Eight Weeks of Clauneck Pact Work: A Log With the Failures Left In
This is a demonstration record, not a testimonial. Eight weeks of Clauneck pact work logged across six workings, with the arithmetic kept on every line, including the two results that were worthless and the one that cost real money. Nothing here claims a spirit moved a bank balance. What it does is show what a working log looks like when it is written honestly, which is the only way to learn whether a method is any good.
What this log is, and what it is not
It is a teaching artifact. The point is to make visible the shape of a defensible record: what you write down, when, and what you are not allowed to leave out.
It is not evidence that Clauneck does anything. Six entries over eight weeks cannot establish causation, cannot rule it out, and cannot distinguish a working from a person who was already going to send that email. Anyone presenting a log this size as proof of supernatural intervention is overclaiming, and the overclaim is the first thing a reader should distrust.
What a log this size can do is tell you whether a method produced more completed actions than a control period would have, and whether the person running it noticed anything they would otherwise have missed. That is a modest claim and it is the one the data here supports.
So the record is written in the second person of the practice rather than the first, and the numbers are the argument. Where a result is ambiguous the log says so. Where an outcome is bad the log keeps it, because a log that only contains successes is a marketing document and teaches nothing.
The five columns, and why those five
Every entry carries the same five fields. Nothing else was recorded, and the constraint is what makes the record useful.
Date of the charge. One line, no exceptions. A log without dates cannot produce a hit rate, and a hit rate is the only number that tells you whether to keep going.
The ask, quoted from the working sheet exactly as written, including the number and the date inside it. Quoting rather than paraphrasing matters, because paraphrase is where memory edits the past in your favour.
The action taken in the following seven days, with a day count. This is the field most people omit and it is the one that carries the mechanism.
The result at fourteen days, with a number where a number exists. Fourteen days is arbitrary and deliberately fixed in advance, because a review date you move is not a review date.
The verdict, written the same day the review falls. Not the mood of that evening. The verdict written later is contaminated by whatever happened since.
Two things were deliberately not recorded. Interpretations, and feelings about the working. Both are real and both are noise in a dataset this small. Anything felt gets written in a separate notebook that is never opened during a review.
Working 1: the invoice, 47 days late
The ask, quoted: the 1,400 dollars from the March invoice, by the 14th, and the written reason was that the work was delivered and accepted in April and payment had not followed.
Four-part test applied before starting. Number: 1,400. Date: the 14th. Action: a written request with the invoice attached, sent once. Failure response: if unpaid by the 14th, a second letter stating the relationship is paused pending payment, sent by the 18th.
The refusal condition was written first and it was specific: if the client says the work was not accepted, the working ends and the dispute becomes an ordinary commercial conversation with no magic in it.
Result at fourteen days: paid in full on the 9th, four days after the charge, plus a late fee of 40 dollars that nobody had asked for. Days to result: 4.
Verdict: worked, and the log is honest about how little that tells you. A 47-day-overdue invoice from a client who had already accepted the work is a coin flip, and coin flips land four-to-one in somebody's favour roughly one time in five. This entry is indistinguishable from luck and would be worthless as evidence if it were the only one. It is in the log because the failure response was never triggered, which is itself information about the client's character.
Working 2: the referral, where no number existed
The ask, quoted: one introduction to whoever handles procurement on the client account, by the 20th, because the current route goes through a person who left in June and nobody has been told.
No number attaches to this ask, which broke the log's own rule. It went in anyway with the verdict field marked unnumbered, and that turned out to be the most useful decision in the eight weeks.
The clause doing the work here is the discovery of hidden things, not any money clause. The route had genuinely gone stale and nobody had noticed. That is exactly the shape of thing a spirit of discovery is described as finding.
Result at fourteen days: two names, both unsolicited. One was a former colleague, useless. The other ran procurement. The project she authorised started six weeks later at 600 dollars, which falls outside the fourteen-day window and is therefore not in the result field. It goes in the notes with a date.
Verdict: partial and slow. The useful part was not the money and it was not on the timeline the ask specified. A rigid fourteen-day review would have scored this as a failure, which is a strong argument for the unnumbered-ask exemption and a warning about scoreboards that are too tight.
Working 3: the call that was never going to be made
The ask, quoted: make the phone call to the supplier about the overdue credit, on the 9th, because it has been outstanding for five weeks and every day of delay makes it harder.
No financial target attached. The charge was done in ninety seconds at a red traffic light, and the number in the result field is zero.
Result at fourteen days: the call was made. The supplier released the credit the same afternoon. The sum involved was larger than the total of the other five workings combined.
Verdict: worked, and the result field says zero, which looks like a failure and is not. This is the clearest illustration in the whole log of why the financial column cannot be the only column. The money was released by a phone call, and the sigil did not cause the phone call. What the sigil might have done is move a task that had been sitting for five weeks to the front of a queue, on a specific morning, with a written reason attached.
Nothing here distinguishes that from an ordinary act of administrative nerve. But the log now contains an entry where a five-week-old avoidance ended on a date that was written down beforehand, and that is the sort of thing worth accumulating.
Working 4: the recharge mistake and the vague ask
The ask, quoted: consistent months at 8,000 dollars, by the end of the quarter. This is the vague-total version and the log marks it as the worst-constructed ask of the run.
Two errors in one entry. The first is the total. There is no line item behind 8,000 dollars a month, so there is no specific action that could produce it, and the working sheet's own instruction to write the action was quietly ignored because writing it was uncomfortable.
The second is the recharge. The sigil was charged on the 2nd, again on the 4th, and a third time on the 9th when nothing appeared to be happening. Three charges for one intention, spread over a week, converted a single decision into a sequence of anxious restarts. Nothing in the record can be attributed to the original charge after that, because the object of study no longer exists as a single event.
Result at fourteen days: nothing. Result at thirty days: nothing. Verdict: failed, and the two errors are jointly responsible. Which of them did the damage is not recoverable from this record, and the log does not guess.
Cost beyond the zero: three weeks during which the actual overdue invoice from the December account was not chased, because attention was on the total. That is the real price and it is a price paid in a currency the log does not otherwise track.
Working 5: the escalation, and what it cost
The ask, quoted: recognition of the added scope at the end of the year, in writing, before the renewal conversation. A line item, correctly framed, and it is the best-constructed ask in the log.
Then the sentence that ruined it: and if that is not possible, a substantial increase. An escalation clause bolted onto a specific request, added after the sheet was written.
The logic of the escalation is that it gives the other party an easy way to say yes to something smaller. What it actually does is convert a defensible request for compensation into a negotiation, and it hands the other party a reason to read the whole thing as a demand.
Result at fourteen days: an increase of 600 dollars a month, which is real money. A second account was also moved to a different supplier at worse terms, and the account relationship was never repaired. The 4,800 dollars of that account is the figure that matters.
Verdict: partial success, net loss. The entry exists in the log because a reader who saw only the 600 dollars would draw the wrong conclusion, and because the mistake is the kind that gets repeated by people who never see it written down. The escalation clause was not the problem. The problem was that the clause was added after the request was drafted, which meant the request was no longer the thing being made.
Working 6: the same target, run again properly
The ask, quoted: the added scope from November, worth 900 dollars a month, confirmed in writing before 1 January, because the work had been running for four months unpaid at the contracted rate.
One charge, on the 18th. No recharging. No escalation clause. The refusal condition written first: if the answer is no, the scope returns to the contracted rate in writing and the account continues unchanged.
Result at fourteen days: agreement in writing at 900 dollars a month, effective 1 January, with the account retained. Days to result: 11.
The comparison with Working 5 is the whole point of including both. Same employer, same substance, same underlying facts about unpaid scope. The better-constructed request returned 50 percent more and kept the account that the escalated one lost.
Verdict: worked, and the log will not claim the sigil did it. The defensible reading is that a specific, bounded, single-ask request is a better thing to send than a broad one followed by a demand, and that this holds independently of any belief about the working. The monetary difference between the two entries is 300 dollars a month and one account.
The arithmetic, kept honestly
Totals across the eight weeks, with the unnumbered and out-of-window results separated rather than folded in.
Invoiced money actually received inside a review window: 1,400 dollars, from one entry. Money received outside a window and recorded in notes: 600 dollars, from one entry. Money released by a phone call rather than a request: not counted, because the mechanism is not the working.
Ongoing monthly change: 900 dollars a month gained, 600 dollars a month that would have been gained and was not. Net positive 300 dollars a month, with one retained account.
Losses: one account, 4,800 dollars, and the opportunity cost of three weeks not chasing a real invoice.
Hours spent: roughly twenty-two, most of it on workings 4 and 5, which produced the worst outcomes in the set. Money spent on apparatus: 0 dollars, because the tools were already owned.
Net defensible result: about 2,000 dollars of invoiced money and one ongoing improvement, out of six workings, two of which were useless and one of which lost money. Written plainly so that nobody has to reconstruct it, and so the next person does not have to guess what a real run looks like.
Four patterns that survived the eight weeks
Line items outperformed totals, every time they were compared. A 1,400 dollar invoice and a 900 dollar monthly scope both produced results. An 8,000 dollar monthly total produced nothing and displaced a real receivable. The mechanism is plain: a line item names an action, and a named action is a thing you either do or do not do on a known date.
One charge beat three charges, twice. Workings 1, 3 and 6 each had a single charge and all three produced an outcome inside the window. Working 4 had three and produced nothing. The log cannot say why, and the most likely explanation is not mystical. A single charge is a decision, and a decision that keeps being revisited is not a decision.
The action was the working in four cases out of six. Working 3 is the extreme case: the sigil plausibly did nothing and the credit still came back, because the phone call is what recovered it. Any model of these results that omits the action is not explaining the data.
The failures carried the information. Every pattern above is visible only because two entries were useless and one was a net loss. A log containing six successes would have supported any conclusion at all, which is precisely what makes it worthless as evidence.
Three mistakes, ranked by what they cost
The escalation clause, which cost 4,800 dollars and 300 dollars a month. Most expensive by an order of magnitude, and the mechanism is simple: it converted a request into a demand, and demands are refused at a higher rate than requests even when the substance is identical. The two entries that differed only in this respect are the cleanest comparison in the log.
The vague total, which cost three weeks and a receivable. A target with no action attached to it cannot be executed, so the time goes somewhere else, and where it goes is always something with a real invoice attached. The December account is the receipt.
Counting the boldness call as a financial result, which cost nothing and was still the most corrosive habit in the run. When a zero-financial-result entry is filed alongside two that paid, the log starts telling a story in which the practice is reliably effective, and the person running it starts making decisions on that story. By Working 5 the framing had already shifted toward escalation, which is what Working 5 did. The distortion did not come from a single entry. It came from the aggregate.
What to copy and what to leave
Copy the five columns, the fixed fourteen-day review, the quoted ask, and the refusal condition written before the charge. Together they cost about four minutes per working and they are the entire difference between a method you can improve and a collection of memories that feel significant.
Leave the escalation clause, the vague total, and the second charge. All three are common, all three feel like commitment in the moment, and all three are ways of avoiding a small uncomfortable sentence about a specific number.
Also leave the interpretive layer out of the dataset. It goes in a separate book, it is genuinely interesting, and it must not be consulted while writing a verdict. This is the hardest rule on the page and the only one that is purely about honesty rather than technique.
One addition for anyone repeating this: number your unnumbered asks. A referral has no sum attached, but it has a date and a count, and those two fields are enough to score it. The unnumbered exemption that saved Working 2 from being scored as a failure only worked because the date and the count were still there.
What the log cannot tell you
Three limits, stated so the record is not read for more than it supports.
It cannot attribute the results. Six entries in eight weeks, drawn from a life in which the person running it was also sending emails, chasing invoices and doing ordinary competent work. There is no control condition here and no way to construct one retrospectively. A log that claims attribution from a sample this size is making a statistical error, not reporting a finding.
It cannot measure the sigil. The draw, the charge, the encoding and the timing were held roughly constant across the run, which means they are not variables, which means this log says nothing at all about whether they matter. Answering that question needs a different design with a real comparison, and the technique is covered separately here.
It cannot tell you whether a repeat run would land the same way. One person, one set of circumstances, one eight-week window, with the surrounding business climate unrecorded. The reason Working 6 outperformed Working 5 might be the construction of the request, or it might be that the second conversation happened after four months of unpaid scope had become embarrassing enough that the employer wanted it resolved. Those are very different worlds and this record does not distinguish them.
What survives all three limits is narrower and still worth having: a specific bounded request beat a broad one followed by a demand, by a measurable margin, in the only comparison the log contains. That is a finding about request construction, it is free, and you can use it without believing any part of the rest.
The tool stack this log actually needed
Kept deliberately small, because the log shows that apparatus spending has no relationship to results and this is the one place on the site where that claim should be demonstrated rather than asserted.
A paper notebook. Four of the five columns fit on one page per entry and writing them by hand is faster than opening anything. A physical log also has a useful property: it cannot be edited retrospectively, which is the main way a log like this gets quietly falsified.
A sigil generator. Free browser tools cover the drawing, and there is a reasonable argument that hand-drawing produces a better object for a beginner, because the encoding step is where the intent gets specified. Where an app genuinely earns its money is storage and versioning, so you can see that the same intention was encoded the same way on two different occasions rather than assuming it.
A moon phase reference, if you want timing as a variable. Free tools cover it, and the log deliberately held timing constant, so this was the one purchase that bought nothing. Bought anyway, because the variable is cheap to track and somebody should eventually test it.
A results tracker, only once you are past the fifth entry and the notebook is genuinely full. Before that, an app is a way of not writing anything down.
Total for the eight weeks as logged: one notebook, and whatever the generator cost, which was nothing. The single most expensive item in the run was the escalation clause, and it was free.
Frequently Asked Questions
How do you keep a ritual log without it becoming fake evidence later?
Write the verdict on the review date, not afterwards, and write it before you read anything you wrote about the working. Quote the ask rather than summarising it, because summary is where the memory adjusts the past. Keep interpretations in a separate book that is closed during reviews. The single most important rule is to allow entries that say nothing happened, since a log that cannot record a failure will drift into justifying whatever you already believe.
How many workings before the log tells you anything?
More than most people will run and fewer than the sample size that would settle a question. A dozen entries with dates, quoted asks and fixed review windows is enough to expose your own construction habits, which is what this log was for. It is nowhere near enough to attribute results to a sigil, and no number of entries gets there without a comparison group, because the confound is your own changing behaviour across the same period.
Should the review window be fourteen days?
Fourteen is a reasonable default because most mundane outcomes in this kind of work resolve inside it, and a short window stops a working from being quietly extended until something happens. Change it if your field runs longer, but set it before the charge and do not move it afterwards. The working in this log that fell outside the window was the referral, and a rigid window scored it as a failure when the outcome was real and six weeks late.
What if the numbers are too small to write down?
Write them anyway. A forty dollar late fee is a real number and it was the most surprising line in this log. Imprecise numbers are still comparable across entries, and vague targets are precisely what the field notes show going wrong, since the vague-total working displaced a real receivable for three weeks. A small specific number also keeps the action attached, which is the mechanism the successful entries shared.
Does this log argue that the sigils did nothing?
No, and it is careful not to. Six entries cannot attribute anything, and the honest position is agnostic rather than deflationist, because the design has no comparison and therefore cannot exclude a real effect. What the log does show is that request construction mattered more than any of the ritual variables, and that at least one entry, the phone call, produced its result through a mechanism that has nothing to do with the working. Hold both of those and stay honest.