Should you add a chatbot to your website? A short framework.

I'll start with the unpopular take: most websites do not benefit from a chatbot. The industry pushes the opposite story because chatbot vendors sell chatbots, but the reality is more nuanced. Some sites get a real lift from one. Others see zero impact and waste time configuring it.

Before you spend money or hours on this, here are the three signals that tell you whether you're in the first group, plus the situations where you should walk away.

Signal 1: enough traffic to learn from

A chatbot is a learning machine before it's an answering machine. Every question typed into it is a data point about what visitors actually want to know. To make that data useful, you need volume.

Below 500 unique visitors a month, the signal is too thin. You'll get maybe 10 questions across 30 days, half of them noise, not enough to pull patterns from. The agent answers fine but you can't use the data to improve your site systematically.

Above 5000 visitors a month, the signal is rich. You'll see the same question phrased ten different ways, you'll spot which pages create confusion, you'll notice which products generate hesitation. This is where chatbots earn their keep beyond the immediate "answer at 11 PM" benefit.

The middle range (500 to 5000 visitors) is judgment territory. If your average order value is high (over $200), even a thin signal can pay off. If your traffic is concentrated on one or two key pages, you'll get useful patterns even with low volume. Otherwise, focus on traffic first, chatbot second.

Signal 2: questions that aren't answered above the fold

A chatbot solves a specific problem: visitors who hesitate because they need information that isn't immediately visible. If your pages already answer all the obvious questions, a chatbot has nothing to do.

The test: open your most-visited product or service page on a phone. Read the first screen (no scrolling). Can a visitor learn the price, the core benefit, what makes you different, and how to act? If yes, your friction isn't information-availability. If no, you have a content problem first.

Most sites fall in between. The price is clear, the benefit is clear, but a specific objection isn't addressed. "Will this fit my situation?" "How does it compare to competitor X?" "What happens if I want to cancel?" These are the questions a chatbot intercepts. If your team gets the same handful of questions every week in support emails, the chatbot has a job.

If your visitors mostly contact you to confirm details that ARE on the page (because they didn't read carefully), a chatbot won't help much. They're not absorbing what's there. Better content design solves this, not more channels.

Signal 3: a visitor pool with high "hesitation potential"

Some products are commodities (visitors decide in 30 seconds). Some are considered purchases (visitors take days to commit). The chatbot impact correlates strongly with how much hesitation exists.

Strong fit:

• B2B SaaS with annual plans
• E-commerce with average order over $100
• Services with custom quotes
• Education / courses
• Anything regulated (legal, medical, financial)

Weak fit:

• Impulse e-commerce (under $30, fashion, accessories)
• Pure content sites monetized by ads
• Sites where the customer journey ends elsewhere (e.g. you redirect to Amazon)

The pattern: if a typical visitor is making a decision that takes more than 60 seconds to think through, a chatbot is useful. If the decision is instinctive, the chatbot adds friction without adding much value.

When you should explicitly not bother

Three cases where I tell people to skip:

Your site has under 200 monthly visitors. Solve the traffic problem first. A chatbot on a dead site is decoration.

You don't have written content to feed it. A chatbot needs a knowledge base, even a small one. If you can't answer the question "what should this thing know?" with at least one document of FAQ-like content, you're not ready. Write the FAQ first, add the chatbot second.

You aren't going to look at the data. The biggest waste I see: a chatbot installed, set up once, then ignored for six months. Your site doesn't improve because you never read the questions visitors ask. If you don't have 30 minutes a week to skim the chat history, the long-term ROI collapses to just "answers questions at night", which is fine but not transformative.

The honest math

Take an e-commerce site with 5000 monthly visitors and a 1.5% conversion rate. That's 75 sales a month at, say, $75 average order value. Total monthly revenue: $5,625.

If a chatbot recovers 5 hesitating visitors a month who would have bounced (a conservative estimate), that's $375 in extra revenue. Almost any plan worth its salt costs less than $80 a month. The math works at this scale.

If you have 500 visitors a month and a 1.5% conversion rate, that's 7-8 sales worth $560 in revenue. Recovering one extra visitor a month is harder to attribute and worth $75. The math gets thin. Maybe still worth it, but borderline.

If you have 50 visitors a month, the math doesn't work no matter what.

Action this week

Skip the marketing comparison spreadsheets. Skip the "top 10 chatbots" listicles. Just do this:

1. Open Google Analytics for your site. Find your monthly unique visitors number.
2. Look at the top 5 pages by traffic. List the most common questions a visitor on each page might have but that aren't answered above the fold.
3. Count how many support emails or contact form submissions you received last month.

If you have over 1000 monthly visitors, more than 3 unanswered questions per page, and over 10 contact form submissions, you're in the camp where adding a chat will probably move the needle.

If two of those three are weak, work on the underlying issue first. The chatbot can wait.