Original research

What original research is

Original research is a piece built on numbers your own business collected, published with the method described in full: who was counted, over what period, how many, and what was left out. It is the highest-effort item in this part of the site and it returns more than any of the others, because a finding nobody else holds is the one thing other writers will point at by name.

It is not a summary of somebody else's report. It is not a roundup of borrowed figures with your logo on the cover. And it is not an opinion with a chart behind it. If the number did not come out of your own records or your own properly run survey, this is not the service, and an argued piece from experience is the honest version of what you actually have.

When you need one

You need this when your trade argues about something nobody has measured. Every trade has two or three: how long a decision takes, how often a job comes back, what the real range of prices is. The firm that counts it first owns the answer and gets named whenever the question comes up.

It also suits a business sitting on records it has never looked at sideways. Quotes issued, jobs completed, service calls and delivery times are data already, collected honestly for another purpose, which makes them better evidence than anything you could design.

You do not need this if you have neither records worth counting nor a way to ask customers a question. Publishing a finding drawn from a handful of replies is worse than publishing nothing, and I will say so on the first call rather than take the work. If the aim is to understand your buyers rather than to be quoted, that is research done for your own use, which is cheaper and does not have to survive a stranger's scrutiny.

What counts as usable data

• Your own operational records. The strongest option. Already collected, already true, and gathered before anyone thought of publishing them, which is what makes them credible.
• A customer survey run properly. One defined population, the same wording for everyone, questions written so they do not lead, and a reply count you are willing to print beside the finding. A survey done carelessly is worse than none: a leading question produces a number you will be asked to defend and cannot, and one careful reader can take back all the standing the exercise was supposed to buy.
• A small structured sample, gathered on purpose. Something like the published lead times of forty suppliers, recorded on one stated date by one stated method. Small is fine. Undocumented is not.
• What is never usable. A poll of your own mailing list presented as a finding about the whole trade, a number whose source you could not reconstruct in a year, and anything you would have dropped had it come out the other way.

What you get

The published piece. Twelve hundred to two thousand words in a shared document you own, with the finding stated in the first paragraph, the method described plainly, and the findings written in words as well as figures so a reader understands them without studying a table.

The method note. A separate page naming the population, the dates, the count, the exclusions and a person to contact with questions. This is what makes the work quotable, and it is the part most businesses skip.

A chart brief and a one-page summary. The brief goes to a designer, either yours or one I bring in and manage. The summary is the version that gets forwarded to someone who will never read the full piece.

The working file. The cleaned data and the steps taken, kept so the study can be rerun next year against the same method. A finding repeated annually is worth more than one published once.

A worked example

An illustration, not a client. A twenty-two person commercial flooring contractor in Halifax has eight years of quote records in an old system: date issued, project size, and whether it was won, lost or went quiet. Nobody in the trade has published anything about how long buyers actually take to decide, and every supplier in the region assumes it is faster than it is.

The data review takes an afternoon and finds roughly four thousand usable quotes once duplicates and internal test records are stripped out. The finding, written in words rather than figures, is that small jobs are decided within weeks while the largest ones take most of a year, and that the quotes which go quiet are almost never lost on price. The piece runs to 1,600 words, states the exclusions in the method note, and gives property managers one instruction about when to start a tender. It is picked up by a regional trade newsletter and cited twice the following year by writers with nothing else to cite. The firm reruns it the next spring with one more year of records and a second question added.

How it runs

1. A half-hour call. What the argument in your trade is, and what you already hold that might settle it. No charge.
2. A data review. Before anything is designed or quoted, I look at what you actually have. If there is nothing usable, I say so and we stop there rather than building a survey to manufacture a headline.
3. A fixed price in writing. Scope, method, whether a survey is being run, and the number of revision rounds.
4. Collection. An extract from your records, or a survey written, tested on five people and sent, with the fieldwork window fixed in advance so nothing is quietly extended to improve a result.
5. The analysis and the draft. Fifteen to twenty working days for a records-based study, longer where fieldwork is involved. Every figure traceable to the working file.
6. The method review. You and I read the method note as a hostile reader would, and anything that cannot be defended is cut before publication, even when it is the interesting part.
7. Handover. Piece, method note, summary, chart brief and the working file.

What it costs

Original research is quoted as a fixed price after the half-hour call and the data review. It is the most expensive item in this group, because most of the work happens before a word is written. The price turns on whether the data exists already, how much cleaning the records need, and whether you want a long downloadable version as well as the article. Everything is in CAD and fixed in writing before collection starts. The pricing page shows the four ways of working here.

What happens next

Three things usually follow. The finding is worth a longer, designed version for download, which is publishing it as a report. It is worth offering to the trade press under your byline through placing articles on sites your buyers read. And the questions it raises make the material for the next year's study, or for a piece about where the change is heading. None of that is assumed. One study, published and defensible, is a complete piece of work.