An Amazon PPC automatic campaign uses Amazon’s matching to discover relevant shopper queries and product contexts through four groups: close match, loose match, substitutes, and complements. Choose a default bid for a simple starting point, or targeting-group bids when the groups need different treatment. Review each group’s underlying traffic before deciding what to keep exploring, reduce, exclude, or transfer to manual targeting.
Platform documentation checked 2026-11-06: Amazon Ads’ Sponsored Products targeting guide. This is a documentation-based operating guide, not an account performance test. The control tables below are working templates; the numerical example is hypothetical.
For your first launch, use the beginner PPC setup. This guide focuses on managing automatic targeting after you have defined the product and advertising objective—not on rebuilding your account.
Amazon PPC automatic campaign: what the four groups control
Amazon’s targeting guide, checked 2026-11-06, defines close and loose match around shopper search terms, and substitutes and complements around related product detail pages. These are matching relationships, not four promises of conversion quality.
| Targeting group | Amazon’s matching relationship | Question to answer in your review | Candidate action |
|---|---|---|---|
| Close match | Search terms closely related to the advertised product | Are the actual queries relevant to this specific variation? | Retain useful discovery; nominate proven queries for manual keyword targeting |
| Loose match | Search terms loosely related to the advertised product | Is broader wording revealing a useful use case or an unrelated need? | Separate promising queries from mismatches before changing the group bid |
| Substitutes | Detail pages of products similar to the advertised product | Would a shopper reasonably consider your product instead? | Evaluate individual product contexts for a manual product-targeting shortlist |
| Complements | Detail pages of products that complement the advertised product | Does the pairing make sense for compatibility and purchase intent? | Retain useful pairings; investigate irrelevant product contexts |
Use that table as a review map, not a predetermined bid ladder. Close match can contain expensive traffic that does not meet your objective. A complementary relationship can be sensible but still fail commercially because the shopper is not ready to buy your item.
For a hypothetical insulated lunch bag, closely related wording might describe the same bag and use case; broader wording might describe carrying meals more generally. Another lunch bag is a substitution hypothesis, while a compatible food container is a complementary hypothesis. These illustrations explain the relationships; they do not predict what Amazon will match or where an ad will appear.
Close match is not exact match
Close match is an automatic targeting group. Exact match is a manual keyword match type. Amazon’s guide, checked 2026-11-06, describes them separately; a close-match label does not mean you selected the shopper’s precise query as an exact keyword.
Preserve that distinction when harvesting. Record the observed query, then decide how to target it manually using the keyword match-type guide. Copying the group name into a keyword plan loses the actual discovery.
Similarly, substitutes and complements are not lists of ASINs you personally selected. The manual destination gives a selected product context its own deliberate targeting decision; it does not turn the original automatic group into a fixed shortlist.
Default bid or group bids: choose a readable starting point
Amazon’s targeting guide, checked 2026-11-06, explicitly offers a single default bid or bids by automatic targeting group. The useful distinction is whether you need one starting instruction or different instructions for different matching relationships.
| Bid approach | Use it when | What to write down |
|---|---|---|
| Single default bid | You want a simple baseline before the groups have differentiated evidence | Starting bid, product, objective, review window, and acceptable discovery spend |
| Targeting-group bids | Evidence or a clear product hypothesis calls for different treatment by group | Each group’s bid, reason for the difference, and the condition for revisiting it |
Do not assign a higher close-match bid merely because its name sounds more precise. Start with the product’s advertising economics, then use observed relevance and returns to decide which group deserves a different bid. The profit-first PPC playbook covers the margin calculation; the automatic-group label cannot supply it.
Keep bid amount, bidding strategy, and budget decision separate in your notes. A group bid expresses your targeting decision, not a guaranteed click price or a commitment to spend a fixed amount on that group. For the behavior of dynamic versus fixed bidding, use the bidding-strategy guide.
When introducing group bids, preserve the old baseline. Record which groups changed and which stayed unchanged. Avoid changing every bid, the advertised offer, and other campaign settings together: even a better result would leave you unsure which intervention deserves credit.
If you lack enough evidence to distinguish the groups, simplicity is useful. Keep the baseline readable and define what evidence would justify a change rather than creating four arbitrary numbers that require four separate explanations later.
Read performance by group before changing the campaign
Build a worksheet with the same date range for every group. Record impressions, clicks, spend, attributed orders, and attributed sales from the reporting available to you, alongside the advertised product and current bid. If you combine exports, preserve identifiers so a similarly named ad group does not receive another group’s results.
Use campaign totals for the overall spending decision and group rows to locate the reason behind it. Then inspect the underlying queries or product contexts before making an exclusion. Amazon’s targeting guide, checked 2026-11-06, identifies the automatic campaign search term report as a source of keyword and product-target discoveries.
A blended result can hide different decisions
The following hypothetical worksheet assumes the same observation window and currency for every row. It is not an Amazon benchmark or an account result. ACoS here means advertising spend divided by attributed sales, multiplied by 100.
| Group | Spend | Attributed sales | ACoS | What to investigate next |
|---|---|---|---|---|
| Close match | $24 | $120 | 20% | Which queries produced the sales, and whether those queries fit the objective |
| Loose match | $30 | $60 | 50% | Whether a few mismatches or broadly weak traffic explain the cost |
| Substitutes | $12 | $80 | 15% | Which individual product contexts deserve a shortlist |
| Complements | $14 | $0 | Undefined | Relevance, clicks, reporting completeness, and the discovery allowance |
| Combined | $80 | $260 | 30.77% | Overall spending outcome, not a substitute for the group decisions |
A combined ACoS of 30.77% does not establish whether the campaign is profitable: that requires the product’s economics and your objective. Nor does the 15% substitutes row establish repeatability without knowing the underlying orders and traffic concentration.
Zero attributed sales makes the ACoS calculation undefined, not 0%. Review the spend and evidence directly rather than interpreting a blank or unavailable ratio as efficient performance.
Separate irrelevance from insufficient evidence
A plainly incompatible use case calls for a different response from a relevant query with little traffic. Review the query or product context against the actual advertised variation: dimensions, compatibility, pack size, intended use, and the offer the shopper sees.
For relevant but thin traffic, record an observation decision and its spending boundary. A low-impression row is not evidence that the group cannot work. For costly but relevant traffic, examine click cost and conversion economics before assuming a relevance exclusion will solve the problem.
Do not invent a universal number of clicks that proves a group should stop. A useful review rule needs the product’s economics, the cost already incurred, and enough reporting context to distinguish a weak opportunity from an unfinished observation window.
Run a controlled review loop
Use a repeatable loop: define the window → inspect the group → inspect its underlying traffic → make a specific change → review against the recorded reason. The purpose is to make the next decision explainable, not to promise an optimal bid.
1. Define the observation before reading the result
Write down the product, marketplace, objective, date range, acceptable discovery spend, and who owns the next review. Check whether the reporting period is sufficiently complete for the decision; do not treat the newest visible figures as final merely because they are available.
Amazon’s targeting guide, checked 2026-11-06, recommends running an automatic campaign for about two weeks before creating a manual campaign. That recommendation is not a guarantee of sufficient traffic or a reason to ignore an obvious mismatch until a calendar deadline passes. Use it as a planning reference, then consider the actual evidence and spend.
2. Give each group one current decision
Choose a primary action for each group: continue observing, investigate specific traffic, adjust its bid, or nominate a discovery for manual targeting. If several changes are necessary, record each separately rather than describing the entire campaign as optimized.
| Review-note field | Entry to keep |
|---|---|
| Source | Campaign, ad group, advertised ASIN, automatic group |
| Evidence | Date range, spend, clicks, orders, sales, and relevant query or product context |
| Diagnosis | Relevant but sparse; irrelevant; costly; or promising discovery |
| Intervention | Exact setting or target to change, with its previous value |
| Expected signal | What improvement or learning the change should produce |
| Review condition | Next comparable window or the discovery-spend boundary |
3. Change the narrowest control that answers the diagnosis
If one irrelevant query explains the problem, investigate a targeted exclusion before reducing an otherwise useful group. If relevant traffic across a group is too costly for the objective, a group-bid adjustment may be the clearer experiment. Negative keywords and product exclusions are different tools; use the negative-targeting workflow to choose the appropriate scope.
4. Compare like with like
At the next review, look for the expected signal and record any offer, inventory, or other campaign changes that complicate the comparison. An improvement after a bid change is an observation, not proof that the change caused it.
Keep a useful change, reverse an unsuccessful one, or extend observation when the evidence is still thin and the spending boundary permits it. Give every extension a reason and a next condition; otherwise, continue observing becomes a permanent nondecision.
Hand off discoveries without losing the source record
Amazon’s targeting guide, checked 2026-11-06, supports using automatic-campaign search term data to find manual keyword and product targets. It also says ASINs generating clicks or conversions can appear in the search term column. Read the value itself before deciding its destination.
| Discovery | Manual destination to evaluate | Acceptance check |
|---|---|---|
| Relevant shopper query | Keyword target with a deliberately selected match type | Query fits the advertised variation and has evidence appropriate to your objective |
| Relevant ASIN context | Individual product target | Product relationship and observed results justify a controlled follow-up |
| Clearly irrelevant query or context | Appropriate negative-targeting action | Exclusion scope matches the unwanted traffic without removing useful intent |
Keep the source group, observation dates, advertised ASIN, query or product ASIN, spend, outcomes, and destination together. For product discoveries, continue with the manual product-targeting guide; for account-level ownership and destination design, use the campaign-structure guide.
Moving a discovery is not the same as proving the manual destination works. Define its product, targeting choice, bid, spending decision, and review owner before deciding whether to exclude that traffic from the source. Check the destination is able to serve and has begun producing useful evidence; immediate source exclusion can remove discovery before the handoff is established.
You can also retain both paths deliberately. Amazon’s targeting guide describes automatic campaigns as a supplement to manual campaigns, checked 2026-11-06. Document what the automatic source is still learning and what the manual target now controls. The decision should follow the purpose of each path, not a blanket rule that every harvested term must disappear from automatic targeting.
Frequently Asked Questions
Does an automatic campaign replace keyword research?
It supplies observed discovery data rather than a complete market map. Use relevant queries from its search term report to inform research and manual targeting; an absent query does not establish that no demand exists.
Must each automatic group have its own campaign?
The documented option to set bids by targeting group lets you differentiate bids without treating every group as a separate campaign. Consider a structural split only when you need additional independent controls; use the campaign-structure guide above for that decision.