Being found: content and search for a technical company

A software company puts up a site, adds a services page, writes four posts about digital transformation, and waits. A year later the traffic is a few hundred visits a month, most of it from the team's own devices and from people looking for a job, and not one inquiry has arrived through it. The conclusion usually drawn is that content does not work for B2B software in this market. The more accurate conclusion is that nothing was published that anyone was searching for.

Inbound is the slowest channel a software company has. It is also the only one that keeps producing after you stop working on it, and the only one where the buyer arrives already half-convinced, having read your reasoning before speaking to you. It is worth building. It is not worth building on the assumption that it will produce a client this quarter.

Buyers search for their problem, not for your category

Almost nobody who becomes a client starts by typing "software company". They start with a sentence describing what is going wrong: stock counts that disagree between the shop and the warehouse, an accountant re-entering the same invoices into two systems, a government portal that has to be filed to and nobody knows how. They type that, in their own words, often as a full question, and often into an assistant rather than a search box.

Only the last row is the query most companies write for, and it is the smallest, most contested and least useful of the five. The decision is formed in the other four, usually weeks or months before a supplier is contacted, and usually by more than one person: the operations manager who has the problem, the finance person who holds the budget, the owner who signs. Each of them searches differently, and each of them needs a page.

One page per problem, written in the words the buyer uses

The unit of inbound is not the blog. It is a single page that answers a single question completely, at a stable address, and does not try to answer four other questions on the way. Ten such pages beat fifty posts about trends, because each one can be matched to a query and judged on whether it actually answered it.

A page that earns its position tends to contain:

• The question in the title, phrased the way it is asked rather than the way you would classify it.
• The answer in the first paragraph — not the history of the category, not a definition of ERP.
• Numbers you can defend in a call: ranges in EGP or USD, weeks of work, how many people the system touches. Give the range and say plainly what moves a project within it, which is what our note on what a custom system costs does at length.
• The cases where your honest answer is "do not build this". A page that never recommends against you is read as marketing and treated as such.
• One link to the next question the same reader will have.

A page that ranks is a page that answers the question and then stops.

The failure mode is circling: eight hundred words that restate the question, define the terms, list four "key considerations" and end by recommending that the reader consult an expert. That page can be written without knowing anything. Both the reader and the engine can tell.

The Arabic search market is thinner than the English one

Search a commercial or technical question in English and you compete with thousands of vendors, publishers and aggregators worldwide. Search the same question in Arabic and the results are frequently machine-translated summaries, directory listings, and forum threads from a decade ago. A page written in Arabic by someone who has actually delivered the system has very little standing between it and the top of the results.

That gap will close, but it has not closed yet, and closing it takes original writing rather than volume. A few practical points:

• Write the Arabic natively. A translated page answers an English speaker's version of the question and reads like a translation from the first line.
• Buyers mix scripts. Someone in Cairo or Riyadh will type "نظام ERP" and "شركة POS", so the Latin term belongs in the Arabic page at its first use, once.
• Modern Standard Arabic reaches Egypt and the Gulf equally. Dialect narrows the page to one country and reads as unserious in a purchase decision.
• Spelling varies — hamza forms, ta marbuta, transliterated product names. Cover the variants where they occur naturally in a sentence, never by listing them.

The technical work that matters, and the work that does not

The technical side of this is small and finite. What matters: one page per topic at a URL that does not change; a title and description that describe the page truthfully, because a description that oversells produces a click and an immediate exit; a sitemap; headings that mirror the question; text that renders on the server rather than assembling itself in the browser; and a page that loads quickly on a mobile connection, which is how most of this market reads. Structured data helps at the margin — Article, Organization, BreadcrumbList — and takes an afternoon to add correctly. FAQPage is worth adding for the machines that read it, but since 2023 it no longer earns a rich result for a company like yours, so do not count on it showing up in the listing.

Bilingual sites raise one question worth answering properly. Each language needs its own URL — /blog/x.html and /blog/ar/x.html — each a real page with its own title and description, paired with hreflang in both directions plus x-default. What a crawler can index is what exists at a URL when it arrives, so an article whose second language appears only after a click in the browser has, for search purposes, one language. A listing page may reasonably carry both — as long as every article it lists has its own URL in each language, and is linked in both. Do not redirect by IP either: a reader in Cairo may want the English page, and a crawler needs to be able to reach both.

What does not matter: keyword density, meta keywords, word-count targets, monthly retainers that buy links, and chasing global search volume for terms nobody in your market types. A page that answers a question fifty people a month ask in Egypt is worth more than a page aimed at a term with fifty thousand searches worldwide and no buyers among them.

Assistants read your pages too

A growing share of this traffic never reaches your site as a click. Someone asks an assistant what a custom accounting system costs in Egypt, and the assistant answers from pages it can read. What makes a page usable that way is close to what makes it usable for a person: the answer stated in a sentence near the top, headings that make claims rather than tease, figures written in text and not only inside an image, a visible date, a stated scope ("this is about companies of this size, in this country"), and content present in the HTML. What gets quoted is what is quotable and attributable.

Your own project record is the strongest thing you can publish

The most valuable inbound asset a software company owns is not an article. It is a page for each system it has delivered: what the system does, what constrained it, what was rejected and why, what it runs on, how long it took, and what changed for the client afterwards. Softwiro publishes its own record this way: each system it has built is described in the terms above, and most of them link to an address you can open and use. The same entry serves a search result and a sales conversation.

This works because a buyer typing "how to connect a POS to an accounting system" is describing something you have already built, and a record of building it answers them more directly than any general article can. It also cannot be copied. A competitor can rewrite your guide in a week; they cannot claim your delivery history. The mechanics of turning delivered work into something publishable are in turning delivered work into proof.

Quarters, not weeks

A page published today is typically indexed within days to a few weeks and settles into a position over the following two or three months. Ten to twenty pages have to exist before the shape of anything is visible. In practice that means a first two quarters with almost nothing, a third and fourth quarter with steady impressions and the first inquiries, and a second year in which inbound becomes a channel rather than an experiment. One good page a month, sustained, beats twelve in a burst and then silence.

Because leads arrive last, judge progress on what moves earlier:

1. Pages published are getting indexed, and the count of indexed pages tracks the count of published ones.
2. Impressions are appearing on problem queries, not just on your company name.
3. Average position on your target questions is moving from the fortieth range into the teens, even without clicks yet.
4. You appear consistently across one family of related queries rather than once on an unrelated one.
5. Inquiries and calls change character: someone quotes your page back to you, or arrives already knowing your price range, and the discovery call is shorter.

Sessions, bounce rate and follower counts tell you nothing about any of this. And none of it replaces the work of going out and finding clients in the meantime — inbound and outbound run on different clocks, and only one of them pays this quarter.

Where to start

Take the five questions you answer most often on the phone and write one honest page for each, in both languages, before writing anything else. If you would rather scope a system than a content plan, the contact form is the shorter route.