The sources
Organic social & dark social
Every platform defines its own metrics, counts them its own way, and grades itself generously. Meanwhile the most valuable sharing you get happens somewhere you cannot see at all.
Organic social is the least comparable data in your entire stack. Not because the platforms are dishonest, but because “reach” on LinkedIn and “reach” on Instagram are different measurements with the same name — and nobody has any incentive to standardise them.
The same words, different meanings #
| Metric | What varies between platforms | Why it matters |
|---|---|---|
| Reach | Some count unique accounts, some count unique devices. Definitions of a “viewed” post differ — one pixel on screen, versus dwell time. | Two platforms reporting 10,000 reach may have delivered wildly different amounts of actual attention. |
| Impressions | Whether repeat views by the same person count, and whether a scroll-past counts at all. | Impressions can exceed reach by 5× on one platform and 1.2× on another, for identical content. |
| Video view | Historically 3 seconds on some platforms, 2 on others, 30 seconds or completion elsewhere. Autoplay may or may not count. | “Views” is the single most misleading number in social reporting. Always check the threshold. |
| Engagement rate | Denominator may be reach, impressions, or followers. Numerator may or may not include saves, shares, profile taps and link clicks. | The same post can be reported at 2% or 12% depending purely on formula choice. Never compare across platforms. |
| Follower | Includes inactive and bot accounts to varying degrees. | A vanity metric with almost no relationship to reach on any algorithmic feed. |
Never compare a social metric across platforms. Compare a platform against itself over time, and compare posts within one platform against each other. Any cross-platform social benchmark you have been shown is comparing measurements that do not share a definition.
Dark social: the traffic you cannot see #
When someone copies your URL and pastes it into WhatsApp, Slack, Signal, iMessage, a private group, or an email, the resulting click usually arrives with no referrer. GA4 has nowhere to put it, so it lands in Direct.
This is not a small effect, and in South Africa it is unusually large. WhatsApp is the dominant messaging platform, and a great deal of genuine word-of-mouth referral moves through it and through private groups. All of it reports as direct traffic.
Your “direct” bucket is not people typing your URL from memory. It is a mixture of dark social sharing, untagged campaigns, app-to-web clicks, redirect chains that dropped the referrer, and a small residue of genuine direct navigation. Treating it as brand strength is wishful thinking; treating it as noise wastes a real signal.
How to get some visibility #
- Use tagged share links. If your site has share buttons, append UTMs to what they copy. This converts a chunk of dark social into labelled traffic.
- Look at landing page depth. Direct traffic to your homepage is plausibly genuine. Direct traffic to
/blog/a-specific-article-title-nobody-would-typeis a share. Segmenting direct by landing page is the single most revealing five minutes you can spend on this. - Use short branded links for anything you post manually, so the click carries a tag even after copy-paste.
- Ask. A “how did you hear about us?” field on your enquiry form is unfashionable, imprecise, and frequently more accurate than your attribution model for exactly this category.
What organic social data is actually good for #
Use it for
Content testing within one platform. Understanding which formats and topics resonate with your audience. Audience demographics. Spotting a piece of content that outperformed, so you can put money behind it. Community and sentiment signals.
Do not use it for
Cross-platform comparison. Revenue attribution. Justifying budget on engagement rate alone. Any claim about business impact that does not route through your CRM or site analytics.
Connecting social to something that matters #
Organic social is one of the hardest channels to connect to revenue, and most attempts fail because they try to measure it the same way as paid. Three approaches that actually work:
- Tag every link you control. Bio links, post links, story links. Consistent UTMs turn organic social from an invisible channel into a measurable one — at least for the portion that clicks directly.
- Watch branded search as a proxy. Organic social frequently drives demand that converts later through branded search or direct. If your social effort steps up and branded search impressions rise in Search Console two to six weeks later, that is meaningful correlated evidence.
- Ask on the form. Self-reported attribution is unfashionable and imprecise, but it is the only method that can see a conference conversation, a podcast mention, or a recommendation in a private group. For channels that are structurally invisible to tracking, an imprecise measurement beats a confident zero.
Attribution models credit what they can observe. Organic social, word of mouth and community are among the least observable channels you have — so they receive the least credit, not because they contribute least but because they are hardest to see. When a dashboard says paid drives everything, part of what you are reading is a measurement artefact rather than a fact about your business.
Common questions #
Why does GA4 show so little social traffic compared to platform analytics?
Because platform analytics counts link clicks and GA4 counts sessions, and a great deal is lost between the two. In-app browsers are aggressive about privacy, many social clicks bounce before the tag fires, referrers are stripped on some platforms, and consent declines apply as usual. A gap of 50% or more between reported link clicks and GA4 sessions is normal for social.
Should I use a third-party social management tool for reporting?
They are genuinely useful for scheduling and for having one place to look. But they read the same platform APIs with the same definitions, so they cannot fix comparability — they just present incomparable numbers in one table, which sometimes makes the problem worse by implying the comparison is valid.
Is direct traffic growth a good sign?
It depends entirely on what is inside it. Growing direct traffic to deep content URLs generally indicates increased sharing, which is good. Growing direct traffic that appears the same week you launched a campaign usually indicates untagged links, which is a bug. Segment by landing page before you celebrate.