Servisez is ours, which is the point. It is not a client. It was built and launched by the people behind We Lead, and that is exactly why every number on this page can be shown: the search figures are read straight from our own Search Console, August 25 to October 3, 2026, and the product figures from our own database on October 5. Nothing is rounded up and the small numbers are left in. A client case study can almost never do this. This is the day-40 chapter; the day-90 chapter is added to this same page in late November 2026.
We built a feature we were proud of. Research told us not to write about it.
The feature is the digital car file: every car gets a record, and the app warns the owner before the technical inspection, the insurance or the road tax expires. It is the thing that makes Servisez different from a listings site, and we expected it to carry the content plan.
Demand research said no. In Romania, the terms people would use to search for it index near zero. “ITP”, the technical inspection, is searched constantly. The road tax is searched roughly 1.8 times as often as the inspection, and had a visible content gap.
So the car file stopped being a topic and became a differentiator, sold inside articles about things people actually search for. The road tax became the first content cluster.
That single decision produced most of the chart further down this page. It is also the decision we are proudest of, because it meant writing less about what we liked and more about what was wanted.
What a Naples business should take from this
This is a Romanian marketplace and you are probably not one. Four things here are not about Romania, and they are the reason the page exists:
Research before writing, even when you are sure. Your best feature may have no audience searching for it.
Timing beats volume. Twelve good articles did less in a month than three pages published the week a law changed.
Measure before you spend. The funnel was instrumented on July 2, nearly two months before launch.
Keep thin pages out of the index. Pages with nothing to show do not help you, and Google notices.
The rest, from the marketplace cold-start problem to recruiting supply by hand and Romanian tax law, is interesting and does not transfer. The sections where that applies are marked.
The three decisions that produced the chart
We published before the law changed
On October 1, Romania changed how its road tax is priced, by emissions class. We published the price table, a calculator and an explainer before the change, then updated all three within a day of the official figures appearing in the Official Gazette.
Those three pages are now the most-seen on the site, with more than 2,000 impressions between them. The calculator brought six clicks from people with a real question in its first nine days. Six clicks is a small number. It is also six people who arrived with intent, on day nine of a brand-new domain.
The transferable version: watch the calendar your industry actually runs on, from regulation changes to season and local events, not just a keyword tool.
We built a trilogy, not a pile
Three “how to check it online” articles, for the inspection, the insurance and the road tax, each with the same comparison table and linked to each other, plus cost guides for inspections, diagnostics, oil changes and winter tyres.
Twelve articles by day 40. Each with a short answer at the top, questions as headings, a sources section and a named author. That structure is what Google and AI assistants can quote, and it is the same structure we use in content marketing for clients.
We kept thin pages out on purpose
The site has pages for 48 cities. Only cities with a verified shop are allowed into the index: two so far, Suceava and Botoșani. The other 46 carry a noindex tag until they have something real to show, so Google sees 39 pages, every one of them earning its place.
We also audited the whole site for orphan articles on September 17 and found none, because one article launched with a single inbound link in week one and we fixed that habit immediately.
The chart, and what 40 days actually looks like
The honest reading of that chart:
- The first month averaged under 40 impressions a day. The site existed and almost nobody saw it.
- September 25 is the content going live, ahead of the law change. The site jumped to 359.
- October 1 is the law changing. 462 impressions, roughly five times the 40-day daily average of 91. One day, not a new baseline, and the chart shows it falling back afterwards.
- Average position 12.5 across 223 queries, 40 days in, on a brand-new domain. That is page two. Most of what ranks is the blog, which is the point: the articles carry the site while the commercial pages earn their authority.
- Brand demand is real but small: 72 impressions and 55 clicks on the name itself. People who had heard of us went looking.
What the product numbers look like at day 40
Small, and we would rather show them than hide them. On October 5 the product had 3 verified repair shops (our two internal test accounts are not counted), 51 cars with a digital file, 13 requests posted by drivers and 14 offers sent back by shops. Billing is still switched off.
The job of the first 40 days was to prove that shops will verify, drivers will post and Google will show the site. It did that. Nothing more.
“Sure, but you were your own client”
It is the right objection and we will not dodge it.
Being our own client removed real constraints. There was no budget approval, no stakeholder who wanted the logo bigger, no legacy website to work around, no agency-client lag, and founder-level access to the database whenever a number was needed. Those advantages are genuine and they are not available on client work.
Here is what they did not change:
- Demand research does not require owning the company. It requires being willing to hear that your favourite feature has no audience.
- Publishing ahead of a calendar event is a scheduling decision, not a budget one.
- Instrumenting the funnel before launch costs an afternoon.
- Keeping thin pages out of the index costs nothing, and most sites still do not do it.
What owning the product bought us was not better marketing. It was permission to publish the evidence. That is the trade this page makes.
Before launch: the decisions that shaped everything after
- One brand, with rules, applied everywhere. One mark, one blue, one lowercase wordmark and a short set of rules: the mark never rotates, there is no mascot, no “temporary” versions. That system went into the app, 25 transactional email templates, the social covers, the field flyers and the onboarding kit before a single customer saw any of it.
- We killed a mascot after one day. A smiling character was designed during the early marketing work and withdrawn the next day. It was likeable, and it would have split the brand into two symbols. The discipline to delete something you like is most of what brand consistency is.
- Measurement before marketing. The conversion funnel (request posted, offer sent, offer accepted, booking completed, review left) was instrumented in product analytics on July 2, nearly two months before launch. Search Console was set up in early July.
- Free during launch, as an acquisition decision. Billing was built and tested, then deliberately switched off. Shops join without a sales conversation.
- Deliverability treated as marketing. Confirmation emails that land in junk are a product nobody can sign up for. The sending domain, its authentication records and the templates were set up and tested before launch, the same way we set up websites and their email for clients.
Launch: supply first, then press, then search
Specific to marketplaces. Most local businesses can skip to the next section.
A marketplace has a cold-start problem. Drivers will not post on a site with no shops, and shops will not join a site with no drivers. So the launch started on the supply side, in one city, with a field kit: a flyer for drivers, a message sequence for shop owners and a “your account in ten minutes” walkthrough. Verification, covering company registration, the tax registry and documents, doubled as a quality filter.
Press took one press kit. A regional daily covered the launch on September 4 and a national startup publication on September 9, both linking back. The angle that got picked up was not the feature list. It was the story: the app where the shops come to you.
Social ran on two kinds of content. Proof posts generated from the database: every newly verified shop became a story and a feed post. And a series of short sketches with two recurring characters, published on an account we already ran, with about six thousand followers, under one rule: the brand is never named in the dialogue. It is in the frame, on the phone, on the screen, and the joke carries the clip.
What happens next, and what we will report at day 90
Forty days is enough to see a slope, not a result. The day-90 chapter lands in late November 2026 and answers these, with the same honesty about small numbers:
- Does the curve hold once the road-tax news fades? The October 1 spike was tied to a law change. The real test is where daily impressions settle in November.
- Do the commercial pages start ranking, or only the blog? At day 40 the articles carry the site.
- Do the clicks turn into accounts, requests and bookings? The funnel has been measured since July 2.
- What happens when billing is switched on? The first paid plans, or their absence, go in the next chapter either way.
- Which cities earn their index entry? Two of 48 are in today. We will report how many follow.
Update log. October 12, 2026: day-40 chapter published, covering August 25 to October 3. Next entry: day 90, late November 2026.
Frequently asked questions
Want a launch measured like this one?
Everything above started with measurement rather than a projection, which is also how we start with clients. Request a free audit and we will show you where you stand and what we would fix first.