Hotjar's founder has said content was the last channel the company got right, because a single post took him a month. Then it built an organic search operation of more than a thousand pieces, judged on paying customers rather than traffic. Hotjar's own account puts non-branded search traffic up 734 percent in three years and new paying customers from it up 1,398 percent. Here is the machine, the method it says it measures with, and the limits of a story told from one source.
The Content Machine is a column about startups and scaleups that run content as a system: what they make, how often, how it is reused, where it goes, who makes it, how it is measured and what it moves for the business. Hotjar is here for a reason the first four were not. It is the clearest case I have found of a company that was already successful without content, said so, and then built a content operation on purpose and reported the result. Every number in this piece is Hotjar's own, from one account it published on HubSpot's blog in 2023, and I will say where that leaves the evidence.
The premise
Hotjar sells behaviour analytics: heatmaps, session recordings and surveys that show a product team what visitors actually do on a page. It was founded in Malta in 2014 by David Darmanin and a small team, grew without outside capital, and was bought by Contentsquare in September 2021 on undisclosed terms, by which point it had more than 190 people in 33 countries, per Contentsquare's announcement. A January 2019 interview teaser on Hackernoon put the earlier run at zero to US$19 million in annual recurring revenue in four years; that is the interviewer's framing, not a filing.
The machine, in one paragraph. A content and SEO team, working with product, product marketing and the company's own user-experience designers, produces and maintains a library that Hotjar says now exceeds a thousand pieces, at a monthly output it says is up to ten times what it was three years earlier. Every piece is built around the product solving the problem the reader searched for. The library is clustered by stages of the user's journey rather than by keyword, the keyword research starts from customer research, and old pieces are treated as a source of growth to be updated rather than an archive. Hotjar measures the whole thing on traffic that converts to paying customers, using what it calls 40/20/40 attribution. Hotjar's own write-up is the primary source and the piece to read.
How it started
The honest origin is that content came last. In a talk recapped by SaaStock in August 2021, Darmanin said Hotjar had "never done content really well before" and that content was the final channel the company added, after paid acquisition, email and communities. His reason was practical: a blog post "took me a month", so the team avoided writing. The channels that built the company were different. Hotjar ran a nine-month beta with a public roadmap, collected about 60,000 email addresses and let 18,000 people in, per the same recap. Those are emails, not sign-ups, and I am keeping the distinction because the number gets rounded up elsewhere.
So the content operation was a second act, started by a company that already had revenue, a free tier and a known product. That matters for reading the figures that follow. A large base of product users and brand searches is exactly the condition in which a non-branded search programme can post very big percentages from a small start. Hotjar's account gives the percentages and not the base.
How the machine works
Hotjar's write-up lists five fundamentals. I have kept its order and put each in plain words.
The product is in every piece. Hotjar calls this product-led content. A piece about, say, finding where users drop off on a checkout page is written so that the product is the answer, with the product's own screens as the illustration. The write-up's claim is that this is what turns readers into trials: the reader arrives with a problem and leaves having seen the tool solve it. It also explains why the operation is judged on paying customers rather than visits. A piece that ranks but does not show the product is, by Hotjar's own measure, not working.
The library is organised around the user's journey, not the keyword list. Pieces are clustered by stage: what a reader needs when they are learning what behaviour analytics is, when they are comparing approaches, when they are setting up, when they are trying to act on data. Hotjar says it maps the clusters with its product and product marketing teams and gives each cluster its own user-experience design, so a cluster reads like a guide rather than a pile of posts that happen to share a topic.
Keyword research starts with customers. The write-up says the topics come from combining several research inputs, qualitative and quantitative, so the team can work out which subjects will land with its audience before writing. In practice that means surveys, session recordings and support conversations feed the editorial plan, which is a sensible thing for a company that sells surveys and session recordings to say, and also probably true.
Old content is where the wins are. Hotjar describes nurturing its historical library as a major source of results. With more than a thousand pieces, the update of a piece that already ranks is cheaper than a new piece that does not, and the write-up treats maintenance as a core activity rather than housekeeping.
The measure is paying customers, through a declared attribution model. This is the part I would ask every content team to copy, even though Hotjar leaves it half explained. On the numbers, the company says it uses "40/20/40 attribution to try and build a fairer reflection" of multi-touch conversion paths, adding that "no attribution is perfect, but this at least helps to spread credit more realistically." The write-up never says what the three numbers stand for. The usual reading of a 40/20/40 model is that the first touch and the last touch each get 40 percent of the credit and everything in between shares the remaining 20, but that is the industry convention, not Hotjar's definition, and I will not put words in its mouth. The company also says it filters results by demand generation versus demand capture, by theme and by persona, which is a more honest way to read a content dashboard than one total.
What the account does not cover: the size of the team, how many pieces ship a month in absolute terms, any newsletter, podcast or video operation, and how the pieces are distributed beyond search. Hotjar's machine, as it describes it, is a search machine.
What it moves for the business
Hotjar's claims, confirmed on the page. "It is how we grew non-branded organic search traffic by 734% in just 3 years (2022 versus 2019)." And the line that qualifies it for this column: "it is also how we were able to turn this traffic into 1,398% growth in new paying customers" over the same three years. Annual non-branded new visitors are "now in the multi-millions". Monthly output is "up to 10x higher than it was 3 years ago".
The limits. Every one of those figures is a percentage with no base, from one first-person account that names no author inside Hotjar, published on a partner's blog. Nobody outside the company has checked them. The three years in question run from before the Contentsquare acquisition to after it, and the account does not say what changed in the business over that time. And the growth in paying customers is attributed to the traffic by the company's own model, which it describes as fair rather than perfect. I take the figures seriously for two reasons. The company chose to measure itself on paying customers rather than visits, which is the harder number to inflate. And the founder's earlier admission that content was the channel Hotjar was worst at is not the kind of thing a company says before a vanity statistic.
What to steal
Put the product in the piece, then judge the piece on customers. If the reader cannot see your product solving the problem they searched for, the piece is reach. Measure every piece on trials and paying customers, and let traffic be a diagnostic.
Organise the library by what the reader is trying to do, not by keyword. Clusters by journey stage read as guides and get maintained as guides. Keyword lists get abandoned.
Budget maintenance as production. Hotjar's wins came from a library of a thousand pieces that it kept updating. Decide what share of each month goes to pieces that already exist.
Declare your attribution model, even an imperfect one. Hotjar names its model and admits its limits. A named model that spreads credit is better than last-click, and far better than no model at all.
Admit which channel you are bad at, then build it on purpose. The most useful sentence in this story is the founder saying content was the thing the company had never done well. The operation followed the admission.
