The Year Every Buyer Tries To Build Their Own Software

Once More Unto the Breach?

We’ve seen it before, and we’ll almost certainly see it again: the software “buy vs. build” conundrum. Back in the 2010s, the big idea was that “buying” meant a company didn’t need to have on-staff resources to deliver tech with speed, with lower upfront costs and predictable OpEx payout, while “building” was for getting really specific on features and functionality, and generally, having total control over the price tag.

H1 2026 is the first time in a long while that “we’ll build it ourselves” has started to feel like the default objection in software sales. To the point where “building” has more than doubled in a year — based on our analysis on 100 anonymized companies — and it’s still accelerating.

Let’s take a look at what’s changed to allow the DIY inflection point to really stick this time around.

Means, Economics, Permission

Data from the latest GTM motions tell a unique story: The advancements in AI development have brought a perfect storm of capabilities, financing, and executive buy-in on an already volatile marketplace. It’s a combination of means, economics, and permission.

Means. Building out tech has become a much lighter lift than in the traditional SaaS era. In a late-2025 survey of 817 builders, Retool showed that 35% had replaced at least one SaaS tool with a custom build, while 78% expected to build more in 2026. Why would they wait, when over half of non-engineers can now build solutions, including customer-facing ones, without involving engineering?

Economics. Why pay 5–7 figures on point solutions when a few hours of building can give you the dashboards and intelligence that you need?

Permission. Top-down AI mandates across orgs are setting the stage for even greater build capabilities. IBM’s 2026 CEO Study found that 76% of large orgs had promoted or hired a Chief AI Officer — up from 26% in 2025 — and 85% of CEOs said every functional leader now needed to become an AI expert in their own domains.

Pavilion’s flagship is back.

September 28 to October 1 — Join the Topline community, including Sam, Asad, and AJ, in New York City to celebrate the best annual GTM event of the year.

You’ll be surrounded by more than 1,000 of the greatest thought leaders and operators, including Geoffrey Moore (author of Crossing the Chasm), as well as Topline regulars like Amanda Kahlow (CEO, 1mind), Anis Bennaceur (CEO, Attention), and many more.

Grab your ticket before they sell out.

What the Sales Data Says

At Attention, we capture, analyze, and execute on sales conversations for hundreds of companies. We’ve watched this objection move in waves through the market in real time: AI-native buyers first in mid-2025; enterprises with real AI, Ops, and internal-facing engineering teams next; and then, a sharp mid-market jump in Q1 2026.

To size it, we took 100 anonymized vendor profiles across ACV brackets and 15 industries. Five major patterns stood out.

Objection rate. For most vendors, the objections have at least doubled in a year, from 2.0% of engaged deals in late 2025 to 5.5% in H1 2026 and 7.2% this quarter. Meanwhile, 82 of 100 vendors are still trending up.

Timing. The trend coincides almost perfectly with Claude and MCP making internal builds accessible to technical Ops teams. (Note: Though our data can’t prove that’s what caused the jump, it’s the clearest explanation we’ve seen.)

Deal size. Objections scale with deal size too, in both frequency and in damage. Vendors selling under $25K see this on about 4% of deals and win those at their normal rate (“we’ll build it” is the buyer’s bluff when there’s no engineer to build it). In the $25–100K bracket, it’s on 6% of deals and whenever the objection is present, it cuts the median win rate by two-thirds. At $100K+, it’s on nearly 9% of deals; the median vendor sees their chance of winning drop by 80% against their baseline, and about 60% of these losses die with “no decision made” (i.e., internal teams say “we could do this” and the evaluations stall). In a sense, then, the build option doesn’t only beat the vendors; it stalls the buyer.

Industry sector. Data and analytics vendors top the panel at 8% incidence; cybersecurity, sales tech, and dev tools sit close behind; proptech and logistics run at 3%. The resounding message here: the more technical your buyer, the more often they see themselves as your competitor.

CRM visibility. Almost nobody has “internal build” as a picklist value, so the losses get filed under “No Decision” or “Other.” The median vendor has nearly 10% of lost ACV tied to an objection that was a rounding error a year ago.

A note for RevOps teams: check your call transcripts, not the dropdown in your CRM!

The Buyer’s Perspective: When to Build vs. Buy

Before they became a customer, one buyer walked us through swapping a $70K vendor contract for a cheap notetaker and an internal agent platform. Initially, it looked like he’d saved $50K. But when we dug deeper, the numbers just didn’t add up.

That’s a common mistake behind most “buy vs. build” math: Buyers price the vendor against the cost of building v1, when they should be pricing it against the total cost of owning v1 through v10 — including the token bills that pile up through every iteration, the pipeline maintenance, a system that breaks more often than you’d like, and the not-insignificant opportunity cost of pulling your best people off the mission they’re there to win.

None of that means that building is the wrong call. For a lot of orgs, it’s increasingly the right one: If the workflow is specific to your company, strategically differentiating, cheap to maintain, and you have the internal talent to own it, you should probably build it in-house. Paying a SaaS company $100K a year for something your team can reproduce in a week is getting harder to justify by the day.

The calculation changes, though, the moment the software becomes infrastructure. That’s when these three factors become a real litmus test:

1. Low cost of failure. If the tech breaks for a day, or produces the wrong answer, the damage is negligible relative to the upside of owning it.

2. Clear ownership. Someone’s job (not just their initiative) is to maintain the tech after launch and through the inevitable employee turnover that can derail a lot of internal processes.

3. Economics that survive v10. You’ve priced not just the initial build but the maintenance, infrastructure, model/API costs, security, integrations, debugging, and the opportunity cost of the people maintaining it instead of running the core mission as a business.

Check all three boxes? Build it. Miss even one, and the “cheap” internal alternative could become the most expensive software you own.

The Vendor’s Perspective: Battling DIY with the Pyramid of Leverage

If you sell software, there are three places to fight the buy vs. build objection, and I stack them into what I call the “pyramid of leverage.” Product has the most leverage and moves slowest, so it’s the foundation. Demand Gen sits in the middle. Sales and Enablement has the least leverage but can move the fastest, which makes it the tip of the pyramid.

My advice is to work all three in parallel, across the different teams: Product shapes Demand Gen, and Demand Gen shapes the sales conversation.

Step One: Start with Sales and Enablement. The raw material is already sitting in your call recordings. At Attention, we pulled “Closed Won” deals where this objection came up and extracted what reps actually said.

Remember: It never serves you to argue that they can’t build it (they can, and you both know it…). Instead, reprice it for them (“When you build it internally, it’s going to break. Who’s going to fix it?”). And while you’re at it, take away one of their reasons to build in the first place: the biggest wins consistently bundle a Forward-Deployed Engineer into the license, so you build that last mile for buyers rather than leaving them to build it themselves.

Step Two: Revisit your Demand Gen. Stop wasting budget on chasing buyers who are only going to stall you. Instead, let the data write your targeting rules. Run three analyses on your panel: who’s been buying from you without the build vs. buy objection ever coming up; who’s still buying from you despite the objection; and who ended in “Closed Lost” under that same objection. (Firmographics, technographics, personas, and intent/awareness signals are the data points that matter most.)

Once you can cleanly separate the first two groups from the third, you have your targeting rule: go hardest after the buyers who look like your wins, not your losses.

Step Three: Product is where you have the greatest leverage. It’s also the slowest layer, and the only one that ends the fight instead of just buying you time.

Here’s the internal matrix we use to think through it. Two axes decide where a feature lands:

1. Does it create proprietary data that didn’t exist before?

2. Is it single-player or multiplayer, and does it get more valuable as more people use it? (Multiplayer also raises the complexity and the cost of failure, for anyone trying to build it in-house.)

The simplest quadrant — single-player, no proprietary data — is a commodity. Don’t ship it, as LLMs already do this out of the box today.

One step more complex — single-player, new data — is a Claude/Codex Skill. You’ve built something real but only for one user, and a tech-forward buyer can replicate it in days (or hours).

Next up — multiplayer, nothing created — is an echo chamber. It looks safe but isn’t. It stores no memory, compounds nothing, and eventually the more technical teams clone it or can easily rip it out.

Finally — multiplayer, new layer of context — is the moat, and it compounds. Every interaction adds proprietary data to an organizational context graph; more users lead to more context, more context drives more value, and the vendor keeps layering intelligence on top.

Feature vs. Moat — Choose Wisely

Of course, none of this settles for good. The pendulum swings back toward “buy” eventually — it always does. Still, the vendors left standing next time won’t necessarily be the ones with the sharpest pitch, but rather, the ones whose product turned multiplayer workflows into proprietary context that can’t be built by an internal team in a sprint.

Almost anyone can ship a feature in 2026. Very few can ship a moat — and that gap is where next year’s winners get decided.

P.S. For both Sales and Demand Gen analyses, I have several scripts I can share with you that will expedite a lot of this work. Feel free to reach out via LinkedIn and I’ll send them over to you. And, if you have interesting data on the topic, I’d be more than glad to discuss it with you!

Subscribe now

Anis Bennaceur is the co-founder and CEO of Attention.com, a Series-B startup focused on maximizing sales through AI automation. After teaching himself to code at 12, he began his career in investment banking and private equity before moving into early growth at Tinder. He later founded Mixer, where he encountered many of the challenges he now solves at Attention. Anis is based in NYC, where the company is headquartered.