The Proof-of-Value Gate: Why B2B SaaS Demos Stopped Closing Deals

HubSpot added about 7,000 net new customers in Q2. It was aiming for 9,000 to 10,000.

That's a company doing $911.7 million of revenue in the quarter, up 20% year on year, with 306,446 customers on the books and probably the best inbound funnel in software. And it missed its customer adds by roughly a third, then cut its own full year guidance to 5,000 to 6,000 a quarter. The stock dropped nearly 5% on the day, 5 August 2026.

The reason is the interesting bit, and Yamini Rangan said it out loud on the call.

"Customers want proof of value before they buy."

The Two Reasons

Management gave two, and they're doing different work so it's worth separating them.

The first one HubSpot did to itself, deliberately. It moved AI agents onto trials in April and shifted to outcome based pricing with price cuts on some products. That lengthens the sales cycle by design, because a trial is a period where the buyer isn't paying you. You can't run a trial-led motion and keep last year's velocity. They knew that, they did it anyway, and I think they're right.

The second one they didn't choose. Budgets tightened suddenly, and mid-sized deals that used to close at one level are now going up for C-suite or board approval. That's not a HubSpot problem. That's the market.

Put the two together and you get the actual headline, which is not "HubSpot had a soft quarter". It's that the largest, most efficient buying journey in B2B software now requires the buyer to see a number before they'll sign, and to take that number to somebody more senior than they used to have to.

The £110,000 Lead Generator

One of my clients hired a physical BDR person last year. Spent £110,000 on him over the year. Got zero leads.

Zero. Not "below target". Zero.

Now, nobody sets out to do that. What happened is that a decision got made on a promise. A CV, an interview, a good conversation about pipeline, and a reasonable assumption that a full-time salesperson would generate some sales. There was no mechanism in that decision that produced evidence before the money went out of the door.

That's the decision buyers have stopped making. Not because they got smarter. Because enough of them have spent their £110,000 and they're not doing it twice.

And if you're selling anything AI-shaped right now, you are asking for that exact leap of faith, from people who have just been burned by it. That's your actual competitive environment. It isn't the other vendors on the shortlist.

The Vanity Version

Here's where I lose patience with most GTM reporting, and it's the same complaint I've had for years.

We booked forty demos this month. Great. I can book a demo. Anyone can book a demo, it's a calendar invite.

The questions I'd want answered instead: how many pilots actually ran? How many of those produced a number the buyer could take to a board? How many of those numbers were good? And of the ones that were good, how many closed?

Because that last funnel is the real one now, and I'd bet most founders reading this can't populate it. Not because they're lazy, but because they've never had to measure it. The demo used to be the qualifying event. It isn't any more.

Who Signs Now

The board approval bit gets skipped over and it's the most expensive change of the two.

If your deal used to close with a Head of Marketing and now needs a board sponsor, three things just changed. Who you're selling to. What they need from you, which is no longer a feature comparison but something they can defend to people who don't know your category. And how long it takes, because board cycles are monthly at best.

The mistake I see is founders treating this as a slower version of the same sale. It isn't. Your champion has gone from being the buyer to being your internal advocate in a meeting you're not in. That means the thing you hand them is different. It's not a proposal. It's the case they're going to have to make on your behalf, written so that a CFO who has never heard of you can follow it.

Most vendors are still handing over a pricing PDF and hoping.

What A Pilot Actually Has To Do

There's no clean answer here, so let me not pretend there is. Pilots are expensive, they eat delivery capacity, and some of them will fail in front of the buyer, which is a genuinely worse outcome than never having run one. That's the trade and you have to accept it.

But if you're going to run them, three things have to be true or you're just doing free work.

The number has to be agreed before you start. Not "let's see what happens". A specific metric, a baseline, and a threshold that counts as a pass. If the buyer won't define success, they aren't buying, they're browsing, and you've just given away six weeks.

It has to be short enough to survive the budget cycle. Six weeks, not six months. A pilot that outlasts the quarter it was approved in dies in the reshuffle.

And it has to produce an artefact, not a feeling. A number, on a page, with the baseline next to it and the date it was measured. That's the thing your champion carries into the board meeting. Nobody has ever walked into a board meeting and successfully argued "the team really liked it".

The Other Way To Read This

I want to be fair to the other side of this, because there's a version where I'm overreading one quarter of one company.

HubSpot's agent adoption went the other way entirely. Data Agent customers up 80% quarter on quarter to 16,000, Prospecting Agent at 17,000, monthly agentic actions tripled since the start of the year, Customer Agent resolving 72% of what it touches. Revenue still grew 20%. The board authorised another billion dollars of buybacks. This is not a company in trouble.

So you could reasonably say: adoption is fine, they just repriced and took a timing hit, and it'll normalise. Maybe. I'd still take the Rangan quote seriously, because a CEO does not volunteer "customers want proof of value before they buy" on an earnings call unless it's been said to them repeatedly by people holding budget.

The lesson is smaller than the headline and more annoying to act on. The demo is no longer the close. The measured pilot is.