Ask ten marketing teams where social media is heading and you will collect ten confident forecasts. Surveys capture intentions. Conference panels capture predictions. Purchase records capture something harder to argue with: what people did when their own budget was on the line.
Over an 18-month window, a social media marketing platform operating since 2020 examined more than one million anonymized orders placed by creators, brands, agencies and resellers across more than 25 countries. Every order names a network and a service, so each one is a small, dated vote about which platform matters and which signal is worth paying for.
Four findings emerged. Each carries a consequence for anyone setting marketing budgets, specifying reporting requirements, or building software that measures social performance.
How the dataset was built
Before the findings, the shape of the evidence, because a study worth citing should be checkable.
- Sample: more than one million anonymized orders placed between January 2025 and June 2026
- Coverage: more than 25 countries
- Exclusions: internal, test and reseller accounts removed, so the figures reflect real end demand
- Geography: derived from anonymized IP geolocation, resolved at processing time and then discarded, with raw addresses never stored
- Reporting basis: aggregate distribution shares and how they changed, never individual records The full dataset and charts are published openly and free to cite with attribution.
Finding one: two networks are absorbing the market
Across the full period, Instagram took 42 percent of all orders and TikTok 26 percent. Together, more than two thirds of everything purchased.
The direction matters more than the snapshot:
| Network | H1 2025 | H1 2026 |
| 34% | 45% | |
| TikTok | 26% | 31% |
| 6% | 6% | |
| Telegram | 10% | 6% |
| YouTube | 8% | 4% |
| Twitter/X | 12% | 4% |
In twelve months, short-form and visual platforms moved from 60 percent of orders to 76.
That is not a gentle slope. It is consolidation, and a second dataset points the same way from the opposite direction: the IAB projects United States social video ad spend at 31.9 billion dollars in 2026, growing faster than connected TV for the first time. One figure tracks advertising budgets, the other tracks organic growth purchases. They measure different things, which is exactly why their agreement is worth noticing.
Consequence for planners: budgets built on a platform mix from eighteen months ago are already describing a market that no longer exists. Long-tail networks are losing share faster than most annual review cycles can register.
Finding two: followers stopped being the product
This one reversed an assumption the analysts started with.
Followers, the metric that defined a decade of social media marketing, made up just 9 percent of orders.
- Views: 42 percent
- Likes: 32 percent
- Followers: 9 percent
- Shares and saves: 7 percent, and the fastest-growing category
- Comments: 6 percent
Between the two halves of the study, shares and saves nearly doubled their share, rising from 5 percent to 10. Comments also climbed.
The interpretation is straightforward. Every major platform now ranks content using interactions that indicate genuine interest: watch time, shares, saves, comments. Buyers noticed, and repriced accordingly. They stopped purchasing a larger number on a profile page and started purchasing the signals that determine whether content reaches anyone at all.
That shift is also visible in the purchasing patterns recorded by Boostero, where demand has moved beyond simple follower growth toward views, likes, shares, saves, and comments. The change offers a useful window into how buyers are adapting their social growth strategies as platform algorithms place greater weight on engagement signals.
Consequence for reporting: if a social dashboard still leads with follower growth, it is tracking the metric the market has already moved past. The people spending real money reorganized around engagement depth roughly a year before most reporting templates did.
Finding three: every network has its own fingerprint
Aggregate figures conceal the most actionable result in the study. Demand looks completely different on each network.
| Network | Leading service | Share | Notable second |
| Likes | 67% | Views 19% | |
| Twitter/X | Views | 57% | Likes 23% |
| YouTube | Likes | 34% | Comments 26% |
| Views | 46% | Likes 31% | |
| TikTok | Views | 42% | Likes 34% |
Two out of three Facebook orders are likes. Twitter/X is dominated by views. YouTube is the only network where conversation itself is the prize, with comments at roughly a quarter of its orders, a share nowhere else approaches.
Each platform rewards a different currency, and the buyers in this dataset behave accordingly.
Consequence for software: this is an argument for per-network reporting as a default rather than a drill down. A blended engagement score averages away precisely the differences that decide whether a campaign works. If your product rolls five networks into one number, it is hiding the signal its users need most.
Finding four: the demand map is moving
Orders came from more than 25 countries. The United States led at roughly 30 percent of geographically identified orders, with the United Kingdom near 9 percent and Mexico at 7.
The outlier was France. A year earlier it sat at roughly 3 percent of identified orders. By the first half of 2026 it had reached about 15 percent, making it the second-largest market in the study.
The data does not explain the shift, and inventing a tidy narrative for a single data point would be the wrong move. What it does establish is that a demand map can reorganize inside twelve months.
Consequence for market planning: geographic assumptions need an annual review, not a triennial one. In this dataset, a yearly check would have surfaced France long before any planning document did.
Three practices that survive contact with the data
- Plan per network, not per channel. “Social” is not one channel. The service mix on Facebook has almost nothing in common with the mix on YouTube, so a single social strategy is really five strategies sharing a label. Budgets, creative and success metrics should be set separately, and the fingerprint table above is a workable starting map.
- Measure what the ranking systems measure. Views, watch time, shares, saves, comments. These are the currencies platforms actually trade in. Follower counts still belong on a dashboard, just not at the top of one.
- Review geography every year. The cost is a morning of analysis. The occasional payoff is catching a market doubling before your competitors write it into their plans.
What this study cannot tell you
Stating the limits is part of making a dataset worth citing.
- It reflects demand on one platform. A large sample is not a census.
- It measures what buyers purchased, not what those purchases achieved. Ordering behavior and campaign outcomes are separate questions.
- Geographic figures cover only the orders where a country could be identified.
- Distribution shifts show the market moving. They do not establish causation.
Why purchase data deserves a place in the evidence mix
Most social media research is built on surveys, platform-reported metrics, or content samples. Each has a known weakness. Surveys record what people believe about their own behavior. Platform metrics are defined and adjusted by the platforms themselves. Content samples reveal what was published, not what anyone was willing to fund.
Transaction records sit in a different category. They are unglamorous, they carry no narrative, and they cannot flatter anyone. That is precisely their value. A million dated purchase decisions, each attached to a specific network and a specific signal, describe a market with a directness that no self-reported figure achieves.
The platform behind this dataset, Boostero, serves creators, brands, agencies and resellers across 23 or more networks in more than 125 countries, and has delivered over 11.9 million orders for more than 209,000 registered accounts since 2020. The findings above are published in full, with charts and methodology, for anyone who wants to check the working.
Related Categories