference

What you want to do	Prompt
Auto-classify competitor ASINs on 4 axes	Prompt 1
Bulk-label Search terms with the Taxonomy	Prompt 2
Manually assign / correct to a specific Label	Prompt 3
Label SP campaigns / DSP / products	Prompt 4
Design / update the Taxonomy, overseas marketplace setup	Prompt 5
Register / infer campaign naming conventions	Prompt 6
Design Labels by search intent (Branded / Discovery / Solution-based / Competitive)	Prompt 7

Related categories

• Review dashboard KPIs — prerequisite for Taxonomy-grouped KPI aggregation
• Decompose sales contribution via N-gram — destination for Taxonomy-based clustering
• SQP funnel analysis — read market positioning by Label
• Create a campaign — destination for using naming conventions and Label design
• Why ACoS alone is not enough — the philosophy behind the 5-axis Label design of offensive / defensive / brand / use / competitor