What I think the data is telling us
Fifteen notes from sitting with the 17 September 2026 report. I have not filled in traffic stories that would need impressions, branded search or a CRM join I do not have.
01
The traffic growth I can see was bursty, not a compounding engine.
The numbers. Q1 2026 reached 9,555 sessions, then Q2 fell 66.8% to 3,176. July 2025 (2,512 sessions) reset 82.7% in August. January 2026 peaked at 5,657 then February fell 66.0% to 1,925.
How I am reading it. What stands out to me is that we can create reach, but I do not yet see a floor I would budget against. The peaks look purchased or anomalous; the troughs look closer to the Direct baseline.
What this means for us. I would not forecast next year from January. I want to understand the mix behind each spike before I treat any of those months as run-rate.
FactGA4Growth report
02
Paid Social is the quality problem I keep coming back to.
The numbers. Paid Social generated 4,773 sessions (21.8%) but engaged at 23.63% with three seconds of average engagement. Direct engaged at 65.23%; Organic Search at 54.00% with 42 seconds.
How I am reading it. This looks like targeting, placement, creative-to-landing-page fit, or measurement. I am not completely sure which of those is doing the most damage yet.
What this means for us. I would look at placement and landing pages before putting more budget into this mix. Scaling it as it stands looks like it would buy more of the three-second sessions.
FactGA4Growth report
03
Direct is carrying the engine right now.
The numbers. Direct generated 8,120 sessions (37.09%) and a 65.23% engagement rate. It led February, May, June, July, August and September 2026 MTD.
How I am reading it. Brand, referrals and untagged demand are doing more work than the campaign reports show. I am not sure how much of Direct is really untagged LinkedIn or ads.
What this means for us. The next thing I would check is UTMs and branded search, so I can tell how much of this floor is actually ours versus leakage.
FactGA4Growth report
04
Organic Search is the channel I would rather grow.
The numbers. Organic Search generated 2,461 sessions (11.24%), 54.00% engagement and 42 seconds average time. July 2026 was its highest month (253 sessions) with paid traffic largely gone.
How I am reading it. Search visitors stay longer and engage more than Paid Social. July looks promising, though one month is not enough for me to call it a new baseline.
What this means for us. I would put more of the content and technical SEO work here, rather than buying another January-shaped spike.
FactGA4Growth report
05
The market we care about is not the one engaging most strongly.
The numbers. Saudi Arabia is 18.2% of users (3,355) but only 27.37% engaged, with eight seconds average time. UAE is 2.57% of users (473) despite being a core commercial market.
How I am reading it. Volume into Saudi is not the same as Gulf buyers who actually stay. I am not sure yet whether that is the audience we bought, or the site they landed on.
What this means for us. I would look at placement and localisation before I spend more on GCC paid. UAE being this small also deserves a closer look.
FactGA4Growth report
06
Mobile is where most people arrive, and where the session is weaker.
The numbers. Mobile supplied 54.4% of users but only a 39.1% engagement rate; desktop supplied 43.8% and engaged at 65.2%, a 26.1-point gap. Android WebView is 36.1% of users and averages nine seconds.
How I am reading it. The majority device gets the weaker visit. Android WebView makes me think a lot of this is in-app paid. I am not sure yet how much is the page versus the traffic quality.
What this means for us. I would test mobile landings and the form first, before buying more in-app traffic.
CalculatedGA4Growth report
07
The pages that already hold people barely receive the campaigns.
The numbers. Pricing, Sales Squad and Case Studies average 47, 52 and 32 seconds respectively, but receive only 177, 189 and 230 views. The homepage alone receives 19,419 views and starts 74.3% of sessions.
How I am reading it. Attention sits on the mixed-intent front door. Proof and offer pages already work; they just do not get the traffic.
What this means for us. I would land paid and LinkedIn on those three URLs, not on the homepage.
FactGA4WebsiteGrowth report
08
One proof post did most of the LinkedIn click work I can measure.
The numbers. The 10 June Gulf banking case-study post generated 169 of 352 measured LinkedIn clicks (48.0%). The other 16 measured posts averaged 11.4 clicks. Median clicks were 12.
How I am reading it. Named proof with a destination beat the typical post. I only have clicks for Apr–Jun, so I would not over-read the rest of the year from this.
What this means for us. I would repeat that format onto crawlable case pages, and tag the destinations so I can see whether it actually reaches the site.
FactNotion LinkedIn HubGrowth report
09
Posting more on LinkedIn did not lift Organic Social on the site.
The numbers. Apr–Jun dated LinkedIn rows rose 1 to 13 while Organic Social sessions were 94, 73 and 74. Coefficients with n=3 are too thin for me to treat as a relationship I would act on (r = −0.66 vs Organic Social).
How I am reading it. These two things did not move together in a way that helps me. Three months is too thin for a coefficient I would trust. I cannot tell from this whether LinkedIn is leaking into Direct, or simply not sending people.
What this means for us. I would stop treating row count as a performance KPI until impressions and UTMs are filled in.
CorrelationGA4Notion LinkedIn HubGrowth report
10
The 13,250 key events are not the commercial story.
The numbers. 13,250 key events coexist with 28 form-submission users. The careers landing page has a 97.5% key-event session rate. form_submission and form_submit both exist. GA4 revenue is £0.
How I am reading it. The key-event set is mixing hiring traffic and generic actions with commercial ones. I would not use that total in a budget conversation.
What this means for us. I want one frozen definition of a lead, then HubSpot joined, before I talk about ROI.
FactGA4Growth report
11
Most people who start the form do not finish it.
The numbers. 313 form-start users versus 28 form-submission users (8.95% start-to-submit; 91.1% did not complete). The contact form requires six fields including phone, company and target market.
How I am reading it. I am not sure whether this is tracking (two form events) or friction (six required fields, weak mobile). Both are plausible. Either way, this is the steepest drop I can actually count.
What this means for us. I would fix the event names and shorten the form before I buy more traffic into it.
CalculatedGA4WebsiteGrowth report
12
The blog did not publish inside this window.
The numbers. The newest visible article is 21 May 2025. Zero original posts fall inside 1 Jul 2025 – 30 Sep 2026. The blog index has 207 views.
How I am reading it. Content supply stopped before the period I am looking at. Residual search from the archive may still sit inside Organic Search; I cannot split that out from here.
What this means for us. If I want search to grow, I need GCC-specific pages again, not a lastmod sweep of old posts.
FactBlogGA4Growth report
13
The case studies are written, and trapped in modals.
The numbers. Twelve documented case studies use javascript:void(0) interactions. The index received 230 views and 32 seconds average time: high intent, 0.85% of views.
How I am reading it. People who find the index stay. Campaigns still cannot deep-link a named story.
What this means for us. I would turn those twelve into crawlable pages with UTMs before I ask LinkedIn or paid to send more proof traffic.
FactWebsiteGA4Growth report
14
July 2026 is the month I would try to understand, not January.
The numbers. Organic Search reached 253 sessions, average time 42 seconds, and paid traffic was minimal. August volume rose with paid return and an Egypt homepage anomaly.
How I am reading it. This looks like a healthier mix. I would want another few months of similar conditions before I call it a baseline. August already mixed paid back in, plus the Egypt spike.
What this means for us. I would try to recreate the July source mix, not the January volume.
FactGA4Growth report
15
Almost everyone I can count is a first visit.
The numbers. New users (18,383) sit next to active users (18,435). Monthly returning-user counts ranged from 26 to 178; January’s 178 is the peak, then 46 to 112 through August.
How I am reading it. The audience is not compounding. Every month is mostly people we have to find again.
What this means for us. I would start nurture and retargeting from case readers and form abandoners, once those pages have real URLs.
FactGA4Growth report