Auto Targeting (Auto Campaign)
Auto targeting is the Sponsored Products mode where Amazon — not the advertiser — chooses which keywords and ASINs to bid on. The lowest-friction discovery engine in the format; underused as a harvest source.
Auto targeting is the Sponsored Products campaign mode where the advertiser does not specify keywords or product targets — Amazon's relevance algorithm decides which queries and PDPs the ad appears on, drawing on the listing's title, bullets, A+ content, category, and historical performance. The result is a campaign that runs without explicit targeting input, and a Search Term Report that surfaces whatever Amazon decided to test.
Auto campaigns are the lowest-friction discovery engine in Amazon PPC. They are also routinely the highest-yield harvest source for new SKUs — and the most underused report in most accounts.
The four sub-targets
Inside a single auto campaign, Amazon splits matching into four sub-targets, each of which can be bid independently:
| Sub-target | What it matches | Use |
|---|---|---|
| Close match | Queries highly relevant to the ASIN | Default workhorse; closest to exact-match intent |
| Loose match | Queries loosely related to the ASIN | Discovery layer; lowest CVR |
| Complements | PDPs of products complementary to yours | Product-targeting equivalent for cross-category |
| Substitutes | PDPs of products that substitute for yours | Direct competitor PDPs |
Bidding all four at the same number is the default — and the wrong default. Close match deserves a higher bid (CVR is comparable to harvested exacts); loose match deserves a lower bid (CVR is comparable to broad). Substitutes deserve a bid set against your visible competitive advantage; complements deserve a bid set against your typical cross-sell rate.
Auto vs. broad as a discovery engine
Auto and broad campaigns serve overlapping but not identical purposes:
| Aspect | Auto | Broad |
|---|---|---|
| Input required | None | Seed keywords |
| Source of matches | Amazon's relevance model on the listing | Your keyword expansion |
| Surfaces ASIN targets? | Yes (complements, substitutes) | No |
| Surfaces unknown unknowns? | Yes — finds queries you'd never brainstorm | No — bounded by your seeds |
| Operational lift | Lowest | Low |
Most professional accounts run both for every SKU. Auto catches what broad misses (especially product-target opportunities) and vice versa. The combined STRs from both are the harvest input.
Harvest workflow from auto
The harvest workflow on an auto campaign has one extra step compared to broad — the targeting type matters:
- Pull the STR weekly.
- For each high-converting search term, identify which sub-target (close/loose/complements/substitutes) caught it.
- Keyword search terms → promote to a manual exact-match campaign.
- ASIN search terms (B07XXXXX...) → promote to a manual product-targeting campaign.
- Negative-out the harvested term in the auto campaign (negative exact for keywords, negative ASIN for products).
Bidding the four sub-targets
A defensible default on a new auto campaign:
| Sub-target | Initial bid (relative to category suggestion) |
|---|---|
| Close match | 100% of suggestion |
| Loose match | 60% of suggestion |
| Complements | 70% of suggestion |
| Substitutes | 50% of suggestion (CVR will be low) |
Then iterate weekly based on per-sub-target CVR.
Common mistakes
- Not running auto at all. "I know my keywords" — but you don't know the ones you don't know. Auto surfaces those.
- Single bid across all four sub-targets. Each has a different CVR profile; each deserves a different bid.
- No harvest cycle on auto STR. Auto becomes a permanent spend leak instead of a discovery engine.
- Pausing auto after it "served its purpose." New search terms emerge continuously as the category evolves; auto should never be paused on an active SKU.
- Setting the auto campaign budget too low. A budget-capped auto campaign produces a thin, unreliable STR. The discovery layer needs enough budget to generate statistical signal.