Updated 2026-08-02
Smart Complete, Organic, Amazon Recommended: the relevance filter nobody reads
Every keyword in a Cerebro export carries its origin story in three columns — Smart Complete, Organic, Amazon Recommended — and that origin is a free relevance filter: Smart Complete keywords tend to be on-topic for your seed, keywords found only through organic data are disproportionately off-topic noise, and Amazon Recommended keywords are precise long-tail. Most sellers never look at these columns. They are the fastest triage you can run on a 10,000-keyword list.
What do the three provenance columns mean?
When Cerebro builds a keyword list from a seed, each keyword got in through at least one door:
- Smart Complete — surfaced by Helium 10's Smart Complete engine: expansions and completions of your seed. By construction these stay close to the seed's meaning.
- Amazon Recommended — keywords coming from Amazon's own recommendation data. Precise, specific, long-tail phrasings.
- Organic — a numeric signal derived from organic ranking data. A keyword can rank organically for many accidental reasons: a product that happens to match your seed also sells in a completely different market, and its other keywords leak into your export.
The three are not exclusive — one keyword can carry several flags. What matters is the only-organic case: no Smart Complete, no Amazon Recommended, organic signal alone.
How strong is the effect in real data?
On a real one-word-seed export of 11,547 keywords:
| Provenance bucket | Share of corpus | Reading |
|---|---|---|
| Smart Complete | 60% | Likely on-topic — read these families first |
| Amazon Recommended (no SC) | 6% | Precise long-tail — few, read individually |
| Organic only | 34% | Likely noise from unrelated products — triage last, with suspicion |
A third of the export enters through the noisy door. If you read the list top-down sorted by volume, that third is mixed invisibly into everything — including generic monsters with six-figure volumes that have nothing to do with your seed.
How should you use the prior in practice?
- Order your reading, don't filter your data. Read Smart Complete families first — that is where the seed's actual market lives. Keep organic-only visible but flagged.
- Suspect, then verify. Before discarding an organic-only family with real demand, look at its page 1 — thirty seconds settles whether it belongs to your seed's world.
- Mind the AR shortlist. Six percent of the corpus, hand-readable in one sitting, and the most specific phrasings in the export. This is where precise sub-niches announce themselves.
Why "prior, not verdict"?
Provenance tells you where to look first, not what is true. Real opportunities do surface through the organic door, and Smart Complete happily expands into on-topic-but-hopeless territory. The prior orders your attention; only demand numbers (read at family level) and the actual page 1 decide. The complete sequence: You ran Cerebro. Now what?
Frequently asked questions
Should I delete organic-only keywords from my Cerebro export?
No — triage, don't delete. Organic-only is a suspicion, not a verdict: real niches do hide in that bucket. Read Smart Complete families first, then sweep organic-only for anything with unusual demand before discarding.
What does the Organic column mean in a keyword-seed Cerebro export?
It is a numeric signal tied to organic ranking data. For niche research its main value is provenance: a keyword that appears ONLY via organic data — with no Smart Complete and no Amazon Recommended flag — frequently comes from unrelated products and reads as noise.
Why are Amazon Recommended keywords so few?
In our reference corpus they are ~6% of keywords. They come from Amazon's own recommendation data and tend to be precise, long-tail phrasings — low volume individually, but unusually specific. Few, and worth reading one by one.