Search Intelligence for Subscription Media

Turn your archive into revenue
with search intelligence.

Bernal is a search engine you add to a subscription media site. It understands what readers mean and reports results by membership type, so editors and the business team can make decisions using first-hand reader data.

Set-up cost ¥0 · The quick check uses only the access logs you already have

Try it as a reader

Search in your own words.

For example, you could try

Choose a word, then press Search (or Enter) to see an illustration of the search below.

Measured at partner media

~75%

of search clicks go to articles published at least 30 days earlier

~56%

click-through rate on search results (about 3.7x the usual rate)

56→81%

of test searches put the expected article near the top (before → after our search improvements)

~25%

of readers who saw the paywall after a search went on to pay (about 1 in 4)

Measured at a partner media site. Click-through rate: September 2026. Payments: September–October 2026 combined. Archive share: as of October 2026. Expected article near the top: test in November 2026.

See more numbers
  • Paying members make up only about 1% of all users, yet they account for more than 40% of all searches.
  • Paying members use search about 57 times as often as non-members (around 7% of paying members use it).
  • Where the search box sits makes about a 130-fold difference to how often it is used (top pages against article pages). On top pages, around 17% of paying members (1 in 6) use search, compared with about 1% of non-members.
  • Article pages make up over 90% of page views, yet generate only around 10% of searches. We think placement explains the gap: the search box sits below the banner on top pages, but at the very bottom of the article on article pages.
  • Once the search box was added to top pages as well, everyday searches unrelated to paid articles (news, sports results and the like) became more common. This is a qualitative observation.

Measured at a partner media site (based on weekly unique users)

The Problem

A shared challenge for subscription media

Past articles that go unused

Paying members use search far more than non-members. But on article pages, the search box is easy to miss.

The rising cost of winning members

Winning new members costs more every year, while helping current members to stay is often left until later.

‘I couldn’t find it’ as a reason to cancel

Not finding the article you wanted is a reason people stop renewing.

How it works

The same word, side by side

Bernal searches the articles and delivers the ones that match the word. Measured at partner media, the click-through rate on search results is about 3.7 times the usual rate.

Gunners: A fan nickname, not the official name. On its own, this word does not say which team you mean.

A non-Bernal search

[Sample] Gunners fan scarf – official shop

[Sample] Arsenal v Chelsea: today’s match report

[Sample] Village gun club: annual dinner

Bernal

[Sample] Arsenal’s title push: how the Gunners rebuilt their midfield

[Sample] Inside the Gunners’ dressing room: a season in ten interviews

[Sample] The Gunners’ academy: where the next star comes from

Illustration only: these are not real search results. The examples are modelled on real searches on a news site’s paid archive. The search on the left looks across the whole site; Bernal looks inside the articles.

The Solution

Bernal turns search into business numbers, in two ways

Search that reads intent

It looks at what the reader wants, not just whether the words match. The aim is fewer searches that find nothing (zero-result searches).

Results by membership type

By matching with your membership data, we split searchers into paying members, free members and non-members, so you can see how search affects paywall conversion, payments and retention.

Pricing

Simple pricing: pay for what you use

Free
Trial (PoC) · 6 weeks

Try it free for six weeks and see the difference it makes. All it takes is one search-box tag and access to your articles.

¥35
per paying member, per month (after the trial; billed in yen)

You pay according to the number of paying members. Set-up cost: ¥0

Weighing up the cost: we compare the monthly fee with the lifetime value of readers who paid after searching, plus the cancellations avoided among members who use search. The final trial report shows this with your own numbers.
How to start

Start small, then grow once you see results

STEP 1

Quick check

We use only the access logs you already have, so nothing new is needed. We can estimate your own version of the ‘about 75%’ figure straight away.

STEP 2

Set up finer tracking (optional)

Once early results look promising, we set up more detailed tracking of page views, time on page and visit frequency.

STEP 3

Keep improving

We compare readers who use search with those who don’t, and widen the work step by step.

Why Bernal

It isn't about technology

What you measure, and who reads the result, is where the difference lies.

about 57x PayingFreeNon-member

We split results by membership type

We show results for paying members, free members and non-members separately. The people who decide to adopt Bernal are the team running the site. Only by splitting results this way did the 57-fold gap come to light.

A person reads every one

We read the zero-result searches ourselves, one by one. That is why we cannot take on many partners at once.

Set criteriaStartRead result

We set the success criteria first

Before we start, we agree in writing what counts as ‘success’ and what counts as ‘no effect’. We don’t move the goalposts once work has begun.

This method was filed for patent in 2017 and has since been granted in both Japan and the United States (Japanese Patent No. 6832584; US Patent No. 11,113,297; inventor: Kazuo Ikeda). We first applied it to advertising, without success. Only when we applied it to subscription metrics did a meaningful difference emerge.
Our view

Where we see this going

A search term is a small signal a reader leaves behind. Bernal reads those signals every day and points media toward their next move.

A reader’s search term

A small signal left behind by the reader.

Bernal reads it daily

Every day, without a break.

Shows the next move

Media can see what to do next.

What we are building

Bernal is building a layer that lives inside a publisher’s search logs and does two things: it puts buried archive content back in front of readers, matched to what they actually search for, and it measures how a reader’s search leads toward a subscription.

Internally, we call this layer the daemon: a process with no screen of its own that stays resident in the background and keeps working without being asked. We are building toward a system that does not wait for a quarterly report, but tells a team the moment a pattern changes.

It is designed to keep evolving. Like a living system, it changes every day to fit what its partners need, lighting the way one step ahead from behind the scenes.

Where it leads

A search box is one of the few places where readers tell a publisher, in their own words, what they want. It shows what a headline failed to explain, what a push notification did not answer, and what a reader could not find anywhere on the site. The more of that we learn to read, the more the daemon can give editorial, product and subscription teams a standing signal at the moment it matters, not at the end of the month.

The goal is not a better search box. It is a media organisation that always knows what its readers are asking for, because something is always listening.

Want to know more? Our founder will talk with you directly. Get in touch.

We are also looking for people to build it with us. See our careers page.

Start by showing us
your search logs.

Even a look at your current search logs lets us tell you about your zero-result searches and where there is room to improve.

Get in touch

For partner and client media

Information publishers need for disclosure pages covering external transmission, where Bernal is installed on their site, is set out in our Privacy Policy. If you need anything confirmed individually, please get in touch.