Insights
Insights
One article per solution, each about a single decision rather than a category — written to be useful before you hire anyone, including us.
These are not case studies and they are not announcements. Each one takes a single decision that comes up inside an engagement — the one that most often gets made badly, early, and expensively — and sets out how to make it well.
They are deliberately not about places. A country page has a job to do and this is not it: the questions here are the same in Berlin, Toronto and Dubai, and writing nine articles nine times with a city name substituted would produce something worse than useless. Where local regulation genuinely changes the answer, the location and sector pages carry it.
Each article corresponds to one of our integrated solutions, and says so. That is partly navigation and partly honesty: we have a commercial interest in these subjects, and it is better to state the connection than to present a sales argument as disinterested advice.
The order they appear in is not a reading order. Each stands alone, and the useful one is whichever matches the decision currently on your desk — scoping an AI pilot, arguing about build versus buy, working out why a pipeline is short, or trying to publish without it consuming your week.
Several of them describe an exercise you can run yourself in an afternoon: sorting a month of support contacts by the cost of being wrong, comparing four numbers to locate a pipeline gap, totalling three years of running cost on both sides of a build decision. Those are the parts worth doing before you talk to any supplier, including this one.
Articles
AI for custom customer care: what it can answer and what it must not
Most AI customer care projects fail on scope rather than technology. The useful question is not whether a model can answer — it is which answers you are willing to be wrong about.
Read more →When to build software instead of buying it
Building is usually the wrong answer and occasionally the only one. The difference is not scale or budget — it is whether the process you are encoding is the reason customers choose you.
Read more →Why more traffic rarely fixes a pipeline problem
A pipeline shortfall has three possible causes and they look the same in analytics. Buying more traffic fixes exactly one of them, which is why it so often fails.
Read more →Lead scoring that sales will actually use
Most scoring models are abandoned quietly within a quarter. The ones that survive share a property: sales helped build them and can see why a score moved.
Read more →Quote to cash without re-keying: where the handoffs break
Between agreeing a price and getting paid there are usually four handoffs, and in most businesses at least three of them are somebody retyping something.
Read more →The order of operations in a digital transformation
Most transformation programmes are not wrong about what to do. They are wrong about the order, and the order is what determines whether the second phase ever gets funded.
Read more →Naming a B2B company without painting yourself into a corner
Naming is treated as the creative part and is mostly a constraints problem. The names that fail do so predictably, and the failures are visible before launch if anyone looks.
Read more →Founder-led content without turning the founder into a content machine
Founder accounts outperform company accounts consistently, and most founder content programmes stop within four months. The difference is whether the process fits how the founder actually works.
Read more →Product data is the e-commerce growth lever nobody budgets for
Advertising and site design get the budget. Attribute completeness quietly determines how well both of them can possibly work, and it is measurable today.
Read more →How these are written
The constraint we set ourselves is that each one has to be useful to someone who never contacts us. If an article only makes sense as a prelude to a proposal, it has failed.
- They name the decision, not the category
- Not "AI in customer service" but which questions you are willing to be wrong about. The category level is where everyone agrees and nothing is decided.
- They include the case against
- Most of them argue against doing something, at least partly. An article about building software that never recommends buying is marketing wearing a lab coat.
- They give you the test, not the conclusion
- Where possible there is a diagnostic you can run yourself — four numbers to compare, a sorting exercise, a cost to total up honestly. You should be able to reach your own answer.
- They say where the regulatory line sits
- Where a decision runs into GDPR, the EU AI Act or sector rules, that is stated as scoping context with a recommendation to confirm it locally, not presented as legal advice.
- They are dated and left alone
- Each carries a real publication date. We do not quietly re-date articles to look current, because that is a signal aimed at a crawler rather than at you.
- They are long because the short version is useless
- Every one of these subjects has a version that fits in a paragraph and helps nobody. The specific mechanism, the failure mode and the counter-case are what make a piece actionable, and they take room.
- They avoid numbers we cannot source
- No invented benchmarks, no unattributed percentages, no industry statistics we have not verified. Where a figure would strengthen an argument and we do not have a sourced one, the argument is made without it.
- They name what we would decline
- Several articles state work we would not take on — regulated device software, eligibility decisions, headcount-reduction promises from a first phase. A capability description with no boundary in it is a sales document.
Why there are nine and not ninety
Because nine is how many integrated solutions we offer, and each article exists to be genuinely useful about one of them. Publishing weekly to maintain a cadence produces filler, and filler is visible — to readers first and to search engines shortly afterwards.
A high-volume content programme is a legitimate strategy for some businesses and it is not what this is. If you want three hundred pages a quarter, there are agencies who will produce them, and we would rather say that plainly than take the work and disappoint you at volume.
Where an article would be better as a section on a service or industry page, it lives there instead. The sector-specific material — what changes in a regulated industry, which constraints alter scope — is on the thirty industry pages, because that is where it is findable when it is relevant.
Nothing here contains invented statistics, unattributed figures or client results we cannot evidence. Where a claim would need a number to be persuasive and we do not have a sourced one, the claim is either qualified or removed. That constraint costs some rhetorical force and we think it is worth it.
Each article corresponds to a solution because that is where the material came from. These are the conversations that recur in scoping calls, and writing them down means a first call can start from the second question rather than the first — which is better for both sides and shortens the sales process in a way we are happy to admit.
None of these are location pages and none carry a country in the title. That is a deliberate structural decision: publishing the same nine arguments against thirty-two markets would produce a few hundred pages differing only by a place name, which is the doorway pattern and earns a manual action rather than traffic. Jurisdiction-specific material lives on the location pages, where it is genuinely different.
Where an article contradicts something on a service or industry page, tell us and we will fix whichever is wrong. Content written at different times drifts, and a site of this size drifts silently unless someone points at it.
Questions
How often do you publish?
Rarely and deliberately. These correspond to our nine solutions, and they are revised when something material changes rather than refreshed on a schedule to look active. A publication cadence is a marketing metric, not a reader benefit.
Are these written by AI?
They are written and edited by us, and they carry positions we would defend on a call — including the ones that argue against buying something from us. Anything that reads as though it could have been generated from a title has not been finished.
Why is there no country angle?
Because these questions do not change by country in any way that would justify separate pages, and producing nine articles multiplied by thirty-two markets would be a doorway pattern rather than a resource. Where jurisdiction genuinely matters, it is on the location and industry pages.
Can I use this material internally?
Yes. Several of them are structured as exercises you can run without us — the enquiry-path trace, the contact-tier sort, the three-year cost comparison. If that produces an answer that means you do not need an agency, that is a good outcome.
Do you write about your own projects?
Only with a client's explicit agreement, and we do not publish invented or anonymised composites presented as real engagements. That means fewer stories than most agency sites carry, which is a trade we accept.
Where should I start?
With whichever decision you are currently facing. If you are not facing one specifically, the article on diagnosing a pipeline shortfall is the most broadly applicable, because it is the situation most businesses misdiagnose first.
What if I disagree with one?
Tell us. Several of these positions came from clients arguing us out of a previous version, and the articles are better for it. A correction we accept gets made visibly rather than quietly.
Do you accept guest posts or link exchanges?
No. Both are transparently attempts to acquire links rather than to publish anything worth reading, and taking part would undermine the only thing that makes this section useful. We do not send those requests either.
Will there be more articles?
When there is a decision worth writing up properly, rather than on a schedule. If we start a tenth solution, there will be a tenth article; if a subject here changes materially, the existing article gets revised and the revision is visible rather than quiet.
Can we reference these with our own team?
Please do. Several clients have used the diagnostic sections as internal exercises before deciding whether to bring anyone in at all, which is exactly what they were written for.
Do you publish sources for the claims in these articles?
Where an article states a regulatory fact it names the instrument, so you can check it yourself. Where a statistic would strengthen an argument and we do not have a sourced one, the claim is qualified or dropped rather than propped up with an invented number.
Not sure where to start?
Describe the problem rather than the service. We will tell you what we would do about it, and whether it is worth doing at all.
Get in touch