Personalized Landing Pages: What Works When You're Not Amazon

A personalized landing page shows different visitors different things, so more of them find what they came for. The idea is sound, but most advice about it is written for companies with data teams and millions of visitors. Part of our series on asking before you sell, this guide covers the three ways to personalize and what each one needs.

Why personalize at all

A landing page has one headline, one offer and one button for everyone. The people who arrive aren't one person: some are beginners and some experts, some want the cheap option and some the complete one, some arrived from an ad about one benefit and some from an article about another.

When the page can only say one thing, it's a compromise for most of them. Personalization means some of those visitors see something closer to what they need. In a controlled test, eBay found that adapting pages to each shopper raised purchases by 7.34% against the same page for everyone (Lo et al., RecSys 2021). That's eBay, with eBay's data and traffic, so read it as evidence that adapting works, not as a number to expect.

The real question for a smaller business is which kind of personalization it can actually run.

Personalization pays when your visitors want different things; the question is which method your data and traffic can support.

Way 1: personalize by where the visitor came from

The page changes based on the link the visitor clicked: the ad, the email, the partner site. Usually this uses the UTM parameters in the link, or simply a separate landing page for each campaign.

What it needs: campaigns that are actually different (one ad about saving time, another about saving money), and either a page builder that swaps text based on the URL or the patience to make a few versions of the page.

What it's good for: keeping the promise of the click. If the ad said "for freelancers", the page should open with freelancers. It's the simplest personalization and often the one with the clearest payoff.

Its limit: it knows the campaign, not the person. Two visitors from the same ad can still need different things.

Match the page to the promise of the click first; it's cheap and it fixes the most common mismatch.

Way 2: personalize by what you already know

The page changes based on data about the visitor: their location, the device, whether they're a returning customer, what they looked at before, what's in their account or your CRM. This is what most "website personalization tools" sell.

What it needs: data and volume. Behavioral personalization learns from many visits, so it needs steady traffic; account-based personalization needs a CRM with clean data; most tools need someone to set up the rules and a developer to install them. Many of these platforms are priced for larger companies.

What it's good for: returning visitors, logged-in customers, big catalogs where recommendations can learn from thousands of sessions.

Its limit: first-time visitors, which on a small site is most of them, look the same to these tools. There's nothing to personalize on yet. And inferring what someone wants from their clicks is a guess, sometimes a wrong one.

Data-driven personalization shines with returning visitors and big traffic, and has little to work with on a small site's first-time visitors.

Way 3: personalize by asking

Instead of guessing, ask. A few questions at the start ("What are you trying to do?", "How big is your team?", "Is this for you or a gift?") and the visitor's answers decide what they see next: a short result written for their situation and the offer that fits them.

What it needs: a clear list of your offers and who each is for, and a handful of good questions. No data history, no developer, no minimum traffic: it works on the first visit.

What it's good for: cold traffic, first-time visitors, sellers with several offers or several kinds of buyer. It also answers the visitor's doubt directly, because the result can say why this option fits what they just told you.

Its limit: some visitors don't want to answer questions, and every question is a chance to leave. It also can't cover a situation you didn't plan a question for. Keep it short, and keep the direct path to your page for people who already know what they want.

This is the approach Snacked takes. It isn't a personalization engine that rewrites your page: it puts a few questions in front of it, written from your own offers, and each result ends on a button to the offer that fits. It doesn't bring traffic, and a funnel holds up to eight offers.

Asking works from the first visit, with no data, which makes it the realistic option for most small sites.

What to personalize first

Whatever method you use, don't personalize everything. Three parts of the page do most of the work:

• The headline. It's the first thing read and the first place a visitor decides "this isn't for me". Naming their situation ("For agencies with 5 to 20 people") keeps them reading.
• The offer shown. Pointing a beginner at the starter option and a team at the team plan changes more than any wording.
• The proof. A testimonial from someone like the visitor (same kind of business, same problem) is far more convincing than the best testimonial from someone different.

Colors, images and layout rarely matter enough to be worth the work. Say you sell bookkeeping to both freelancers and small companies: showing each a headline, a package and a testimonial that match them is the whole job. Everything else can stay the same.

Personalize the headline, the offer and the proof; leave the rest of the page alone.

Which one fits you

They combine well. Match the page to the campaign; ask the visitors who still need to choose; let returning customers see what they bought before. Most small businesses should start with the first and the third.

Start with the methods your traffic can support today, and add data-driven personalization only when you have the data.

Recap

• Personalizing a landing page helps when visitors want different things; eBay's test shows adapting can lift purchases, at eBay's scale.
• Matching the page to the click is the cheapest start; data-driven tools need returning traffic, volume and setup.
• Asking the visitor works from the first visit with no data, as long as the questions are few and lead to one clear offer.

For the full case for asking first, read ask before you sell. And before you personalize anything, check that the page itself is clear with the free sales page grader.