Reading successive filings

Everyone tells you to read the current FDD. That is the document the fourteen-day rule attaches to, and it is the one you would take to a lawyer. It is also a snapshot. The three years in its tables are the years it still wants you to see.

A franchisor that has been offering for a few years has filed several of these, same twenty-three-item template, same system. Put last year’s next to this year’s and the form becomes a control: anything that moved is something they chose, or were required, to say differently. One filing tells you where the system is. Two tell you which way it is moving, and what the current packet no longer shows.

Almost nobody does this, which is the entire value of doing it. The prior filing is not secret. It is just not in the email.

Four comparisons from public filings. Several of the changes are the correct way to prepare the document. They are still the difference between reading a story and reading the edit.

The output is a written question with a page behind it, not a theory about the brand. “Your 2024 Item 19 covers 2020 through 2023 and your 2023 Item 19 covered 2019 through 2022. Please confirm the 2019 figures and explain how the affiliate performed in the year that is no longer shown.” If the answer is boring, that sentence was cheap. If it is not, counsel and an accountant now know where to spend the hour.

Case one: the window rolls, and the loss year rolls off

375° Chicken ‘n Fries filed on 24 February 2023 and again on 30 April 2024. Both documents make a financial performance representation. Both present it the same way: a single aggregate income statement for the corporate outlets rather than figures for individual units.

The 2023 filing’s representation covers calendar 2019 through 2022. The 2024 filing’s covers 2020 through 2023. Set the two tables against each other and the overlap is identical, year for year, which tells you the underlying accounting did not change. What changed is the frame.

Year	Sales	Net income	Margin	Appears in
2019	$701,815	−$42,106	−6.0%	2023 filing only
2020	$809,425	$46,970	5.8%	both
2021	$2,355,698	$772,366	32.8%	both
2022	$3,879,935	$682,480	17.5%	both
2023	$3,782,437	$804,218	21.3%	2024 filing only

FY2019 is the only loss year in the sequence, and it is the year that is not in the current document. Read the 2024 filing alone and you see four consecutive profitable years at margins between roughly six and thirty-three per cent. Read both filings and you see the same four years plus a fifth in which the business turned $701,815 of sales into a $42,106 loss.

Nothing was concealed. Item 19 speaks to the franchisor’s recent fiscal years; a filing prepared in 2024 covers the years a 2024 filing covers, and the earliest year in the previous document falls off the back exactly as it is supposed to. Nobody chose to remove FY2019. The calendar removed it. That is precisely what makes the case instructive: the mechanism that hid the weakest year from the current reader is the mechanism working as designed, and no amount of careful reading of the 2024 document alone would have revealed it. Only the prior filing does.

The comparison also gives back something the newer document keeps: FY2023 sales of $3,782,437 came in slightly below FY2022’s $3,879,935 while net income rose from $682,480 to $804,218. Sales down, profit up, in the same table. That is a question about mix, pricing or cost control, and it is a considerably more interesting question than “is the trend good.”

Two further details fall out of the same comparison. The reporting entity is headed 375 Ventures LLC in the 2023 statement and 375 Enterprises LLC in the 2024 statement. Because the overlapping years carry identical figures, this reads as the same lineage under a new name rather than a different business — but you can only reason that way if you have both statements to compare, and the entity a financial statement belongs to is worth knowing precisely. And the whole representation is an aggregate corporate income statement, which means unit economics cannot be derived from it in either year. Five years of data in two documents still does not tell you what one restaurant makes.

The 2024 filing also carries a royalty footnote reading “five percent (6%)” against 6% in its own Item 6 table. That is a drafting slip rather than a year-over-year change, and it belongs on the same list of written questions, because the answer determines which figure the agreement actually charges.

Case two: survivorship, disclosed in a sentence above the table

Mad for Chicken filed on 13 September 2023, on 3 May 2024 and on 12 March 2025. All three make a financial performance representation covering affiliate-owned and franchised outlets, and all three report revenue only, with no costs and no profit measure.

Read in sequence, the outlet counts inside those representations move like this: affiliate outlets 4 in FY2021, 6 in FY2022, 12 in FY2023, then 10 in FY2024; franchised outlets 0, 2, 3, then 2. Rapid expansion, then contraction. No single filing shows that arc, because no single filing covers more than two fiscal years. The 12 comes from the 2024 document and the 10 from the 2025 document, and the shape only exists once you have laid all three on the desk.

The 2025 filing then explains part of it, in its own words, in the prose above the performance table. Four affiliate outlets “have been excluded from the table below because they closed and did not operate the full year”, and two franchised outlets “have been excluded because they closed and did not operate the full year”. The filing adds that the excluded outlets “were open only two (2) to eleven (11) months during our most recent fiscal year.”

Six outlets closed during FY2024, and the performance table shows the units that survived the year.

Sit with how good that disclosure is before deciding how to feel about it. The franchisor said what it excluded, how many, of which type, and roughly how long the excluded units had operated. It did not bury the fact in a footnote on another page. And the treatment itself is defensible on the merits: a table of full-year revenue that included a unit which traded for two months would be reporting a number nobody could use, and the reader would have to unpick it. Excluding a partial year from a full-year table is arguably the right accounting choice.

It is also survivorship, and survivorship is what makes a table of healthy-looking revenue mean less than it appears to. The remaining units may be perfectly strong. But the population in the table is not the population of the system; it is the population that made it to 31 December. Every average, every high and low, and every impression a reader forms of “what a unit does here” is computed on the outlets that did not close.

The reason this case belongs in an article about successive filings is not the exclusion, which a careful reader of the 2025 document alone would catch. It is what the exclusion means once you have the other two filings. A reader of only the 2025 filing knows six units closed. A reader of all three knows that the estate had reached twelve affiliate outlets a year earlier, that it had been four two years before that, and that the six closures came at the end of the fastest expansion in the brand’s disclosed history. Those are different facts and they support different questions.

The same-unit rows are where the sequence pays off again. Between FY2022 and FY2023, as reported in the 2024 filing: Flushing $3,333,431 to $2,885,923; Brooklyn $1,392,756 to $1,097,591; Bayside $3,591,148 to $3,240,511; Sunnyside $753,334 to $2,150,959; Astoria $1,035,433 to $1,062,335; Chelsea $755,182 to $1,037,237. Between FY2023 and FY2024, as reported in the 2025 filing: Bayside $3,240,511 to $3,272,236; Flushing $2,885,923 to $2,845,751; Williamsburg $1,097,591 to $963,955.

Three of the FY2022-to-FY2023 moves are downward and three are upward, one of them nearly a tripling at Sunnyside. Anyone who reads a filing comparison as an exercise in finding decline will misread that table badly. The aggregate for the same brand went from $9,918,732 in 2021 to $10,861,284 in 2022. Systems move in several directions at once. Know which units moved which way, and ask the operators of those units why.

One row in that list is a lesson in itself. A unit labelled Brooklyn in the 2024 filing and a unit labelled Williamsburg in the 2025 filing carry the identical FY2023 figure of $1,097,591. Labels in performance tables are informal; they are not defined terms and nothing requires them to be stable between documents. Match your rows by figure as well as by name, and where a name changes, ask which address is which before you conclude that a unit appeared or vanished.

Two more items in this sequence are worth citing, not because they matter economically but because of what they teach about reading. The 2024 filing’s Item 19 prose introduces the tables as showing “the 2022 and 2021 Gross Revenue” while the tables it introduces are headed 2023 and 2022. And the 2025 filing carries a page footer reading “Rev. April 2, 2024” although the document was issued 12 March 2025. Both are drafting artifacts of the kind that appear when a document is built by revising last year’s. Neither changes a number. Both are excellent evidence that these documents are edited, not regenerated, which is exactly why a diff finds so much: the parts nobody revised stayed identical, so the parts that did change stand out.

Outside Item 19, the same pair of filings moves elsewhere. On-the-job training rose from 106 hours in the 2024 filing to 196 in the 2025 filing, and the Item 7 range moved as well. A near-doubling of required on-the-job hours is a real change to what an owner must supply in labour and time before opening, and it is invisible to anyone reading a single document, who simply sees a number and assumes it is the number.

Case three: a representation that appears, narrows to one shop, and then vanishes

German Doner Kebab’s US filings run 7 February 2018, 19 August 2021, 20 July 2023, 3 September 2024 and a fifth document registered with the Wisconsin Department of Financial Institutions on 24 September 2025 under filing number 639752 — five documents from one franchisor across seven years, which is the longest run this guide has to work with, and the fifth of them is downloadable free from a state register.

The first two make no financial performance representation at all. Both use the standard formulation: the franchisor does not make any representations about a franchisee’s future financial performance or the past financial performance of company-owned or franchised outlets. The 2023 filing introduces one. The 2024 filing keeps it. The 2025 filing removes it entirely.

Filing	Representation	Outlets covered	Period	Disclosed
7 February 2018	None	—	—	—
19 August 2021	None	—	—	—
20 July 2023	Yes	1 franchised	FY2022	$1,491,322 gross revenues, 58,674 transactions, $25.42 average ticket
3 September 2024	Yes	1 franchised	FY2023	$1,383,053 gross revenues, 64,721 transactions, $21.37 average ticket
Registered 24 September 2025	None	—	—	—

The 2025 Item 19 is worth reading closely, because of what survived the deletion. It consists of the FTC’s standard explanatory paragraph followed directly by the sentence “Other than the preceding financial performance representation, we do not make any financial performance representations.” There is no preceding representation. No table, no measurement period and no figure appears anywhere in the Item; the carried-over sentence is wording left behind from the previous year’s document, which did contain one. Reports are directed to Daniel Bunce in Dallas. A reader who took that sentence at face value would spend a while looking for a table that is not there, and a reader who has last year’s document knows precisely what used to sit above it.

So the sequence is: two filings with nothing, two filings disclosing a single mall unit whose revenues declined between them, and then a filing that discloses nothing at all. Assemble the first four documents and stop, and you would conclude that the representation had been added and retained — the more encouraging of the two directions — and you would be wrong, not because the four documents were misread but because a fifth existed. Confirm you hold the newest filing before drawing any conclusion about change. State registers are searchable and free, and the whole cost of that check is a few minutes on one of them. A conclusion from a stale document is not a conservative conclusion. It is a wrong one.

The middle two filings still reward subtraction. Both representations cover the same single outlet: the unit at F1 American Dream Way in East Rutherford, New Jersey, which the 2024 filing describes as having opened on 21 August 2021 and as “the only open GDK Outlet for the entire 12 months ended December 31, 2023”. Each filing explains that because only one outlet is disclosed, it has not given high, low, median or average figures — the honest thing to say when the sample is one. The 2024 document also restates a Q4 2021 period for the same unit: $409,279 in gross revenues, 17,612 transactions, a $23.24 average ticket.

Now put the two full years beside each other, which no single filing does for you. Gross revenues fell by roughly $108,000. Transactions rose by roughly 6,000. The average ticket fell by about four dollars. That is arithmetic on the franchisor’s own two tables, and it is a much more specific picture than either year alone: more customers, less money, a materially smaller basket. Whether that reflects menu changes, price positioning, channel mix, a shift toward delivery, or something about the venue is not in the document. It is a question for the franchisor and for the operator, and it exists only because there were two tables to subtract.

The representation also never widened. Item 20 of the 2024 filing reports franchised outlets at year end of 1 in 2021, 1 in 2022 and 7 in 2023, and across both filings that carry a representation the disclosed performance covers exactly one outlet — the same mall unit each time. It was introduced when the system had one unit and it still described one unit when the system had seven, so a 2024 buyer was reading the same single mall location that a 2023 buyer read against a system seven times the size. Then it went, in the year the count rose again.

The 2021 filing contains a plan, and the later filings contain the outcome. This is the comparison that only a long run of documents makes available. A note to the audited statements in the 2021 filing — a note written by management about the company’s ability to continue, not a finding by the auditor — says that the company “has two franchised locations in operation as of December 31, 2020”, that it “plans to have an additional five franchised stores opened by December 31, 2021”, that it is “actively working with existing franchisees on the development of 66 additional stores”, and that after year end it signed a development agreement for 15 stores in the Houston metropolitan area. Item 20 of the 2025 filing puts US franchised outlets at seven at the end of 2024. Set the 66 and the 15 against the 7 and the comparison makes itself; no adjective is required. A development schedule is an intention, disclosed as an intention, and the only way to learn how a particular franchisor’s intentions convert into open restaurants is to read what its own later filings report.

And Item 20 of that same 2021 filing contradicts the note bound into it. Table 1 reports zero franchised outlets at both the start and the end of 2018, 2019 and 2020, while the note quoted above says two franchised locations were in operation as of 31 December 2020. One of those is wrong, or the two use different definitions of an outlet, and the document does not say which. It is the best argument in this guide for reading Item 20 and Item 21 against each other rather than trusting either alone, and it is a discrepancy inside a single document that a reader could find without obtaining anything else.

The 2025 outlet summary does not add up. Item 20 of that filing extends the series through 2024 and again reports zero terminations, zero non-renewals, zero reacquisitions and zero outlets that ceased operations. But Table 1’s “Franchised” row shows 2024 beginning at 7 and ending at 7, with a net change of zero, while Table 3 and Table 1’s own “Total Outlets” row show 7 rising to 9 with a net change of +2 — and the company-owned row is zero at every point in the table, so there is no second outlet type for the difference to be hiding in. Projected openings as of 31 December 2024 are one signed but unopened agreement in New York. Do not repair that arithmetic in your own notes. Copy both figures. The document gives two answers. The Item 20 page makes the same point about tables that fail to add up inside one filing.

What Item 20 cannot answer, and where the answer lives instead. Item 20 covers outlets of the US franchisor through the last completed fiscal year. So the 2025 filing is silent about calendar 2025 and 2026, and silent about the estate outside the United States, where Item 1 of the 2024 filing says the parent and its affiliates franchise 170 outlets across the UK, UAE, Canada, Saudi Arabia and Sweden, with 5 opening soon and 15 under development. A brand can close units steadily and still present a clean US Item 20, and a run of zeros in the closure columns is not evidence that nothing closed. It is evidence about US franchised outlets in completed fiscal years, which is a narrower statement than most readers hear.

Closures do exist here and they are documented outside the FDD. The Courier reported that the Stirling GDK on Murray Place, opened in 2022, shut permanently after a “temporary” closure, was delisted from the GDK website and was being marketed to let by TSA Property Consultants, with the company’s chief operating officer, Sofia Dimen, quoted apologising for the closure and saying the company was working with the landlord. That is a named unit, a named publication and a quoted officer of the company, which is the standard a closure claim has to meet. Third-party directories described the system as “over 140” and “147” locations in the same period, so a single closure sits against a large base and does not describe a trend on its own. A second rumoured closure, in Brighton, was checked and not substantiated — the unit was still listed with current hours in mid-2025 and carried a customer review dated July 2026 — and so it is not cited here. That is the discipline: cite the specific unit, the date and the source, do not aggregate into “numerous closures” without a count you can support, and do not offer a clean Item 20 as proof of the opposite.

Four further changes fall out of the same five documents.

The buyer being described changed shape. In the 2024 filing, “you” is defined as a person who buys the right to operate five or more outlets, and the Item 7 range is per outlet inside that minimum. So the single disclosed unit is one unit of a commitment of at least five. A performance table covering one mall store reads differently when the smallest thing you can buy is five stores, and that relationship between Item 19’s denominator and the offering’s minimum is the sort of thing that only becomes visible when you are already reading across documents rather than down one.

The narrative count and the table disagree, inside one document. Item 1 of the 2024 filing states that in the United States “we have 9 GDK Outlet franchises open and 1 under development” as of issuance, while Item 20 of the same filing reports 7 at 2023 year end. Those are different as-of dates and both can be accurate; a system can open two outlets between a fiscal year end and an issuance date. Put both numbers on the sheet. Anyone who has been trained by comparing documents is the person who notices it.

The franchisor kept moving. Item 19 of the 2023 filing directs performance-related reports to an address in Concord, Massachusetts. The 2024 filing gives the principal business address as Auburn Hills, Michigan. The 2025 filing gives 11015 Beauty Lane, Dallas, Texas, a trade name of “Doner Kebab Outlet” and Daniel Bunce as Global Chief Operating Officer. Any one relocation is administratively ordinary. Three addresses in three documents is a fact about where support sits, who staffs it and which of last year’s team is training this year’s openings, and it costs nothing to ask.

Item 21 ran in one direction throughout. The audited statements across this run show six loss-making fiscal years out of the six with figures on file, totalling roughly $7.47 million, against an accumulated deficit of $7,609,195 at 31 December 2024, with the auditor’s report carrying an emphasis-of-matter paragraph — an unmodified opinion, not a going-concern qualification — pointing at the company’s dependence on working capital advances from its ownership group, disclosed at $5,936,215 at the end of FY2024. The overlapping years are worth lining up here too: the FY2023 accumulated deficit is $6,095,843 in the 2024 filing and $6,095,561 in the 2025 filing, a $282 difference in the same fiscal year across two documents. Item 21 sets out the figures in full and the auditor’s report covers why the distinction between an emphasis paragraph and a going-concern finding has to be got right.

Read together, the five documents describe a franchisor that requires a five-outlet commitment, reported a loss in every one of the six fiscal years with figures on file, is funded by advances from its ownership group, disclosed one mall unit’s declining revenues for two filings, and now discloses no unit performance at all. Every clause of that sentence is the franchisor’s own disclosure, and not one of them is visible in any single document.

Case four: the comparison that does not work

The most useful case in this article is the one where the technique fails, because a diff run on the wrong pair of documents produces confident nonsense.

Atomic Wings filed on 30 April 2024 and again on 29 April 2025. Neither document makes a financial performance representation: the 2024 filing states that the franchisor does not make any financial performance representations, and the 2025 filing uses the standard formulation about future and past performance. There is no withdrawal to report and no addition. A diff of Item 19 across that pair is correctly empty.

Item 20 is where the trap is. The 2024 document is an area representative offering, and its outlet table counts a type called “Area Representatives”: 1 in 2021, moving from 1 to 5 during 2022, and 5 in 2023. The 2025 document counts “Franchised” outlets: 9 to 15 in 2022, 15 to 18 in 2023, and 18 to 20 in 2024.

A reader who lines up the totals sees a system going from five to twenty in two years and writes down a growth rate. That number is meaningless. An area representative is a party with development rights over a region; a franchised outlet is a restaurant. The two tables are counting different things, and the arithmetic that connects them does not exist. The tables are not wrong and neither document is misleading — they are answering the questions their own offerings pose. The error is entirely in the comparison.

So the first step of a filing comparison is not to open Item 19. It is to establish that both documents describe the same offering: the same legal franchisor, the same type of right being sold, the same unit definition, the same format. If the offering changed, you have not found a trend, you have found two different products, and the honest output is a note saying so. That single check is what separates this technique from the kind of analysis that produces a confident chart from incompatible data. It is the same discipline the comparison worksheet applies across brands, turned to point at one brand across time.

When there is only one filing

A first-year franchisor has nothing to compare, and it is worth being clear about what that costs and what it does not.

Doner Shack’s 2025 filing is a US offering from a franchisor that began offering franchises on 5 September 2024, with zero franchised and zero company-owned US outlets at the start and end of 2022, 2023 and 2024. There is no prior US filing to diff, so every technique in this article is unavailable. What the document supports instead is a baseline: a careful record of the passages you intend to read again next year.

Item 13 is the natural place to start with a young brand, because trademark status is one of the few disclosures that changes in a direction and on a timetable. As at its issue date the filing discloses that the principal mark has no federal registration and that an application has been pending since 3 May 2024, and it states the consequence in its own words:

Currently, we do not have a federal registration for our principal trademark. Therefore, our trademark does not have many legal benefits and rights as a federally registered trademark. If our right to use the trademark is challenged, you may have to change to an alternative trademark, which may increase your expenses.

The same Item states that no litigation over the marks is pending, that the franchisor is “not aware of any superior rights in, or infringing uses of” them, and that there are no effective material determinations of the USPTO, the Trademark Trial and Appeal Board, a state trademark administrator or any court adverse to its rights, nor any pending opposition or cancellation proceeding. Copy that as the document states it, with its date attached, and no further: a pending application is a pending application. Paraphrase it into an outcome and you have invented a fact.

And then check the register, because that is the whole point of a disclosure with a future. The USPTO’s Trademark Status and Document Retrieval status view is public, needs no account, and settles what has happened since. Retrieved on 16 August 2026, the record for serial 79/411,340 — the stylised DONER SHACK mark, filed 3 May 2024 by the Madrid Protocol route on International Registration 1,826,161 — shows a non-final action mailed on 20 December 2024 as a refusal sent to the International Bureau, a response received on 5 March 2025, a letter of suspension on 19 March 2025, suspension checks in September 2025 and March 2026, approval for publication on 25 March 2026, publication for opposition on 21 April 2026 with no opposition filed, and US registration 8,290,085 on the Principal Register, issued 9 June 2026, live and active in all five classes it covers. A separate standard-character application for the words alone, serial 99/401,785, was filed on 19 September 2025 and is suspended as of 7 April 2026, with no registration.

Two disciplines apply to reporting that, and they pull in opposite directions. The first is that the registration is narrower than a registration number suggests: the words “DONER SHACK” are disclaimed, so the registrant claims no exclusive right in them apart from the mark as shown, and what issued protects the composite logo rather than the name. The separate attempt to register the words as words is the one still on hold. The second is that nothing here supports a statement about why the refusal issued or what was cited against it. The status record gives dates and outcomes; the office action itself is not in hand, and a reader who fills that gap with a theory has left the evidence behind.

What the pair of records teaches about method is the useful part. The FDD’s sentence was accurate on 29 April 2025 and is out of date now, because registration issued more than a year later. Both facts belong in the file, each with its date: publishing only the filing’s version would leave a reader asserting something the public register contradicts, and publishing only the registration would erase a disclosed risk that was real for the entire period the document was being handed to prospects. That is what a disclosure with a future looks like when the future arrives, and the reader who copied the language down in 2025 is the one who can see it.

The same filing offers a second lesson that applies to any document, comparison or not. Item 1 names one affiliate as the owner of the marks and licensor to the franchisor, while Item 13 names a different affiliate as the party applying for registration of the primary word and design marks. Both entities are given the same Glasgow address. Reading two Items of one document against each other is the same skill as reading two documents against each other, applied at a shorter range — and it is worth doing first, because a discrepancy inside one filing is a question you can ask without obtaining anything.

The method, compressed

Get the documents, and establish that the newest one is newest. Two is a comparison and three is a trend. Finding prior-year filings covers where they generally live and what to ask for. Search the state registers for a document more recent than the one you were furnished before you write down anything about direction — the GDK case is what that check is for. Record for each one the legal franchisor, the issue date, the format offered, and the fiscal years its tables cover before reading a single figure.

Confirm they are the same offering. Same franchisor entity, same right being sold, same unit type, same format. Atomic Wings is the reason this step is first and not fourth. If the offering changed, stop and write down that it changed; that is a finding on its own.

Diff structurally, not impressionistically. Work Item by Item in numerical order rather than skimming for what looks different, because the changes that matter are frequently the least dramatic on the page. Where change shows first sets out which Items repay the effort and what a movement in each one tends to mean.

Record what left, not only what arrived. New disclosures announce themselves. Departures do not: a year that rolls out of Item 19, an outlet that is no longer in a table, a paragraph of risk language that was tightened, a subsidiary that stopped being named. The 375° case is the whole argument for reading in this direction, and it is the direction almost everyone forgets.

Keep the periods attached to every figure. A number without its measurement period and its document date is not usable in a comparison and will eventually be used as though it were current. This is the discipline that stops a filing diff from becoming a rumour.

Know which questions the documents cannot answer, and go elsewhere for those. Item 20 covers the US franchisor’s outlets through the last completed fiscal year, so it is silent about the current year and about every outlet outside the United States. Closures in those places are real events with public records — trade press, the operator’s own site, mapping data — and each one gets cited individually, by unit, date and source. A rumour that cannot be substantiated is left out, not softened into a hedge.

Turn each difference into one written question. Not a theory. A question with the document, the Item and the page in it. The franchisor question list is the format; the year-over-year worksheet is where the raw comparison goes before it becomes questions.

Take the survivors of that list to the people who can answer them. Operators from the Item 20 lists — including the former-franchisee list — can tell you what a change felt like from inside. Counsel can tell you which changed clause actually alters your position. Neither conversation is well spent on a question you could have answered by reading.

What a comparison cannot do

It cannot tell you why. Every case in this article ends at a question, and the documents contain none of the answers: not why a ticket fell four dollars, not why six units closed in a year that followed the fastest expansion in the system’s disclosed history, not why a training requirement nearly doubled. Filings record outcomes in a prescribed format. Causes live with the people who were there.

It cannot tell you whether a change is good. A rolling window is the rule working. Excluding a two-month unit from a full-year table is arguably correct treatment. A relocated head office may be a growth step. Added training hours may be the franchisor responding to exactly the problem you would have worried about. These findings are questions, not scores, and they are the best-targeted hour of reading available.

And it cannot substitute for the current document. The FDD that governs a transaction is the one you are furnished, with the receipt in Item 23 proving which version arrived and when. Prior filings are context. They sharpen the questions you bring to the current document, the operator calls and the professionals. They do not replace any of them, and nothing on this page is legal, tax or investment advice.

What the technique does provide is a genuine asymmetry. Almost every other buyer walking into a discovery day has read one document. The one who has read three knows which year is missing from it.

Related reading

• Where change shows first — the Items that reward a diff, and what a movement in each one means
• Finding prior-year filings — how earlier documents generally become available
• Year-over-year worksheet — the side-by-side sheet these cases were built from
• Item 19 — population before metric, in a single filing
• Item 20, outlet tables — movement, and how to rebuild it
• Item 21, financials — the franchisor’s own results across the same run of filings
• The auditor’s report — reading the report by its headings, and what an auditor change might mean
• Comparison worksheet — the same discipline applied across brands rather than across years