The answers we give on calls, written down.

A short document, one per idea. It contains a question nobody has answered, why your company in particular can answer it, and where the numbers would come from — named sources, not "industry data".

It does not contain the answer. That is deliberate, and it is the part people find strange at first. The answer is the piece of work you would publish, and it does not exist until somebody goes and gets the numbers. Anything claiming to hand you a finished finding on day one is either repeating something already published or making it up.

No. We stop at the brief, and that is a boundary rather than a gap in the offer.

Writing is the part your team, or an agency you already use, does better than we would. They know the voice, the audience and the review process. What none of that machinery does is decide what is worth writing — the decision made earliest, with the least information, and at the highest cost if it is wrong.

It depends where the data lives, and working that out is part of the brief rather than a question left for later. Every brief says which of three cases it is, before you commit to anything.

Reading and deciding, mostly. Someone senior enough to say "yes, we could publish that" or "no, and here is why". An afternoon here and there rather than a standing commitment.

The heavy part is producing the piece itself, and that only starts after you have decided the question is worth it.

Whoever owns the research or content programme. If nobody does, the person who ends up deciding what gets published this quarter.

Downstream of that you need what you would need anyway: somebody who can pull or assemble the data, often a few days of an analyst or an engineer, and somebody who can write. No new capability, no new headcount, nothing for anyone to learn.

That is worth stating plainly because it is unusual. There is nothing to adopt here. If the brief is wrong, you have lost an afternoon of reading, not a rollout.

For a lot of this, yes, and you should. A model is very good at summarising a field, drafting angles, and telling you what the standard treatment of a subject looks like.

The problem is the one question that matters most here: has anyone already covered this? That is the question a model is least able to answer and most confident about, because it cannot tell the difference between "I did not find it" and "it does not exist". Both come out as the same sentence.

And the gap between them is enormous. Models are trained on a slice of the web, months or years old. They do not see reports behind a form, research sold rather than published, conference decks, trade-press archives or anything posted last week. A study can be sitting in plain sight and be invisible to the thing you asked.

Then you have the hard part already, and we are not selling you a replacement for it. An in-house team is better placed than any outsider to run a study, clean the data and write it up in a voice the market recognises.

Choosing the subject is a different job from doing the research, and it is usually nobody's job. In most companies it arrives from somewhere other than evidence: last year's study repeated because the series exists, the loudest opinion in a planning meeting, a topic sales asked for. Everything downstream is then executed well, and none of that execution can rescue a subject that was not worth the money.

The second thing has nothing to do with skill. An in-house team is inside the company. Knowing whether a question has already been answered means researching continuously outside it — competitors, but also universities, trade bodies, journalists, regulators and adjacent industries touching the same subject from another direction. That is a standing job nobody is assigned.

If your team is already strong, that is the best possible case for this. You have the capability to build almost anything, so the only thing standing between you and a piece that gets cited for years is pointing it at the right question.

Keep them. A good agency executes far better than we would, and when a brief is ready an agency is very often exactly who should build it.

The gap is not execution. It is that an agency is paid to run the study, which makes it a poor place to ask whether the study is worth running. Nobody in that room is rewarded for concluding that this quarter’s research should not happen.

There is a second thing, and it is more specific. When an agency runs fieldwork, the agency is also the party that checks the fieldwork. The topline lands, the client sees the summary, and the raw file is rarely opened by anyone independent. That is a normal arrangement and it is not fraud. It just means nobody outside the transaction tests whether the sample held up, or whether the most interesting split in the data was the one that got reported.

We sit on the other side of both. We decide what is worth proving before money is committed, and we are happy to look at what came back afterwards.

They point in opposite directions, which is easiest to see in what each one starts from.

SEO starts from demand that already exists. Somebody is typing a question, the volume proves it, and the work is to be the best available answer to a question already being asked. That is a real discipline and it works.

This starts from something with no demand attached, because the fact that would create the demand has not been published. Nobody searches for a number that does not exist yet. The objective is not to rank for an existing question but to become the thing everyone has to cite when the question comes up.

They are complementary rather than competing. Once you have published something original, all the usual work of making that page findable applies — and applies better than usual, because the page now has something no other page has.

A researcher you hire is capacity. They can run whatever study you point them at, and they will run it better than a process can. What they inherit is the pointing.

A new researcher walks into a company where the subject usually arrives from somewhere other than evidence. Their first study is chosen by the same mechanism that chose the last one, and being excellent at execution does not change that.

The second difference is scope. Deciding whether a question is already answered means researching continuously outside your company and re-checking it as things get published. That is not a task a person can hold alongside running studies, and it is the task that decides whether the study is worth running.

If you have a researcher, this points them at better questions. It does not replace them.

We are not an SEO tool and we are not an AI-visibility tool. We do not track rankings, we do not monitor citations, and we do not sell a position. If that is what you are shopping for, the tools above do it and we do not compete with them.

What is true is that the material this produces performs well in both places, for one reason that applies to each: both systems reward content that can be checked.

A search engine and an assistant are solving the same problem from different ends. They need something they can point at. A page saying a process is slow and frustrating can be paraphrased by any other page saying the same thing, so nothing is owed to it. A page saying a process took a measured number of days across a named population, with the method shown, is the only place that fact exists. Anything wanting to use it has to reference where it came from.

That difference compounds in a way opinion never does. A specific number gets quoted, and each quote is another route back — including by people who disagree with it. The practical consequence is that this is worth doing even when you are not chasing visibility. If you are, the mechanism is the same either way: be the source rather than another summary of it.

No, and the cheaper move is often a page you already have.

Most companies are sitting on pages that rank, get traffic, and make a claim with nothing behind it. Those pages are already doing the hard part, which is existing and being found. What they lack is the one thing that would make anyone quote them.

Adding evidence to a page you already have carries advantages a new piece does not. There is no new slot to win in the calendar and no new mandate to argue for. The page has history, which search treats better than something published this morning. And you find out quickly whether the evidence was worth producing, because you already know how that page performs.

In practice this is often the best place to start. It is a smaller commitment, it uses work you have already paid for, and it answers whether this does anything for you before you build a programme around it.

No, and treating it as one is the most common way this kind of work gets underused.

The difference is in what you end up holding. An opinion piece is spent on the day it is published. A number you produced is an asset, and it can be spent many times without being used up.

• The piece itself. The report, the study, the index — the thing that is cited.
• Several pieces after it. Each cut of the data is its own argument. A finding that differs by company size, by region and by segment is three articles, not three paragraphs.
• A launch rather than a post. Original data is the one thing a journalist can use without checking with anyone. That is a press conversation you cannot have about an opinion.
• Sales and talks. A number your team produced is usable in a pitch, a deck and a conference talk, and stays usable for as long as nobody has replaced it.
• Next year. Measure the same thing again and it stops being a study and becomes a benchmark. The second edition is easier, better attended, and harder for anyone else to start competing with.

This is also why a handful of these a year is not a small number. The unit is not an article. It is a subject you own for as long as nobody outmeasures you.

We start by conceding the thing most people in this position would skate past. If a report is gated, we cannot get to it either. Nobody outside the company can.

We are not looking for the document. We are looking for evidence the work exists. That is a much easier target, because a gated report is gated in order to be promoted. The whole point of putting a study behind a form is to send people to the form. So it arrives with a public landing page carrying the headline finding, usually a press release, a post from whoever ran it, a webinar, and often coverage quoting the number. Any one of those is enough to tell you the subject is taken.

The same holds everywhere else a model does not reach. A conference talk has a programme entry and a speaker’s post about it. A report sold rather than published has a sales page describing what is in it. Research that leaves no public trace at all is research nobody is getting value from, which also means it is not competing with you.

We search across several engines rather than one, because they disagree far more than people expect. In our own checking, one engine on its own reaches roughly a third of what five reach together. Using a single search is close to not searching.

Every source named in a brief has been fetched and checked, and the brief records the date it was fetched. If something could not be reached, it says so rather than quietly disappearing.

This matters more than it sounds, because an invented but entirely plausible source is the characteristic failure of anything built on a language model. A fabricated citation does not look wrong. It looks like exactly the sort of study that ought to exist, with a title that makes sense and an institution that is real. The first person to find out it does not exist is usually a journalist you pitched.

So nothing reaches you on the strength of a model remembering it. If we cannot open the page, it does not go in the brief.

Because they are built from your company specifically rather than from your category. The starting point is what you can measure that others cannot: your product’s own data, the population you can see, the thing you happen to be instrumented for.

A generic idea is one any competitor could also run. That is not a matter of taste, it is a test with a yes or no answer, and anything failing it is dropped for that reason alone before it reaches you.

It is worth saying what this rules out. Anything drawn from your category rather than your company is generic by construction, however interesting it sounds. If the same brief would have worked for three of your competitors with the logo swapped, it is not a brief, it is a trend piece.

The opposite, and the check runs in the other direction. Every idea is tested against what has already been published on the subject before it reaches you, and anything that collides is dropped rather than sent with a note.

The nuance worth understanding is that a competitor being active in a subject is not the same as the subject being taken. If they have published an opinion and you can publish a measurement, that is not a collision, it is an opening. What kills an idea is somebody having already put a defensible number on it — because then anything you publish is a second opinion about their number.

It starts from what you publish, which is a different thing.

How far from home an idea goes is a dial rather than a setting, and being close to home does not make the work derivative. Your own argument, measured in a way nobody in your industry has tried, is new work. The measurement is the new part.

The failure this question is really pointing at is a system that looks at your blog and suggests more posts like your blog. That is a real failure mode, and it is why the second half of the research exists: everything you already say gets checked against everything already published elsewhere, and an idea only survives if it is both yours and unsaid. One of those alone is not enough.

It is close to the opposite, and the confusion is worth taking seriously because the two sound identical when described quickly.

Reactive content is fast by design. Something happens, you have a view on it that day, and the value is entirely in being early. That is a real skill and it is not what this is. Our lead time is a quarter.

What we do is grounded, which is a different word. Grounded means the question comes out of what your field has established and what it has left unsettled, researched continuously. The news is one of the things we look at. It is not the thing being chased.

The second half of the worry is usually about originality rather than timing. The closest kind of idea is a claim you already make repeatedly with no number behind it. Further out is the next question your last piece raised, then territory you could credibly enter, then a relationship everyone in your field implies and nobody has stated. The furthest is a capability you already have and have never pointed at anything. An angle can sit close to home while the instrument is completely new — making an argument you have made six times, and this time measuring it, is not derivative work.

For most of what people want from that question you do not need us, and for the rest you would license it yourself rather than through us. What we can do is tell you which one, at which tier, actually covers what you need, so you are not buying a subscription in order to find out.

There is a more useful distinction underneath, and it decides every version of this question.

One more thing dissolves most of these questions. A lot of what people want from Statista is a republication of somebody else’s statistic, and the primary source is usually public and free. Tracing back to it is cheaper, and a primary citation carries more weight with the journalists and analysts who actually link to things.

This is the right question to ask, and it is the one the brief is designed around.

Before anything is committed, every idea has to survive a test we apply on purpose: is there a story whichever way the number falls? If the piece only works when the answer comes back high, it is not an idea, it is a bet. Those get dropped at the brief stage, which is the cheap place to drop them.

So by the time you are looking at a brief, the boring outcome has already been considered and written down. You can see what the piece is if the number is surprising and what it is if the number is mundane, before you spend anything.

Then we tell you, and that is a result rather than a failure.

It is worth being blunt about how unusual that is. Almost every service in this space is paid to produce output, which means "there is nothing here" is the one conclusion the arrangement cannot reach. Ideas get padded to fill a deliverable. You end up with a list, and the weak items on it cost the same as the strong ones to act on.

A quarter not spent on the wrong study is worth more than a study nobody cites. If the honest answer this cycle is that nothing cleared the bar, you get that answer with the reasoning attached, and the material behind it does not disappear. Something that fails today because the data is not reachable can become viable when it is.

Then you correct it, and that is the most useful thing you can do early.

Everything we build at the start comes from outside your company: what you have published, and what your market has published. Some of it will be wrong, and it will be wrong in ways only somebody inside can see. A product line described as it was two years ago. A capability we took as central that you are quietly retiring. A constraint nobody outside could know about.

None of it is a verdict about you. It is a draft addressed to you, and the parts we get wrong are the fastest route to the parts that matter.

It can happen, and pretending otherwise would be the wrong answer to give you.

Two things reduce it. Ideas are re-checked rather than cleared once, so a subject that was open when the brief was written gets looked at again as the work proceeds. And the continuous research exists partly for this: if something appears in your category that closes a question you are actively working on, that is one of the few things worth telling you the same week.

If it happens anyway, the question usually gets re-cut rather than abandoned. Somebody publishing an opinion on your subject does not close it. Somebody publishing the number does, and at that point the honest move is to say so and put the work somewhere else.

No. Nothing is connected, nothing is installed, and we do not ask for a login.

Everything we build comes from what your company has already published: your site, your documentation, your public filings, anything else with your name on it that anyone can read. That is a design choice rather than a limitation we are working around. It means there is nothing to approve before we start, and nothing to unpick if you stop.

If a brief needs something only you can see, it says so explicitly. Getting that number is your decision, done by your people, and it stays inside your walls.

Often true, and telling us early is genuinely useful rather than a disappointment.

The constraint is usually narrower than it first sounds, and the distinction changes what is possible. "We cannot share customer data" rarely means "we cannot publish an aggregate finding across ten thousand anonymous events." Those are different questions to a legal team, and the second is frequently fine where the first is not.

And where it is genuinely blocked, we would rather find out in the first week than build you three briefs you have to refuse. There are usually public routes to the same subject.

The first map of your company and your field is the slowest part, because it is the part everything else rests on. You see it before anything is recommended, and correcting it is the first thing we ask of you.

After that, ideas arrive as they survive rather than on a schedule. Something obvious and well-sourced can be in front of you quickly. Something that needs a source located, checked and confirmed to still be reachable takes as long as it takes — and being early with a fragile idea is worth nothing.

You stop. There is nothing installed to remove, no integration to unpick, and no data of yours sitting in a system you need access to.

You keep the briefs, the research behind them, and the record of what was considered and rejected. That record is worth more than it sounds: it is a written account of which questions in your field are already closed and why, which stays useful whoever does the work next.

Decide this before you publish rather than after, because the honest measure is not the one most content is judged on.

Traffic is the wrong yardstick here. A piece of original research can be read by a small number of exactly the right people and be worth more than something with ten times the visitors. The things worth counting are whether other people cited it, whether it came up in sales conversations, whether journalists or analysts used it, and whether anyone in your category had to respond to it.

There is a longer-term one that only becomes visible in the second year. If you measure the same thing again, it stops being a study and becomes a benchmark, and the value of a benchmark is that people come back to it without being prompted.