Case: rewrite the "12% a year" landing page

Before any claim about returns goes public, run it through four checks.

1. Type: is the number nominal or real, yearly or total, an average or a single path? "12% a year" is nominal and yearly.
2. Source: where does it come from, and is that source described honestly? Our Tessaro history is synthetic and must be labelled so.
3. Certainty words: "safe", "guaranteed", "risk-free" are never true of stocks, and in many places they are also a regulatory problem.
4. Checkable facts: user counts, fees and features are fine when they are true today.

A claim can fail one check and pass another. Keep what's true, fix what's misleading, and drop what can't be made honest.

Misleading investment copy usually breaks one of three rules.

Units and horizon. A yearly nominal rate with no time frame hides inflation and the swings between years. The fix is a real range over a stated horizon.

Certainty. Words like "safe" and "guaranteed" turn a probability into a promise. The fix is a plain risk line: values can fall, and you can get back less.

Sourcing. A number with an impressive source can still be the wrong number: nominal instead of real, a lucky period, or, as in line 3 of the draft, a synthetic history presented as the real world. The fix is to say exactly what the data is.

When you rewrite, fix all three. A page that drops "12%" but keeps "guaranteed" is still dishonest.

It is tempting to show only the middle and the top of the range. But a user who starts in a bad stretch will meet the bottom of the range first, and if the app never mentioned it, they lose trust and often sell.

Our simulation gives the honest bottom: after 10 years of 150 dol. a month, the 10th percentile is 16,877, which is 1,123 below the 18,000 paid in. After 25 years it is 57,544, well above the 45,000 paid in.

Showing the bad case does two jobs. It sets expectations, so a fall feels like part of the plan rather than a broken promise. And it shows why the horizon matters: the same bottom percentile goes from below the money paid in to well above it.

Present it calmly, with numbers, and next to the good case, not in a footnote.

People read landing pages in a pattern: headline, the biggest number, a button. Anything below the fold or in small grey text is effectively invisible.

If the range is in large type and the risk line is in a footnote, the page communicates the range only. The honest design keeps them together: the range, the horizon, and directly under them "not a guarantee; you can get back less than you put in", at a readable size.

The full legal disclaimer still belongs in the footer, and your company's compliance rules decide its exact text. But the one-line risk statement next to the numbers is a product decision, and it's the one users actually see.

Test it: cover everything except the hero and ask a colleague what the app promises. If they don't mention risk, the design failed.

The strongest honest claim an investing app can make is not about the market. It's about the user's own behavior: start early, put in a little every month, keep going through bad years. Those are true for everyone and fully in the user's control.

That makes a good headline: "Invest a little every month and let the years do the work." It promises a process, not a result.

Numbers move to the sub-line, where there is room to label them: "in our example", "at an assumed 6% after inflation", "simulated on synthetic history". One number is enough; three numbers in a sub-line read like a table.

A headline with a rate invites comparison shopping on the one thing nobody controls. A headline about the habit sells the thing the app actually helps with.

Every number on a money page answers an unspoken question: "says who?" The label is the answer.

In this course the market history is synthetic: generated to behave like real markets, with crashes, flat stretches and bubbles, but not a record of any real market. That makes it great for teaching and testing designs, and wrong to present as evidence of what a real fund did.

Good labels are short and sit right next to the number or chart: "synthetic history", "simulated paths", "not a forecast", "after inflation". They also protect the company: a regulator or a journalist reading the page sees exactly what was claimed.

When you use real past data in a real product, the same rule applies with different words: name the period and the source, and say that past results don't predict future ones.

A finished honest hero section combines everything from this module.

• Headline about the habit and time, with no rate.
• Range from the simulation, rounded, with the monthly amount, the horizon and "today's dollars": real, not nominal.
• Risk line right under the range, including the bad case in plain words.
• Checkable facts that are true today, like the beta user count.
• Footer disclaimer: synthetic data, not a forecast, after inflation, educational not advice, check local rules.

Then sell it internally. Stakeholders accept honest copy faster when you speak their language: trust lowers churn, users who expected falls don't panic-sell, and a page you can defend is cheaper than a page you must take down.