UK investing research under the microscope: an editorial policy read as a specification
Every research site publishes one and almost nobody opens it. Assessed the way a specification gets assessed, most promise nothing anyone could fail to do.
Show me the record
Somebody whose job is deciding whether written commitments mean anything opens an editorial policy looking for one thing, and it is not tone. A policy says what ought to be true. A procedure says who does what, in which order, and what record the doing leaves behind. Only the third can be produced when asked for, which is why assessment runs on it and why the standing request is show me the record.
Put to a page of promises the request becomes a drafting question: what artefact would each clause generate? A dated correction stating what was wrong rather than a quiet repair. A change history against a comparison table. A note that a firm objected, and what followed.
Clauses generating nothing are unverifiable from outside and unenforceable from inside as well, since nobody in the building can point at anything either, and a rule nobody can point at is a rule that gets admired rather than followed.
A promise nobody could fail is not aimed at anybody
Behind that sits a cruder test, and it comes second because everybody already knows the answer to it. Describe the act that would put the publisher in breach of this sentence. Our standards are high. We take accuracy seriously. Describe a departure from either and you find yourself inventing a villain, since nothing in the wording rules out anything an ordinary publisher would want to do. A sentence nobody can fall short of is not a commitment, it is a description of good manners.
Naming the company names nobody
The familiar diagnosis is the missing actor, and it is correct as far as it goes. The rung past it is the one that changes anything. Naming the company does not help, since a company cannot be spoken to and cannot discipline itself. Clauses that survive an examination name a role. The reviews editor. Whoever signs off a change after publication. Then a breach has somebody attached to it, and the follow-up becomes available: if this happened, who would be spoken to, and where would the conversation be written down?
Be blunt about how weak all of this is on its own. A well-drafted document can sit above a newsroom ignoring it, and a woolly one above people who behave scrupulously from habit.
| The sentence you usually meet | The same promise with somebody in it | The act that would breach it | What a reader could see |
| Content is produced independently | No one on the commercial side may reorder a comparison table | A reorder requested by a sales manager, and made | Nothing, unless the table carries a change history |
| We test the products we write about | Every platform scored here was funded and dealt on first | Scoring a platform nobody opened an account with | Partly: a funded account generates a bill |
| We work with selected commercial partners | No partner sees an article, or hears of a downgrade, before publication | Sending a draft out for comment | Nothing at all, from outside |
| Our reviews are impartial | No score changes after publication except by re-testing, and the change is logged | Editing a score to match a new arrangement | Nothing, unless scores carry a history |
| We welcome feedback from the firms we cover | A disputed finding is re-tested and the outcome published either way | A quiet edit with no note | Only where corrections run as corrections |
Stock clauses on the left, rewritten on the right until each one names somebody and can be breached. Four of the five breaches would leave no trace outside the building.
Who is allowed to move a row in a comparison table?
A comparison table’s order is the most commercially valuable object a research site owns, and the one a partner most wants moved. So the clause that matters says nothing about who writes the reviews, since nobody with a sales title is writing them. It concerns influence over which firms appear, the sequence they appear in, and how fast a decided downgrade reaches the page.
A version satisfying an examiner names the role holding the pen and the roles that may not touch it, then says what becomes of a request arriving anyway. Whether such a request gets written down is the difference between a rule and a preference. A publisher unwilling to put that on paper has told you the conduct is undescribed rather than bad, which is still something you take on trust.
Does anyone outside the building see it before you do?
Advance sight is the practice this sector describes least, and it has a respectable version. A firm shown a draft can point out that a fee was withdrawn, an account type renamed, the wrong tier tested. Reviews are more accurate for it, and a publisher who checks nothing with anybody is not obviously the more careful operation. A flat prohibition is therefore the wrong ask, and the clause has to work harder than one. The line falls between correcting a fact and arguing a verdict, and it holds only where the exchange is recorded: what was put, what changed, what was left standing. Without the record the two are indistinguishable afterwards, including to the people in it.
When a tested firm says the finding is wrong
Sooner or later a tested firm disputes the result, and three things can follow. The finding stands and the objection is noted. The finding was wrong and is corrected where the original ran. Or the finding stops being there. The third is the common one and the only one invisible by construction, since a deletion leaves the trace of never having published.
That is the widest of the three gaps, so try two documents. One is the editorial policy of the comparison site you already use, unopened. The other is this page, published by a site whose stated method is that it opens and funds live accounts with its own money to test UK trading platforms rather than compiling rankings from providers’ published fee schedules. None of the three questions above is an input question. A funded account settles what a platform charged and keeps the receipt; none of it establishes who holds the pen over an ordering, which no outsider can settle, here or anywhere.
Why the weak version keeps getting written
Nobody drafts these badly by accident. A specific commitment is a hostage. Write that no partner sees an article before publication and you have made a sentence a departing employee can contradict and a regulator can read back to you. Write that content is produced independently and you have made nothing, which is why review converts a role into a construction with nobody in it.
The demand side deserves its share. Almost nobody opens these pages, so nothing rewards the publisher who spends a fortnight and a legal fee on a harder version, and the sensible business decision is boilerplate. No figure exists for how many UK investing sites name a role in theirs. Supplying a made-up count would be a strange way to close an argument about unverifiable sentences, so take the reading as a reading.
The four sentences that would settle it
None of this needs a long document; four sentences answer the examination. Who holds the pen on the published order, and which roles may not touch it. Whether any commercial party sees an article before publication, and what is recorded when one does. What happens to a disputed finding, and where the outcome appears either way. Who may alter a score afterwards, and what that leaves behind.
Such a document would not make a research site honest. It would make a departure describable, the only material an outsider has.
An empty result is worth having too: the strongest thing this publisher would commit to writing was that content is produced independently, and you now know what weight that will bear.
…
Read More →