Summarize with AI
Pick a free trial, a Sign up button, when a stranger can reach a useful result on their own in under ten minutes. Pick Book a demo when they cannot, because the value depends on their own data, a setup step or a group decision. Make one of them the main button on each page and turn the other into a quiet link. Two buttons of equal weight hand the choice to the visitor, who knows the least about your product.
This matters for a small AI team because the main button decides where a new customer spends the first ten minutes. Either they are inside the product, learning it alone, or they are on a call, learning it from you. You can show your product in four minutes because you built it. Whether a stranger can is a different question, and most founders have never timed it.
Get a Demo and Start Free Trial, the same purple
Picture an AI catalog tool for Shopify stores, used here as an example. It writes product descriptions and image alt text for a whole collection at once. The hero reads Smarter product content, powered by AI. Under it sit two buttons side by side, the same width and the same purple: Get a Demo and Start Free Trial. A grey line below says No credit card required. On the right is a gradient card with a sparkle icon and the words 48 descriptions generated.
Neither button says who it is for. A store owner with 60 products sees Get a Demo and wonders if she is too small to matter. A merchandising lead at a brand with 9,000 products sees Start Free Trial and wonders if his catalog will break it. Both stall on the same pair of buttons. In this picture, plenty of them press neither, scroll for a moment and close the tab.
The demo is a four-minute setup on screen share
Now open the founder's calendar. It holds twelve Get a Demo calls a week, thirty minutes each, and he takes every one himself. The calls run the same way. He shares his screen, clicks Connect your Shopify store, picks a collection called Summer Linen with 48 products, and presses Generate descriptions. About three minutes later a review table fills in, one row per product, each with a green Approve button. That is where every call turns.
A lot of those calls end with the buyer asking whether they can run it on their own store tonight. He pastes the trial link into the chat. He has said in standup that he will hand demos to someone else. The next week he took the calls again, because nobody else runs the setup as smoothly.
Read the calendar plainly. The product reaches value in about four minutes. The demo is not a sales conversation. It is onboarding, done by the most expensive person in the company, one buyer at a time. He knows which collection to pick and where Approve sits. A stranger knows neither, and the homepage never showed her.
What twelve calls a week cost before the raise
Count the hours first. Twelve calls at thirty minutes is six hours. Add no-shows, reschedules and follow-up emails, and the demo path eats close to a full working day of the founder's week. Over a quarter that is about thirteen working days, taken from the product, from hiring and from the raise.
Then count what those hours buy. Product-led benchmarks, measured on SaaS companies, put opt-in free trials at around 18 percent free-to-paid and freemium at 2 to 5 percent. Those rates assume people actually use the trial. On the same benchmarks, activation runs at 20 to 40 percent for most product-led products, and a ten point gain in activation typically drives a 15 to 25 percent increase in free-to-paid conversion. A demo call activates one buyer, because the founder does the activating for them. It does nothing for the visitors who never booked.
The larger cost shows up in the numbers investors read. For AI startups the Series A bar is around 3.5 million in ARR, up from roughly one million three years earlier (Carta, Q1 2026). Across all sectors, the median gap from seed to Series A is around 616 days. Revenue that grows one founder-led call at a time cannot climb toward that bar, and it stalls the week he starts pitching.
The ten-minute stranger test, and the size of the plan
Most advice on free trial vs book a demo stops at time to value. That is half the rule. The other half is what a year of the plan is worth, because that decides whether a thirty-minute call can pay for itself. Use both answers together.
For time to value, ask someone who has never seen the product to start from your homepage with no help from you. Time them to the first useful output. Not to the dashboard, but to the thing they came for, like the first approved description. Under ten minutes means a stranger can get there alone. Then read your answer off this grid.
- Under ten minutes, small plan. Sign up is the only main button. Leave the demo off the homepage entirely. A call cannot pay for itself on a plan a store owner pays by card each month.
- Under ten minutes, large contract. Sign up stays the main button. Add a small Talk to sales link for buyers who need a security review or an invoice, plus an Enterprise column on the pricing page.
- Over ten minutes, large contract. Book a demo is the main button. The value depends on the buyer's own data or a team decision, and a call gets them there faster than they would alone. The second option is a recorded walkthrough, not a trial.
- Over ten minutes, small plan. This is not a button problem. The first run is too long for the price, so fix the product before the homepage.
That last case is common, and no headline can fix it. If the slow part is inside the product, the work is in the first run, which our guide to product-led growth design covers from the product side.
Why two equal buttons choose neither
The usual advice says to show both so every buyer finds a path. It sounds generous. In practice it moves your decision onto the visitor. She has to work out what kind of customer she is before she has learned what the product does, and a lot of people stop at that question.
Equal buttons also blur your own data. When someone presses Get a Demo, you cannot tell whether she wanted a conversation or simply did not trust the trial to work on her store. The calendar fills up and it looks like demand for demos. Much of it is demand for the product, routed through you.
So here is the claim. If a stranger can reach value alone in under ten minutes, a demo button of equal weight is losing you customers, not catching extra ones. The few who truly need a call will find a small link. The ones who wanted to try it tonight will not wait three days for a slot. The headline and the visual around that button are their own job, set out in our guide to SaaS hero section design. This post is only about which button leads.
Sign up first, talk to us second
For the catalog tool, the fix keeps both paths and gives them different weights. The main button now reads Write 20 descriptions free. It is the only filled button on the screen. Below it, in plain grey text, sits a link: Over 5,000 products? Talk to us. That link opens a short form with two fields, store URL and number of products, and the calendar appears only after it.
The gradient card goes too. In its place is a real before and after: a linen shirt with the supplier's one-line description on the left, and the generated description and alt text on the right. The visitor sees the output before choosing a path, so pressing the button feels like a small step instead of a bet.
The call also moves inside the product. Once the first collection is approved, a slim banner offers a twenty-minute call about bulk import and brand voice rules. The founder still talks to the big accounts. He just meets them after they have seen the result, and the call is about scale rather than setup. That call is worth his time.
We saw a version of this switch on the website we designed for Chariot, a speech AI lab. After our engagement, its Chariot-1 TTS model moved from private enterprise access to a public research preview. The action the site repeats is Try Chariot TTS, once in the top navigation and again under the opening line. When a product opens to strangers, the main button should say so in plain words.
Moving the button this week without losing the pipeline
You can make this change in five working days, and none of it needs a redesign.
- Monday: run the stranger test with two people outside the company. Write down the minutes each one took to reach a first useful output.
- Monday: go through last month's demo calls. Mark every one where you mostly did setup on screen share. If that is most of them, your demo is onboarding.
- Tuesday: work out what a year of your typical plan is worth and set it against an hour of your time. Use the grid above to pick one main button.
- Wednesday: rename the main button with the action and the object, like Write 20 descriptions free, not Start Free Trial. Turn the other button into a text link or move it to the pricing page.
- Thursday: add an offer of a call inside the product, after the first result, for buyers who want help at scale.
- Friday: put your calendar link in the day-two welcome email, so nobody who wants a call loses the option.
Then watch three numbers for two weeks: signups, the share of signups who reach a first useful output within a day, and demo requests. If demo requests fall while first outputs rise, the button was the problem. If signups rise and first outputs do not, the first run is the problem, and the homepage had been hiding it. If you would rather hand the page to a team, compare landing page design agencies for B2B AI SaaS companies.
Studio Maydit designs websites for AI founders that a stranger can read and act on in the first few seconds, and we carry on into the product screens those buttons lead to. For a self-serve company we look at the first run before the homepage, because a clearer button that sends people into a confusing first screen only moves the leak. A fixed-scope project takes three to four weeks, built in Framer, Webflow or custom code, and you can see how we work on website design for AI companies. If your calendar is full of demos that are really setup calls, book a 30-minute call with Studio Maydit.
