Good questions, plain answers.

We build websites with the tracking set up properly. Our main service is the tracking and website ecosystem. We also offer four separate services: ads management, social media management, web development, automation and AI agents. Everything we set up is tested before we hand it over, so you can trust the numbers you see.

Separate agencies each report their own numbers, and the numbers often do not match. We set up the tracking first, so ads, social, website and automation are all measured the same way, in one place. You can compare them honestly and see which one earns its budget.

A person does. Every piece of content passes a human editor and nothing publishes unread. Every automated workflow keeps a log, has a route to a named person when something looks odd, and has an off switch that stops it in one action. We test accuracy against your real past cases before launch.

Thirty minutes. You tell us what you are building, what you are spending and what has not worked. We tell you what we would do first and whether it is worth doing at all. Two of the last engagements we scoped ended with a process change and no build.

Engagements are scoped and quoted rather than sold as packages, because the number depends on what the process mapping finds. The work runs in four steps and you can stop after the second, and the map of how your process actually runs is yours whether or not you build anything.

Five: SaaS, B2B, AI, e-commerce and fintech. The services do not change between them and the constraints do, so each industry page sets out what is different in that market and which service usually comes first. Nothing is a separate product.

It is our main service. We build your website and set up the tools that measure it: Tag Manager, GA4, Search Console, Bing Webmaster Tools, Clarity, and Meta ads tracking. We add consent, test every tag, and hand it over, so the numbers you see are numbers you can trust.

It sends the actions that matter, like a form sent or a sale, from your server to Meta, as well as from the visitor's browser. Browsers and privacy settings can block the Pixel, so the second route helps Meta see more of what your ads actually led to. We set up both and check they agree.

Yes, where the law asks for it. We set up a consent banner so analytics and advertising tools wait until the visitor agrees. Which rules apply to you depends on where your visitors are, so we check that with you during scoping.

Usually, and often that is the better choice. We can add the tracking and consent to the site you have, and replace only the parts that hold you back. A full rebuild is the right answer less often than agencies that sell rebuilds suggest.

We scope and quote each project, because the work depends on your site and the tools you need. We have not published fixed prices. On the first 30-minute call we tell you what we would do first and whether it is worth doing at all.

The work overlaps a lot with SEO done properly, which is why we do not sell it separately. What is new is making it clear who you are to a machine, shaping content so it can be quoted, and checking whether assistants name you at all. If an agency quotes a separate GEO retainer on top of SEO, ask which deliverables differ.

No, and nobody can. Answers vary by question, by session and by release, and no vendor controls what a model has learned. What we can influence is how clear you are, how well other sources back you up, and how quotable your content is. We track the result on a fixed list of questions, and you see the movement yourself.

You can buy one, and sometimes you should. They do different jobs on different clocks: ads give a readable signal in weeks, while search builds up over a year or two. Together they help each other, because search ad data is the fastest read on which topics are worth building pages for.

Ads give a readable signal in the first few weeks, because spend buys data straight away. Search and AI visibility move more slowly, so we judge them first on early signs: whether your pages are indexed and healthy, how clear your business is to machines, and whether assistants name you on a list of questions we agree with you up front.

Monthly, in one read across ads, search and AI answers, with every line tied to leads or sales. You see keyword positions, how healthy your pages are, the AI question list including the gaps, how each paid channel is paying back, and the funnel step that is currently losing the most people.

Only if it is written without grounding, which is the usual failure and the reason most AI content is easy to spot. We draft from your positioning, product notes and real customer language, write your voice down as clear rules, and have a person edit every piece. You approve a batch of examples before anything scales.

No. Every piece passes an editor before publication, and you can see the queue before it goes out. We say this on the about page too, so it is not a promise we can quietly drop when volume gets inconvenient.

By how much their audience overlaps with your actual buyers, using platform and third-party data. Follower count is close to irrelevant, and in business markets it often works against you. Every partnership has its own tracking and a measured result, and the ones that perform become long-term ambassadors.

By whether people come back. We look at how many join in the first week and do something, how many take part again, how many questions members answer before staff do, and how much revenue comes from referred accounts. A large silent community is a cost with good-looking numbers.

You do. The channels stay in your workspace under your logins, the content is yours on delivery, and the plan lives somewhere you can read it without asking us. Nothing is held in an agency account you would have to negotiate for on the way out.

How it is designed, not how many features it has. An AI-enabled product bolts an AI call onto a screen that was designed without one. An AI-native product treats the AI side as a core part of the plan: the right information flowing to it, answers fetched where the data already lives, and screens that adapt to what comes back.

Typed JavaScript from end to end by default: a front end rendered on the server (React or Astro, depending on how much of the page is an app), a content system, a Postgres database and managed hosting. Apps ship to iPhone, Android and the web from one codebase. We will use your existing setup if it is sound.

Usually, and usually that is the better call. Most projects start with a check of speed and structure on what exists, then replace the parts that are actually holding you back, step by step. A full rewrite is the right answer less often than agencies selling rewrites suggest.

You do, from the first commit, in your own repository and on your own hosting where you want it there. Handover notes are part of the launch, written while the reasons are still fresh, so your own developers or your next agency can pick it up without us.

Yes, product and screen design run inside the same team as the build, so the design is made with what is possible in mind. We decline brand identity work, such as logos and brand guidelines, which is a different discipline and one we would be pretending at.

A chatbot waits for someone to type a question and answers it. An AI agent works without being asked. It starts from a queue, a message from another system or a schedule, reads the case, uses your tools to do the work and passes anything unclear to a person. It leaves a log of every step and has an off switch.

Trigger-and-action tools run a fixed sequence and break on any input that does not match the path. An agentic workflow reads the case, chooses the next step and passes on what it cannot resolve. We use both: fixed steps stay fixed, because an AI is the wrong tool for rules that never vary.

Limited permissions, so a helper can only reach the tools its task needs. A route to a person on every branch. Scored tests on real past cases before launch, shadow mode on live traffic before it acts, and a rollout in stages. Plus a full record of every run and an off switch that stops everything in one action.

Either. Regulated and sensitive work usually runs in your cloud with your keys and your retention rules, which is the default for our fintech work. Everything else runs on ours unless you would rather own it. The design is the same in both cases.

Two weeks to a process map and a test set, four more to a workflow running in shadow mode, then a rollout in stages. Six to eight weeks to production on a first workflow is normal. Later ones are faster because the connections already exist.

The tests are re-run on a schedule and whenever the model changes, so a problem shows as a lower score on the next run, before your team feels it. Connections are defined and versioned, and failures retry under a clear policy before they are passed to a person.