Behind the numbers
In an era where information is abundant but trustworthy data is scarce, PartyTrends is committed to transparency, accountability and evidence-based research. Here is exactly how it works — from the machines that gather the data to the people who decide what it means.
A multi-layered research methodology
PartyTrends combines quantitative analysis, large-scale data harvesting, trend monitoring, market intelligence, historical analysis and predictive modelling. No single source determines our reports, forecasts or rankings — we combine many to build a more complete view than has ever existed for this industry.
The machines behind the research
Most trend reports are written from a handful of sources. Ours sit on top of an always-on data harvesting operation engineered to collect, store and protect data at industrial scale.
Google data & the Google API
We pull live and historical search demand directly through the Google API, and archive Google historic data that has since aged out of the public APIs — irreplaceable records of what South Africa was searching for years ago.
A 24/7 AWS server farm
Our scraping infrastructure runs on a multi-node server farm in AWS, harvesting around the clock, every day of the year. The workload is so relentless we routinely burn out hard drives keeping up with the volume.
Large-scale web scraping
Distributed scrapers collect millions and millions of data points from search engines, retailers, social platforms, public datasets and our network of sister sites — continuously, not in one-off snapshots.
Polls & sister-site signals
Live polls and aggregated signals from millions of nodes and partner sites turn real audience responses into structured, queryable intelligence that grounds every report.
Behavioural signal capture
We read consumer behaviour across the open web — search behaviour, installed Chrome applications, Bing and DuckDuckGo searches, and engagement across Facebook, Instagram, YouTube and TikTok.
Long-tail platform coverage
Beyond the giants we track FreeTube, Rumble, Vimeo, Nebula and Dailymotion — with particular focus on Rumble and YouTube, where niche trends and emerging audiences often surface first.
Permanent historical archive
Every signal is time-stamped and stored permanently, so we can reconstruct demand, pricing and behaviour years into the past and measure exactly how the market has shifted.
Big-data processing pipeline
Raw data is cleaned, de-duplicated, normalised and correlated at scale, so patterns invisible in small samples become measurable, repeatable signals.
Our data pipeline, step by step
Every statistic we publish travels the same disciplined path — from collection through to a finding a human is willing to put their name to.
Collect
Distributed scrapers, the Google API, polls and behavioural feeds gather data continuously from hundreds of sources across search, social, retail and public datasets.
Clean & normalise
Data is de-duplicated, validated, de-noised and standardised — volumes adjusted for population, prices adjusted for inflation, partial years projected using known seasonal patterns.
Store & time-stamp
Everything is archived permanently with a timestamp, building the historical depth that lets us measure change rather than just describe a single moment.
Correlate & model
Big-data and machine-learning systems cross-reference sources, detect anomalies and surface patterns, growth signals and relationships hidden in the noise.
Interpret
Human analysts review the output, sanity-check it against multiple sources and decide what it actually means. AI assists — it never has the final word.
Publish & cite
Findings become charts, reports and rankings — free to quote and embed, with a clear methodology behind every number.
Four layers of intelligence
No single discipline tells the whole story. We stack four together so each one validates and deepens the others.
Search intelligence
Before a customer books a venue, hires a supplier or chooses a theme, they search. We analyse search demand, growth and decline rates, seasonal behaviour, regional interest, long-tail keywords and consumer intent — often revealing trends months before they become visible in the real world.
Market & pricing intelligence
We monitor pricing from thousands of suppliers and retailers to understand how costs evolve over time — balloons, helium, décor, venues, entertainment, catering and equipment hire — inflation-adjusted to reveal real affordability. This forms the basis of our Party Economy programme.
Trend lifecycle analysis
Most trends emerge, grow, peak and decline. Our deep historical archive lets us place a trend precisely in that lifecycle — whether it is emerging, growing, mature or fading — so businesses act on momentum, not yesterday's hype.
Predictive analytics & AI
Machine-learning systems detect patterns, anomalies and growth signals hidden within large-scale datasets, and project partial data into reliable forecasts. Artificial intelligence supports the research process — but it does not replace human oversight, interpretation or editorial judgement.
Hundreds of signals. One objective.
Our rankings are not based on a single metric. They are the result of analysing hundreds of independent signals designed to measure excellence, influence, growth, innovation, trust and impact.
Search demand
Brand and category search volume, demand trends, seasonal and regional patterns.
Growth & momentum
Search, brand, category and geographic expansion measured over time.
Authority & trust
Brand authority, industry recognition, citations and digital authority.
Reputation
Review volume, sentiment, consistency and customer advocacy signals.
Innovation
New services, technology adoption, differentiation and trend creation.
Geographic reach
Service areas, provincial coverage, national reach and expansion activity.
How we keep it honest
Scale means nothing without integrity. These principles guide everything we publish.
Independence
Findings are not influenced by advertisers, sponsors or award applicants.
Objectivity
Research uses measurable, evidence-based indicators wherever possible.
Fairness
Businesses are evaluated using consistent, repeatable methodologies.
Accuracy
Reasonable efforts are made to verify information before publication.
Transparency
Methodologies and criteria are explained wherever appropriate.
Human oversight
AI assists the analysis, but people make the final editorial call.
Data first. Evidence first. Integrity first.