Integrated digital storefront. The request enters the systems, not a mailbox.
The website is there, the form works, the request arrives. And it arrives as an email in a shared mailbox, where someone has to read it, understand it and copy it somewhere. This piece builds the door, and connects it to what is behind it.
The request arrives already prepared. Which page it comes from, what the person who wrote was looking at, which product it refers to. Whoever replies starts further along.
Performance is a requirement, not an opinion. The thresholds are public: main content within 2.5 seconds, interaction response within 200 milliseconds, measured at the seventy-fifth percentile.
The indicator is useful requests, never visits. A much-visited page that produces no qualified requests is an elegant cost.
This is the page for a single piece of the system. The other pieces, and the criterion for choosing which to start from, are on the services page.
The problem is that the form produces one more email
A disconnected website moves work around instead of removing it. Whoever writes from the form generates a message that ends up in a mailbox with everything else, and from there the real work starts: working out who they are, what they want, whether they are already a customer, where it has to be recorded.
The second effect shows in the year-end numbers. Nobody knows how many requests came from the website, how many got a reply and how many turned into something, because the request's route breaks off inside a mailbox and leaves no trace anywhere.
A website that isn't connected to anything produces work. It looks like a channel, and in practice it is a collection form with a nice cover.
What changes in practice
Whoever replies starts further along. The request arrives saying which page it came from, what the person who wrote was looking at and which product or service it refers to, so the first reply is already relevant instead of being a request for clarification.
The second change is that nothing gets lost. Every request leaves a log entry with its arrival time and status, and a request still without a reply after a certain time becomes visible instead of slipping to the bottom of a list.
The third is that the website becomes measurable on a number that matters. Not visits, which go up and down for reasons that don't depend on you, but the requests that turned out to be useful, counted one by one.
The look decides trust before anyone reads
It has been measured, and the number is higher than people think. In the research on website credibility carried out at Stanford, 2,684 people explained in their own words why they did or didn't trust 51 real websites, and the answers were classified by hand.
The most cited category of all is the look. The study report puts it this way: «The most frequently mentioned category was design look and feel of the site, accounting for 46.1% of the comments», ahead of information structure and content.
Whoever doesn't know you decides whether to take you seriously by looking, and does so before reading a line. That is why care over the graphic design isn't a luxury on a page that receives new buyers, and it is also why this is not where to save.
Speed has thresholds, and they are public
The three measures Google uses to judge the experience of a page have declared numbers. According to the official documentation the main content must appear within 2.5 seconds, the response to an interaction within 200 milliseconds, and the visual shift of elements must stay below 0.1.
How it is measured matters as much as the threshold: «a good threshold to measure is the 75th percentile of page loads, segmented across mobile and desktop devices». That is, you look at the worst load for three quarters of visitors, not the average, and mobile and desktop are counted separately.
The cost of going over shows in behaviour. The data collected by Google on mobile sites measures that 53% of visits are abandoned when the page takes more than 3 seconds to load, that is, before the visitor even knows what you sell.
What the system does, step by step
The work splits into five points, and the fifth is the one almost nobody builds. Each step leaves a log entry, so at the end of the month you know what happened instead of trying to remember it.
| Step | What happens | What you get |
|---|---|---|
| The pagesone question per page | The structure is built on the questions that really reach you, with the answer in the first sentences of each page. |
Whoever arrives finds their answer without looking for it, and stays long enough to write. |
| Performancepublic thresholds | Images, fonts and code are kept within the thresholds, measured at the seventy-fifth percentile and split by device. |
Visits from mobile stop being lost before they see the page. |
| Entry pointswhere people write | The forms ask for the bare minimum, and carry with them the context of the page they start from. |
Fewer fields to fill in, and more people reaching the end of the form. |
| The connectioninto the systems | The request enters the internal route already classified, instead of generating a message in a shared mailbox. |
No retyping, and whoever replies sees straight away who wrote and what they were looking at. |
| The logwhat happened | Every request has an arrival time, a status and an outcome, and those stuck beyond the threshold are flagged. |
At the end of the month you know how many requests came in and how many turned into something. |
The connection is the part almost nobody builds
Almost all websites stop at the form, because the connection requires knowing what happens inside the business afterwards. It is analysis work before it is programming, and it is done by looking at how a request is handled today, step by step, including the steps nobody has ever written down.
The real connection takes three forms, depending on what already exists. Where there is an internal system that can be accessed from outside, the request is written directly; where it can't, the system prepares the row in the format that system imports; where there is nothing, you start from a shared list with the same fields.
The third form is the most common in small businesses, and it isn't a reduced version. A list with the right fields and a status log does ninety per cent of the work, and stays compatible with whatever comes later.
What goes out on its own and what waits for a person
Everything that repeats already approved information goes out on its own: the confirmation of receipt to whoever wrote, the classification of the request, the assignment to whoever handles it and the flagging of what has been stuck for too long.
Nothing that commits the business goes out on its own. A quote, a price, a confirmed availability and a delivery promise are prepared with the context gathered and wait for a person. The first reply to requests that arrive from several channels is a piece of its own, handled by Inbox AI. The reasoning about the human constraint is on the page about the principles we build with.
The same thing changes name with the trade
The mechanism is identical; the request isn't. It is worth looking at your own case, because that is where you see which information gets lost along the way today.
| Sector | The request that comes from the website | Where you see it |
|---|---|---|
| Food and agricultureand export | A foreign distributor asking about formats and terms, who today gets a generic reply because nobody knows which page they came from. |
|
| Restaurantsand bars | A booking for twelve people on a Saturday, which arrives as an email and is copied by hand into a notebook. |
The service of a dining room where the phone rings unanswered |
| Hospitalityaccommodation and events | A quote request for a date, where the number of people is always missing and it takes two emails to get it. |
What this piece doesn't do
It doesn't bring traffic on its own. A well-built website answers whoever arrives, and the work of getting it found is a different piece: presence on search engines and AI assistants. Building the first without the second produces a door on a street where nobody passes.
It doesn't write the content. The pages need texts that state your numbers, and those are produced with sales copy and materials, starting from the same base of facts.
It doesn't replace the internal systems and doesn't impose new ones. It connects to what you have, and where there is nothing it starts from the simplest form that works, chosen to be replaceable without redoing the website.
Questions and answers
What difference does a connected website make compared with a normal one?
The difference lies after the form, and it is the only one that shows in the numbers. On a disconnected website the request becomes an email in a shared mailbox, where someone has to read it, understand it, sort it and copy it somewhere.
On a connected website the same request enters the route with its context already attached: which page it came from, what the person was looking at, which product it refers to, and with a log entry saying when it arrived. Whoever replies starts further along, and no request gets lost at the bottom of a list.
How fast does a website need to be?
The thresholds are public and are measured at the seventy-fifth percentile of page loads, separating mobile and desktop. Google considers good a main content display time within 2.5 seconds, an interaction response within 200 milliseconds and a visual shift of elements below 0.1.
The reason to take them seriously is that Google's data on mobile sites measures 53% of visits abandoned when the page takes more than 3 seconds to load.
Does the look really matter or is it a matter of taste?
It matters, and it has been measured for twenty years. In the research on website credibility carried out at Stanford, where 2,684 people explained in their own words why they did or didn't trust 51 websites, the most cited category of all is the look: it appears in 46.1% of the comments, ahead of structure and content.
Whoever doesn't know you decides whether to take you seriously before reading a line, and they make that decision by looking.
Does everything have to be redone from scratch?
Not always, and the choice is made by looking at two things. If the current website meets the performance thresholds and has a sensible structure, it often makes sense to connect it rather than redo it, adding the entry points and the log.
If it is built on a set-up that doesn't allow you to work on loading, the time spent patching it comes back as a cost every month. You measure how it is today first, with the same public indicators, and decide on that basis.
How do you measure whether it's working?
With the requests generated by the website that turn out to be truly qualified, agreed as the indicator before starting, never with visits. Visits go up and down for reasons that don't depend on you, and a much-visited page that produces no useful requests is an elegant cost.
The second number is the share of requests that arrive complete, that is, that didn't need a round of emails to understand what the person who wrote wanted.
Notes on sources
- The 46.1% comes from the research on website credibility by B.J. Fogg's group at Stanford, on 2,684 participants and 51 websites: full report, academic version How do users evaluate the credibility of Web sites?. The study dates from 2002 and concerns US consumer websites: we cite it because it measures the order in which people evaluate, and that order is the point, not the exact figure.
- The thresholds of 2.5 seconds, 200 milliseconds and 0.1 and the seventy-fifth percentile are in Google's official documentation on the Core Web Vitals metrics, quoted verbatim. They are the thresholds Google uses to judge, so they should be taken as a requirement and not as a convenient target.
- The 53% of visits abandoned beyond 3 seconds comes from Google, September 2016, on aggregated and anonymised data from about 3,700 mobile sites worldwide that had agreed to the comparison. It is a 2016 figure and concerns sites that use Google's advertising tools: we cite it for the order of magnitude, and mobile networks have improved since.
- We don't publish any percentage increase in requests or conversions. The figures going around on this subject come from sectors, offers and websites that have nothing to do with yours. The number to look at is how many useful requests arrive today, counted over a real month before any quote.
- This page doesn't report results obtained for a client, because this piece hasn't yet been delivered to a client. The tests cited are functional checks carried out in testing.
The other pieces in this group
Customers who are out there, and don't reach youFifteen minutes, with your case in front of us.
How many requests came from your website last month, and how many turned into something? If the answer requires opening a mailbox and counting by hand, that is already the diagnosis. In fifteen minutes on the phone we look at it together and tell you where it makes sense to start, even if we don't end up working together.
You get Mattia Esposito, who then builds the system: there's no salesperson in between. If you'd rather measure on your own before talking, the Diagnostico (in Italian) is twenty questions and five minutes.