Three questions wearing one coat
“What did this cost me?” is the first thing a business asks after it finds something, and the sentence is carrying three questions stacked on top of each other. They have different answers, they cost different amounts to reach, and only the first is a matter of observation.
What happened to the business? Revenue, conversion rate, close rate, cycle length, how people behave on searches that carry the company name, before and after a date. These are readings taken off instruments you own. No model, no assumption, nothing to argue with — which is why this tier is the cheap one and the one that holds up under scrutiny.
What caused it? A different order of difficulty entirely. Answering it means finding something that stands in for the world where the content was never posted, and then showing that the gap between that world and yours is not explained by the eleven other things that moved in the same quarter.
What would the business have earned otherwise? This is the number owners actually want, and there is nothing to measure. It describes a world that did not happen. It can be constructed — from assumptions about growth, margin, retention and how long an effect runs — and each assumption is a place where two competent people disagree in good faith.
Nearly all the disappointment in this subject comes from asking the third question and being handed the first with a confident total bolted on. A firm that produces a dollar figure inside a week has not measured anything; it has chosen assumptions you never saw. Whether a statement is legally defamatory is a question for a defamation attorney, and a formal loss figure prepared for a proceeding is a separate discipline with a written methodology behind it. Everything below is about what a business can establish about itself.
The record you either started or did not
Here is the part that decides how much of this is even available to you, and it was settled long before you knew there was a problem.
Nothing in this subject fills itself in backwards. A search property reports from the day somebody verified it and not one day earlier. An analytics property holds what its configuration happened to be collecting, in the shape it was set up to collect. A rank tracker holds the queries a person thought to type into it. A CRM holds the fields that existed on the day the deal was lost. And the page of results for your own company name, on the morning the article showed up, exists only if a human being took a picture of it.
So a business that finds the content in week one, reacts perfectly and starts recording that same afternoon still cannot produce the twelve months in front of it. Those months are not expensive to recover. They are unrecoverable, at any price, by anybody. The missing baseline is the one loss in this whole subject that the person attacking you did not cause.
That is the honest case for keeping a record, and it is worth being exact about what sort of case it is. Recording protects nothing. It deters no poster, shortens no request, improves no outcome. What it does is make the period before an event exist in a form somebody else can inspect, carrying dates that were not assigned in hindsight. When the question arrives — what changed, and when — you either hold that or you hold your recollection of it.
If nothing was kept, say so plainly instead of building a substitute. A baseline assembled afterwards, from a report generated later covering a period nobody was watching at the time, is the most fragile object in this field, and it gives way in precisely the meeting where it was supposed to help.
What to freeze in the first week
Measurement loses a fight in the first week that it can never win back afterwards, and knowing the order in advance is most of the cure.
What happens in that week is that everyone wants to act. Somebody drafts a letter. Somebody replies to the review. Somebody pauses the campaign, edits the page the article linked to, or switches off a tag that was reporting something embarrassing. Each of those is a reasonable instinct and each of them removes part of the record — including the record of the content itself, because the usual reason a post gets rewritten or deleted is that the subject noticed it and said something.
Four things belong before anything is sent to anyone:
- Capture the content in full, with its address visible in the same frame, on the day it is found. It may not be there next week, and a review can be rewritten in place with nothing on the page indicating that it ever said anything else.
- Export the numbers as they currently stand — search performance, analytics, sales, ad spend — into dated files that live outside the tools, before anybody changes a setting inside any of them.
- Fix the date, and write down how it was fixed. The day the content was posted, the day it began appearing on searches for your name, and the day a customer first raised it are three different days, often months apart.
- Open a change log. Everything the business alters from here, dated, in one file. In six months this is what stops a redesign being blamed on an article.
None of that is urgent in the sense a vendor means when using the word. It is simply first, and it costs an afternoon.
Who the account is for sets how good it has to be
There is no single standard for this work, and pretending otherwise is how businesses either overspend badly or produce something nobody can use. The standard comes from whoever is going to read it.
An owner or a board deciding whether to spend anything at all needs a decision, not a proof: what happened, how much of it plausibly belongs to this, what it would cost to narrow that down, and what could be done in either case. Weeks of analysis supporting a decision that comes out the same way regardless is more money lost to the same event.
An insurer, a lender or a buyer conducting diligence wants provenance above all. Dated documents, sources named, exports that plainly were not generated for the occasion, and a clean line between what was observed and what was inferred. Here the record outranks the conclusion, and a modest, well-sourced account beats an impressive one nobody can verify.
An attorney, if it reaches that point, needs to know what is knowable before advising whether anything is worth doing. Harm is the element everybody assumes will be easy. It is not, and discovering how thin the record is at the start costs far less than discovering it later.
Notice who is absent from that list. No platform appears on it. Not one operator decides anything by reference to what content cost a business — they test content against their own written rules, and financial loss is not a category anyone can file under. If the hope was that a well-built damage estimate would move a review, that is the wrong instrument for the job.
The parts no budget buys
Four things stay out of reach no matter what is spent on them, and hearing that at the start is worth more than most of the analysis that follows it.
The counterfactual. What the business would have done instead is a construction rather than a measurement, and it is worth exactly what its assumptions are worth.
How many people saw it. Impressions on somebody else's page are that platform's data, and platforms do not report them to the subject of the content. View and share counts, where they appear at all, are self-reported by the platform and audited by nobody. Any reach figure for content on a site you do not operate was estimated by whoever handed it to you.
What the earlier version said. Where a post or a review was edited before anyone captured it, the previous wording is gone — not withheld behind a process, gone. There is no revision history to request and no fee that produces one.
What never happened. The call that was not made, the introduction that quietly did not get offered, the buyer who read something and simply never got in touch. This is frequently the largest share of the damage and it leaves no trace anywhere at all.
Instruments have limits as well, and the operator of the best one available publishes its own:
“The chart totals can sometimes differ from the table totals. This is usually due to differences in aggregation (property vs. page).”
— Google Search Console Help, “Performance report (Search results)”, read 12 August 2026
That is a small thing and a useful calibration. The most authoritative search data a business can reach publishes the circumstances in which it contradicts itself, which is a healthier attitude to the numbers than most reports built on them display. Google revises these help pages without announcement, so check the current wording before quoting one in anything that matters.
What an honest account of the cost looks like
Strip away the presentation and a defensible account of what an attack cost has five parts, none of which is a headline figure.
- A fixed date, with the reasoning attached. Where the date is genuinely uncertain, a stated range beats invented precision, and the reason for the range is itself information.
- Direct observations from property you control, sourced, with the export dates on the files.
- Every competing explanation that was considered, written out, each marked as excluded, still live, or impossible to test with the data that exists.
- Documents connecting the content to a decision by a named human being — an email, a call note, a procurement question, a cancellation reason typed at the time.
- A plain statement of what could not be separated, in the same typeface as everything else rather than in a footnote.
What that account is not is a total. A single number is what a business wants and it is the part that cannot be supplied honestly on the evidence most businesses hold, which is why the firms that supply one anyway are able to work so quickly.
And none of this reaches the content. The review, the article or the thread sits exactly where it sat before, worded the same way, indifferent to how carefully its effects were documented. The reason to do the work regardless is that the questions turn up later from people who will not settle for somebody's recollection — an insurer, a board, a lender, an acquirer, an attorney weighing whether there is anything here worth pursuing. A record answers them. A memory does not.
Frequently Asked Questions
Where do I start if I want to know what an attack cost?
Start by freezing what exists rather than by analyzing anything. Capture the content itself with its address visible, export search, analytics, sales and ad-spend data to dated files kept outside the tools, and write down the date you believe the event started along with how you arrived at it. Then open a log of every change the business makes from that point. Analysis can wait a fortnight without losing anything. The raw material cannot, because people react to an attack by changing settings, pages and campaigns within days.Can a baseline be reconstructed after the event?
Not really, and the attempt tends to backfire. Instruments report from the day they were switched on: a search property from verification, analytics from the shape it was configured in, a rank tracker from the queries somebody entered, a CRM from the fields that existed. A report generated afterwards covering an earlier period is not the same object as data recorded at the time, and anyone examining it will say so. The honest move is to state that the period before was not recorded, and start recording now so the next year exists.Is measuring worth doing if the content is going to stay up?
That depends entirely on who will ask about it later. Measurement changes nothing about the content — no operator weighs a business's losses when deciding whether a post breaks its rules. What it produces is a dated, sourced account of what happened, which is what an insurer, a lender, a board, an acquirer or an attorney asks for when the subject comes up months afterwards. If nobody in that list is ever going to ask, the money is better spent on the routes that can actually move content.Is this the same thing as a damages calculation?
No, and the difference is worth keeping straight. What this describes is a business measuring itself: observations from its own systems, the alternatives that were checked, and documents linking the content to specific decisions. A formal loss figure prepared for a legal proceeding is a separate discipline with a stated methodology, assumptions set out for challenge, and an opposing expert waiting for each one. That work is not what this site does, and treating a business's own internal account as though it were that is how a fragile number ends up in a filing.Should I measure before or after trying to get the content removed?
The recording starts first, because the act of asking is what most often makes the content change. Posts get edited, accounts get deleted, and reviews get rewritten in place shortly after their subject makes contact — so a capture taken afterwards may be of something different, or of nothing. Export the data before anyone alters a campaign or a page in response, too. After that the two run in parallel: removal attempts proceed on their own timetable, and the measurement simply carries on in the background.How long should I wait before drawing any conclusion?
Long enough that the comparison is not dominated by noise, which usually means several full business cycles rather than several weeks. Two considerations set the floor: search and analytics data arrive with a lag of a few days, so the tail of any window is incomplete rather than falling, and any comparison against last year needs the same weeks of the year to line up. Early readings are worth taking to catch obvious breakage, but treating a fortnight of data as the answer produces a number that moves every time it is refreshed.Published