The first forty-eight hours, published

Publish the opening, or hide it inside a quarter

An insurer usually waits. Early numbers are small, the mix is strange, and a public figure becomes a number other people repeat without the denominator. Ada’s argument was the reverse. If the company only publishes the good quarter, it has taught itself to hide the launch. The first two days are the cleanest record it will ever have of what the product does to strangers.

The question the record had to answer was not “did the internet like us.” Most of the internet could not buy. The question was, among people in the one state, how many who asked for a price then paid, and whether that was anywhere near the bets the team had been willing to say out loud.

Everything the first two days could hold

Every visit kept its source, its state, an age band, whether a quote was started, and whether a payment cleared. The homepage action was counted apart from the quote, because a click into pricing is not a policy. Previous insurer was kept when the person used the switch: cancel the old policy, take the refund, and bind the new one without retyping the house.

The internal pool was part of the dataset. People at the company wrote a number for the first forty-eight hours before the site opened. Those slips were not edited after. A launch story that compares itself to a memory is not a measurement.

Who was allowed into the rate

Visits from outside the state stayed in the traffic count. They were removed from quote-to-payment. Putting them in the denominator would have made a curious crowd from other states look like a failed funnel. Unpaid quotes were not counted as policies. A quote is a price. A policy is money.

The product-board crowd was tagged, not deleted. It skewed the room. At the open, 82.64% of visitors were men and 17.36% were women. The board itself was 88.89% men and 11.11% women, which is enough to bend a launch. As that spike faded, the split moved to 74% men and 26% women. Women who stayed spent about 9.1% longer and went further into the pages. Dropping the board traffic to make the gender split look intended would have hidden the channel. Keeping it, and saying so, is the cleaning.

Who actually arrived

Age followed the shape a new insurer hopes for and then did not stop there. 43% were 25 to 34. 29% were 35 to 44. About 7% were 55 or older, which is more than a product aimed at a first apartment expects, and it is why the later rise in the house premium matters. The opening was not only renters.

The switch log showed where the paid policies came from. People were leaving large incumbents, the household names, not arriving uninsured. Most of the buyers already had a policy of this kind. The new product did not create the category for them. It gave them a way out of the old one that did not require a phone call. That is a different company from one that congratulates itself on educating a market.

One action, three policies

The homepage that opened had a single action into the quote. Behind it, three policies: renters, which is belongings and liability; apartment, which adds the unit; house, which is the building as well. The blueprint is that screen. The three prices from the first forty-eight hours sit on the grid. Get a quote carries the heat. The rest of the page was kept quiet on purpose.

Almost 30% of the people who saw the page took that action. The brand argument was about color. The measurement argument is about number of choices. One next step, in the only place the eye is trained to look, will be taken by people who are not sure yet. Several next steps will be taken by people who already are. The launch traffic was not sure. The rate says the page did not ask them to be.

The hot band is the action

The homepage heatmap from those two days is a band on the action and almost nothing elsewhere. That is the picture that has to sit next to the 30% click rate. Heat spread across a hero, a video, and three competing buttons would have meant the rate was a crowd clicking around. Heat on one control means the page was legible.

Write the bet down before the number exists

Ada wanted the miss in public if the miss was what the two days contained. Leah’s split was the state line: most visitors cannot buy, so they must not be in the conversion rate. The button argument was settled by a rule rather than by taste. If the written premium was not a multiple of the highest bet, the story to publish was the miss, not the button.

Four times the most optimistic slip

The launch was the test. Written premium against the pool was primary. Quote-to-payment inside the state was the hold. There was no second homepage to compare. There was a number, written earlier, that the live number had to beat honestly.

It did. 142 policies. $14,302 in gross premium written, four times the highest slip. Of the people in-state who reached a quote, 14.8% paid, and that rate did not collapse when the press traffic faded. In the following two weeks the average house policy rose from $691 to $1,120. The one action stayed, because the record said it was doing the job, and because the company had agreed to publish the record either way.

What the two days do not prove is a year. They prove that low-intent traffic will pay when the product can be bought in one step, that a crowd from a launch board is not the customer, and that a bet written down beforehand is the only scale that makes 142 policies mean something.