Google Trends Bulk Compare – 1,000 Keywords on One Scale

Google Trends compares only 5 keywords at a time, and every batch is scaled 0–100 on its own, so numbers from different batches can't be compared. This Actor breaks that limit: it runs as many 5-term batches as needed, links them through shared anchor terms, and rescales everything onto one common scale (the biggest term peaks at 100).

Use it for: keyword research across hundreds of terms · product / brand demand comparison · content planning · market sizing · seasonality and growth screening.

What you get per keyword

Field	Meaning
mean, peak, peakDate	Average and highest interest on the shared scale
growthPct	Last quarter of the period vs first quarter
timeline[]	Every point: raw (Google's own batch value) and value (stitched)
lowConfidence	true when Google's raw value was < 10 (rounding noise)
estimatedMonthlyVolume	Optional: give one term's monthly search volume and every term is estimated proportionally
bestBatchId, scaleFactor	Audit trail of the math

Views: Overview (one row per term) and Debug. Output mode perTermDate gives one row per term per date (pivot-ready CSV).

Real sample (run on 2 Nov 2026: 12 coffee/tea terms, US, past 12 months, 6 batches)

term	mean	peak	growth %	low confidence
coffee	75.60	100	-0.1	no
tea	35.55	45.0	-12.3	no
espresso	9.42	14.50	-17.8	no
cold brew	2.02	3.52	+39.6	no
dandelion coffee	0.016	0.043	-31.9	yes

"Dandelion coffee" is ~4,800× smaller than "coffee" and is still placed on the same scale. One Google batch can't show that, because it rounds the small term to 0.

{"term":"espresso","geo":"US","timeframe":"today 12-m","mean":9.4151,"peak":14.4974,"peakDate":"2026-04-12","growthPct":-17.8,"lowConfidence":false,"bestBatchId":"A2","scaleFactor":0.144974,"timeline":[{"date":"2025-11-02","raw":51,"value":7.3937}]}

How the stitching works

1. Pass 1: the anchor term plus 4 new terms per batch.
2. Pass 2 (automatic, when needed): terms are sorted by size and re-queried in overlapping "ladder" batches, so tiny terms are measured next to similar-sized terms instead of next to a giant.
3. Scaling: batches become a graph that shares terms. The strongest links (a maximum spanning tree) carry the scale factor mean_A / mean_B. Each term is taken from the batch where Google measured it most precisely.

Input

terms (list), anchor (optional; default is the first term), geo (e.g. US, DE, empty = worldwide), timeframe (today 12-m, today 5-y, 2024-01-01 2024-12-31…), category, property (web, YouTube, news, images, shopping), bridgeMode, calibration fields and proxy.

Tips & limits

• Terms far below the others (raw peak 0 even next to the smallest linkable term) come back as belowResolution: true with mean: null, never as 0. Add 1–3 terms of intermediate popularity to bridge them.
• Tested on 12, 50, 100 and 216 terms (US/GB/WORLD/DE; 1-m to 5-y): all terms linked; ratios match Google's own 5-term comparisons within 0–10%.
• Values are relative interest, not search counts. Use the calibration option for estimated volumes.
• Google may throttle large jobs. The Actor retries with new sessions; a residential proxy is recommended above ~100 terms.
• Terms flagged lowConfidence are directionally right but noisy.

Pricing

Pay per stitched keyword: $1.50 per 1,000 on Starter, $1.25 on Scale, $1.00 on Business and Enterprise ($2.00 on the Free plan), plus $0.01 per run. Platform usage and residential proxy are included. A 12-keyword run costs about $0.03, and keywords without data are never charged.

Not affiliated with Google. Uses publicly visible Google Trends data. You are responsible for complying with Google's terms in your use case.