- Women's Jewelry
- PROSPECT
- NEW · every 48h
- 1/day
- https://graceparkerjewelry.com
Step 3.1 · Keyword harvest & classification
- What happens
- Collect candidates from four sources, deduplicate them against the shared volume cache, work the remaining misses through PinClicks, then classify: parent interests, generic keywords, topic clusters, seasonal type and publishing windows.
- Where
- Pinterest search bar and bubbles, the interest taxonomy in this dashboard, competitor annotations, and a PinClicks session for the volume lookups.
- Owner
- Media buyer
- Done when
- A classified keyword bank where every term has a cached volume, a type and a seasonal classification.
Candidates from the search bar
Type the seed keyword and note the suggestions including their order. That order is ranked by volume: a free proxy.
Seed "Jewelry" → 12 new, 0 dupes.
Candidates from bubbles
The coloured filters above the search results and the related searches at the bottom.
Bubbles + related → 16 new, 0 dupes.
Candidates from the Interest Sheet
Search the official Pinterest taxonomy in the dashboard. If an interest exists, there is demonstrable audience.
Interest-taxonomy picks → 3 new, 0 dupes.
Candidates from competitor annotations
From the PinInspector exports. Note: annotations are not keywords until they have volume.
Mined 91 annotation candidates → 91 new, 0 dupes.
Recognise a cloaked niche
In sensitive niches Pinterest hides the suggestions. Workaround: run the top 3 pins through PinClicks pin stats and read the annotations there.
CLOAKED — route to PinClicks pin-stats workaround.
Deduplicate against the cache
Volume is a property of the term, not of the client. Anything already looked up does not need looking up again.
Deduped 135 candidates: 18 hits (13%), 0 stale, 117 misses.
Generate work list
Cache misses only, prioritised. That keeps the manual session manageable.
Queued 117 PinClicks lookups (18 hits saved ~18 lookups).
PinClicks session
Work through the list. Fill in volume per term. If you find interesting related keywords along the way, add them: that is where the good finds are.
PinClicks: queue cleared; +33 related finds.
Assign parent interests
The taxonomy threads. At least five. These become your board names.
Parent interests: Assigned broad Pinterest taxonomy interests that directly match Grace Parker Jewelry's core assortment. Parent interests selected: Jewelry, Rings, Earrings, Necklaces, Bracelets and Gemstones. These will be used as the main taxonomy threads and board-name structure for the keyword clusters.
Determine generic keywords
The test: does this apply to every product? If not, it belongs in a cluster. Five to ten of them.
8 keywords passed the "applies to every product" test.
Form topic clusters
At least three clusters of 10 to 15 keywords. Cluster by product, moment, colour, size, material or season. Moment usually works best on Pinterest.
3 clusters formed (Necklaces, Earrings, Rings).
Seasonal classification
Evergreen, seasonal or micro-trend, read from the curve in Pinterest Trends. Micro-trends never become a board.
Seasonal classified: 10 (0 MICRO_TREND excluded from boards).
Calculate publishing windows
Six to ten weeks before the peak. For Black Friday and Christmas, earlier still. Publishing too late is the most common mistake.
Computed ramp-up windows for 2 seasonal terms (peak - 10 weeks, the start of the 6-10 window).
Align keywords with the client
Some terms perform well but clash with the positioning. Better to know now than afterwards.