Marketing analytics 02 / SaaS / Content performance
Traffic That Converts Versus Traffic That Just Arrives
Content-topic segmentation, conversion analysis by intent, and a fairer way to measure time on site for a Singapore VPN app that scaled to 45,000 monthly visits.
- My role
- Project lead and planner
- Period
- Apr 2024 - Sep 2025
- Engagement
- Project delivered during my Golden Owl Solutions role
Business objective: Build organic visibility and direct high-intent content readers toward product downloads.
My ownership: I led planning and delivery: translated management objectives into project KPIs and milestones, allocated work and resources, trained and guided contributors, and owned progress against the agreed targets.
Consumer software and product intent
Different content themes attract people at different stages of product need. Topic-level download analysis informs SEO priorities and the route from an article to product interest.
Delivered during Golden Owl · Apr 2024–Sep 2025. Downloads are an intent signal, not verified activation or retention.
Turis VPN began SEO from scratch in April 2024 and grew to 45,000 monthly visits. That growth created a harder question: the highest-traffic articles were not necessarily the ones producing downloads. Marketing needed to know which content themes actually carried commercial intent, and needed a measurement of content quality that traffic volume and ranking position could not distort.
Traffic and conversion are close to uncorrelated across topics. VPN, Browsing and IP convert strongly on every channel — VPN at 2.15% organic, 2.58% direct and 3.86% referral — while Streaming and Cross-border Applications generate the largest session volumes at a fraction of that efficiency. Three themes (Cancel Subscription & Removal, Scam & Dark Web, Blockchain) attract traffic and convert essentially never. Separately, GA4 average engagement time was shown to be misleading: most topics are strongly right-skewed, so the mean alone hides the typical engagement pattern. Duration is an engagement measure, not a direct measure of content quality.
What the evidence says
What the topic-level data showed.
Commercial intent clusters in a small set of themes
VPN, Browsing and IP hold the highest conversion rates across organic, direct and referral traffic simultaneously, which is what separates genuine product intent from incidental interest.
Evidence: sessions and download conversion by topic and channel, GA4 1 Apr – 7 Jul 2025The biggest traffic themes are the weakest converters
Streaming contributes 26,874 organic sessions at 0.26% conversion; Cross-border Application contributes 21,098 at 0.49%. They function as top-of-funnel information content and need an explicit conversion path rather than more volume.
Evidence: topic-level session and conversion tableSome themes produced no meaningful recorded downloads
Cancel Subscription & Removal, Scam & Dark Web and Blockchain attract measurable traffic but produce no meaningful downloads on any channel, making them candidates for deprioritisation.
Evidence: zero-conversion topics across all three channelsAverage time on site was measuring the wrong thing
All eleven topics are right-skewed. Scam & Dark Web, Gaming, IP, Blockchain and Safety & Protection are extreme enough that outliers inflate the mean. Topics with compact boxes and mean close to median — VPN, Browsing, Blocking & Faking — have means closer to their typical session duration. Neither mean nor median alone establishes content quality.
Evidence: box-and-whisker distribution of session duration by topic
- Scale VPN, Browsing and IP content, which already convert across every channel, and build internal links from high-traffic articles into them.
- Give Streaming and Cross-border content an explicit conversion path — how-to framing, in-article CTA, and a download incentive — before writing more of it.
- Deprioritise Cancel Subscription & Removal, Scam & Dark Web and Blockchain unless they serve a brand purpose beyond acquisition.
- Report median and interquartile range alongside download conversion as engagement context, keeping the mean only where skew is mild.
- Treat the anonymous-proxy article as a template: it shows high intent, close product adjacency and is worth scaling deliberately.
- Every URL mapped to one of eleven content topics, so performance could be compared at theme level rather than page level.
- Sessions and download conversions split by organic, direct and referral to separate channel effect from content effect.
- Keyword ranking tracked by market (Singapore, Malaysia, global) across the top 5 through top 100 bands.
- Session duration analysed with box-and-whisker plots in Excel to expose median, dispersion and outliers instead of relying on the GA4 mean.
- Translated-URL performance reviewed separately to size the contribution of the localisation programme.
- Download events are the conversion proxy; downstream activation and retention are outside the dataset.
- Topic assignment is a manual mapping, so boundary pages can affect topic-level rates.
- The window is a single quarter and does not capture seasonality in VPN demand.