Keyword generation runs at two speeds. The fast pass expands your seeds once and hands you hundreds of phrases in a couple of minutes. The deep dig keeps expanding its own results, and an afternoon later you own the whole topic map.

Our search form names the speeds Silver and Gold, and the metals fit: silver is the everyday workhorse, gold is the expensive dig you save for the right occasions. Here is when each one wins.

What the fast pass actually does

A fast pass takes each seed and runs one round of autocomplete expansion: the alphabet sweep, the question prefixes, the platform variants. One level out, no recursion.

That single level is deceptively strong. It carries the highest-signal phrases, the completions people type most, and it is fast enough to run on a whim, which means you actually run it.

What a deep dig adds

Depth means feeding results back in. Every phrase the first round surfaces becomes a seed for a second round, and the tree fans out.

sourdough sourdough starter sourdough recipe sourdough discard starter not rising starter smells like acetone starter feeding schedule recipe without dutch oven recipe for beginners discard crackers discard pancakes discard pizza dough the seed fast pass stops here the deep dig lives here
Level two is where the specific, winnable phrases hide. It is also 10x the volume to sift.

The payoff is coverage: the specific, low-competition phrasings that never surface from the original seed directly. The cost is volume, both in wait and in sifting, because depth multiplies noise just as faithfully as signal.

Gold, and the re-seeding technique

On the generator, the deep dig is the Gold option: it feeds every first-round phrase back in as a fresh seed automatically, so one seed fans through its whole second level while the counter keeps climbing long after a Silver run would have finished.

The manual version of the same move is re-seeding: run Silver, pick the ten strongest phrases, and run those as fresh seeds yourself. A third round on the best of round two goes deeper than most paid crawls.

The difference is aim. Gold digs everywhere, which wins when you are mapping a topic completely; hand re-seeding digs where you point, which wins when only some branches showed promise, the way the new-niche workflow expands only the branches with autocomplete evidence behind them.

When fast beats deep

SituationRunBecause
Weekly content ideationFastYou need 5 ideas, not 5,000 rows
Checking if a topic has demandFastThe first level answers it in minutes
Mid-writing, hunting subtopicsFastMomentum beats completeness
Planning a new site or hubDeepYou are buying the whole map once
Exhausting a money topicDeepThe winnable long-tail lives at level two
Quarterly list refreshDeepRare enough to justify the sift

The pattern: fast is a habit, deep is an event. Most weeks are silver weeks, and that is not a compromise, it is correct sizing.

The diminishing-returns curve

expansion depth level 1: dense signal the phrases everyone types level 2: the sweet spot specific, winnable, still real level 3+: the tail duplicates, junk, ghosts
Two levels capture most of the value. Past that you are sorting noise for sport.

This curve is why "always dig deepest" is bad advice. The marginal keyword at level three is usually a duplicate with a preposition moved, and your sifting time is worth more than it is.

Sizing the dig to the job

The honest workflow: silver by default, gold on purpose. Explore fast, and when a fast pass keeps hitting rich veins in one direction, that is the signal to schedule a deep dig there, with the trick stack and modifier matrix pointed at that one branch.

Depth you chose beats depth you defaulted into. Miners do not dynamite the whole mountain.

The one-line takeaway: fast passes are for finding veins, deep digs are for mining them. Run Silver as a habit, run Gold when a vein deserves the full dig, steer with hand re-seeding when only one branch does, and let the second level, not the tenth, be where you stop.