Cloud growth without a cloud majority
A vendor's license ledger records rising cloud intake alongside an installed base still dominated by local deployments.
2026-08-24
Briefs
What a Lead Number Still Hides Once the Period Is Added
Oleksandr Koliesnikov has now dated his 2024 paid lead figures month by month.
What a Storewide Conversion Rate Cannot Tell You
A furniture store's blended conversion rate and one costly set's purchase rate answer different questions, leaving product-level performance unresolved.
What changed when one store removed cash on delivery
A Ukrainian store reported steady sales and fewer returns after requiring prepayment, but its carrier data stops at the parcel.
What the gap between a haircut price and the average bill can prove
Public ratings can reduce booking risk, but the records do not show how much of the average bill pays for that reassurance.
What a custom mattress really costs in Moldova
A shop ledger turns an awkward mattress size into a clear supplement and production window, with strict limits on what the figures prove.
Fifteen hundred readings, no ads
A Toronto numerology web app reports about 1,500 completed paid readings in six months to 28 July 2026 and no separate ad campaign.
Six implants, one arch
A Ternopil aesthetic dentistry studio reports 125 full-arch cases on six implants in one operating year and 604 professional hygiene visits.
What delay costs co-founders
A row-by-row test for co-founders: settle exit, authority, money, equity, workload, information, review triggers and related-party rules before optimism becomes dependency.
What doing nothing costs
An engineering office near Munich puts one German renovation at about €160,000, the KfW grant at almost €42,000, and the twenty-year operating saving at more than €320,000.
Who do people believe
Scientists top Edelman's 2026 global trust ranking at 76%, 27 points above government leaders, yet the sharper finding is proximity: people trust their own CEO 12 points more than CEOs as a category, and their own employer 14 points more than business as an institution.
Thought leadership, measured
B2B buyers say expert content earns more trust than a vendor's own marketing materials and moves what they are willing to pay, but the two most recent annual studies behind those numbers surveyed different populations, and the companies producing the content still cannot measure most of its effect.
The half-life of a web page
Half the hyperlinks the U.S. Supreme Court cites as legal authority no longer lead to the material they cite, a 2014 Harvard study found, and a decade-wide Pew Research sample shows why that is not a legal-citation quirk: roughly a fifth of a given year's web disappears within two years, and well over a third within ten.
Global Talent, counted
Britain's Global Talent visa approves 99% of applications once they reach the Home Office, but only 72% of people who applied for endorsement between 2020 and 2023 went on to hold a visa, and at wildly different rates depending on field and nationality.
Where the clicks go: position by position
Position one on Google still wins the most clicks, but the margin is shrinking fast: four independently collected datasets, from Sistrix, Ahrefs, Pew Research Center and Seer Interactive, show organic click-through falling 32-58% by rank whenever an AI Overview appears, with early-2026 data suggesting the ranked but uncited page loses the most.
How patients pick a doctor
Patient surveys from 2020 to 2026 agree that most Americans now check online reviews before booking a new doctor, with estimates running from 71% to 90% depending on question wording, but they disagree on trust: the share who treat a review like a friend's recommendation has fallen from an 84% peak in 2016 to under half today.
What one star is worth
Four independently designed studies, covering Yelp restaurants, hotel pricing across 11 cities and Amazon book sales, put a number on what an extra star is worth: 5 to 9 percent more restaurant revenue, up to 11.2 percent more hotel pricing power, with a consistently asymmetric effect in which a one-star loss costs more than a one-star gain earns.
Ukraine’s business ledger, wartime edition
A closures slowdown, not a founding boom, explains Ukraine's swing from a net loss of sole proprietors in early 2025 to a net gain of 39,000-plus a year later, while company formation still runs a third below 2021 and three front-line regions keep shrinking while the rest of the country grows.
AI at work: adoption by the numbers
Ask how many U.S. businesses use AI and the answer ranges from 18% to 78%, depending which federal survey answers first: the 60-point gap is not measurement error but a difference in how "adoption" is weighted, worded and reported, and this brief lays out exactly where each number comes from.
The invalid traffic tax
Four widely cited estimates of invalid ad traffic range from 0.5% to 35% of spend, not because the fraud problem is unclear, but because each source measures a different point in the supply chain: a brief on why the numbers disagree, and what the disagreement itself reveals about the true cost of invalid traffic.
What a click costs in 2026
The average Google search ad click ran .42 in the US over the year to March 2026, more than double a decade ago, but the gap between industries is far more useful than the average: legal clicks cost six times what arts and entertainment clicks cost, and conversion rate, not price, decides which ones pay off.
Who Do AI Assistants Actually Recommend?
We pulled the available citation data. The answer has little to do with how good you are, and a great deal to do with where, and in what form, you are written about.
Datum Review publishes industry analysis built on numbers. Every figure carries its source. Read our methodology, editorial standards and corrections policy.
Updated 2026-08-03.