SaaS Sprawl Data: Why Portfolios Add Apps Monthly

TL;DR: The SaaS sprawl data is blunt: the average company adds 9 new applications a month, enterprises add 21, and portfolios grow about 34% a year unmanaged (Zylo, 2026). Every app that lands without integration becomes another walled-off store of data. That monthly app accumulation is the mechanical reason enterprise data silos exist, and it is the mess any AI search or knowledge layer has to untangle later.

Nobody decides to build a hundred data silos. They arrive one purchase at a time. A team swipes a card for a new project tool, marketing signs up for its own analytics, finance adds a forecasting app, and none of it talks to anything else. Multiply that by every department and every month, and the silo problem is just sprawl that nobody stopped to connect.

How many apps does an enterprise actually add each month?

Here is the number that surprises people. The average company brings in 9 new unique applications every month, and at enterprises with 10,000 or more employees that figure jumps to 21 a month (Zylo, 2026). It sounds small until you annualize it: 103 new apps a year for the average company, and 257 for a large enterprise.

That growth lands on a stack that is already big. The average company runs about 305 applications, and the large-enterprise average is 696 (Zylo, 2026). Stack the monthly additions on top and you get roughly 34% annual portfolio growth for the average company and 37% for large enterprises (Zylo, 2026). A portfolio growing a third every year is not a one-time cleanup problem. It compounds.

SaaS sprawl is the term for this: the uncontrolled spread of software across a business because nearly everyone is now a software buyer. It is not malicious. People buy tools to get work done. The trouble starts when nobody connects or governs what gets added.

Why does sprawl happen so fast?

Procurement stopped being a gate. Today, lines of business and individual employees drive 87% of application purchases and 85% of software spending (Zylo, 2026). A manager with a credit card can stand up a new cloud tool in an afternoon, no IT ticket required.

A few forces keep the meter running:

• No purchasing policy, or a poorly communicated one. People buy a tool because they do not know an equivalent already exists three desks over.
• Cloud-everything. Almost every business function has a menu of SaaS options that deploy in minutes.
• AI tool rushes. AI apps are now among the most-expensed tools by employees, often adopted for a quick productivity win without an integration plan (Zylo, 2026).
• Nobody removes the old apps. New tools get added; redundant ones rarely get retired, so the count only climbs.

The spending backdrop reinforces it. Gartner forecasts worldwide software spending to grow 14.7% in 2026, staying above $1.4 trillion (Gartner, 2026). More budget flowing into software means more apps, and more apps mean more places for knowledge to hide.

How monthly app growth becomes data silos

This is the part that matters for anyone trying to find information later. Each app in a portfolio holds some slice of company data. The apps bought outside IT are usually poorly integrated, if they connect at all. So the data sits siloed across hundreds of separate tools (Zylo, 2026).

A data silo is information locked inside one tool, team, or account that the rest of the organization cannot easily reach. Sprawl manufactures silos on a schedule: every unconnected app added this month is a new silo by next month. The downstream cost is exactly what you would expect. Employees do not have the full picture they need to make a decision, or they burn time manually searching across tools for an answer that already exists (Zylo, 2026).

The fragmentation shows up in the operational numbers too. The average organization uses only 54% of its provisioned licenses, and just 21% of apps sit behind single sign-on (Zylo, 2026). Half-used tools and unmanaged access are the same story as silos: the portfolio grew faster than anyone’s ability to see across it.

There is a redundancy tax as well. The average company carries about 14 online-training apps, 10 project-management tools, and 10 team-collaboration apps doing overlapping jobs (Zylo, 2026). When sales lives in one project tool and marketing lives in another, there is no shared place to see who is doing what. The data is technically there. It is just scattered across tools that were never meant to reconcile.

A concrete example: Vantage Health

Vantage Health is a fictional 12,000-person provider network. At enterprise scale, it adds about 21 apps a month, so over two years its portfolio climbs by several hundred tools, most bought by individual teams.

A clinical-operations lead, Priya, needs one answer before a vendor renewal: which teams already use the scheduling tool up for renewal, and what did last year’s exception agreement say. The data exists. It is spread across an expense system, two project trackers, a contract folder, and a Slack thread nobody can find. Priya spends most of a day pinging three departments to assemble a picture that should have taken minutes.

Nothing about Priya’s problem is a knowledge-creation failure. Vantage Health created all of it. The failure is connection. This is the gap a unified semantic layer is built to close. SemanticOS connects fragmented tools into a knowledge graph so a single query can traverse the contract, the usage data, and the past decision at once, instead of leaving a person to stitch them together by hand. The sprawl still happened; the silos just stop being dead ends.

Key takeaways

• SaaS sprawl data shows the average company adds 9 apps a month and enterprises add 21, driving roughly 34% to 37% annual portfolio growth (Zylo, 2026).
• Sprawl is decentralized by default: lines of business now drive 87% of app purchases, mostly without IT integration (Zylo, 2026).
• Every unconnected app added is a new data silo, which is why employees lack context or waste time searching across hundreds of tools.
• The fix is connection, not another tool: a knowledge graph plus AI search makes the data already trapped in the portfolio findable across systems.