Scanned Purchase Agreement: Filled Blanks, Boxes, Addenda

To get reliable text from a scanned purchase agreement, treat it as a form and not as prose. Convert the printed form, each counteroffer and each addendum as separate files, then read every filled blank and every checkbox against the page image, because the printed words are the same in every deal and the entries are the deal.

Why is a scanned purchase agreement so often a filled-in form?

A purchase agreement for a house or a small business can be prepared on a form that someone else published, such as a state commission, a brokers' association or a forms vendor. The parties fill blanks, check boxes, attach addenda and sign. What reaches your file is a scan of that completed form, not a document drafted for the deal.

Texas shows how this works. The Texas Real Estate Commission publishes a One to Four Family Residential Contract (Resale), form 20-19, and its page for the form says: "This is the most frequently used contract form." The approval notice printed after the form's signature lines states that "TREC forms are intended for use only by trained real estate license holders." Other states and other kinds of sale have their own publishers, so look at the header and footer of the form in your file for the publisher's name, the form number and the revision date.

A form of this kind changes what conversion has to deliver.

Two consequences follow.

The printed text is the easy part. It is clean type, and if you know the form number and revision you can read a blank copy from the publisher alongside the scan. Confirm the revision in the footer matches, because forms are revised: the approval notice on the Texas form says it "replaces TREC NO. 20-18".

The entries are the hard part. They are short, they sit on rules and inside boxes, and they were written by hand, typed, or placed by form-filling software. They are also the only words the parties chose.

Sales of businesses come both ways. An asset purchase on a broker's printed form behaves like the property form described here. A negotiated agreement drafted by lawyers behaves like any other long contract, and the form problems do not arise.

Filled blanks carry the deal on an asset purchase or a stock purchase. Copy the price, the closing date, and each condition precedent from the page image. Checkboxes beside included assets or shares need the same care as the dollar figures. Write the initials at the foot of a page into your log with that page number.

Typed, handwritten and e-filled blanks in a purchase contract

Recognition treats the three kinds of entry differently. Entries placed by form-filling software are clean type and read best. Typewritten or stamped entries read well when they clear the printed rule. Handwritten entries are the weakest, and a short handwritten number on a line is the case to check first.

The entry loses its label

Suppose a form prints "Within ____ days after the Effective Date". In the converted text an entry of 10 may appear as "Within 10 days", as "Within days 10", or on a separate line after the paragraph. Reading order around a blank is a general form problem, and the post on keeping each answer beside its field label explains it. For a purchase agreement, the practical step is to list the blanks from a clean copy of the form before you look at the output, so you know how many entries to find.

The empty blank matters too

A blank left empty is an answer. On the Texas form, for instance, the paragraph on the option fee has a blank for the dollar amount, and the form states what follows "If no dollar amount is stated as the option fee". Record an empty blank as "blank", and confirm on the image that it is empty and not faint pencil.

Handwriting over print

A buyer writes "$5,000" where the form's dollar sign is already printed, producing "$$5,000" or a figure whose first digit is hidden by the printed sign. Someone strikes a typed date and writes another beside it with initials. Changes of that kind have their own method, in the post on handwritten changes to a contract. Schema Law's home page says recognition is strongest on clean print and weaker on handwriting, so plan to read handwritten entries from the image whatever tool you use.

Software-filled forms that were printed and scanned

When a form was completed in form software, signed on paper and scanned, the entries look like the printed text. Compare against a blank copy of the same revision to see which words are entries. If the sender still has the filled PDF from before printing, ask for it: that file has real text and is the better source for everything except the signatures.

Which checkbox choices in a purchase agreement change the deal?

The boxes that change the deal are the ones that select between alternatives the form prints side by side: how the price is financed, whether the buyer may walk away after inspection, who pays for what, which items stay with the property, and which addenda are part of the contract. Read every one of them on the image.

The Texas resale form gives concrete cases. The text here is taken from the commission's published copy of form 20-19.

Your form will differ. The method is the same for any of them:

1. List every box on a blank copy before reading the scan. Count them per page.
2. Mark each as checked, empty or unclear from the image, at a zoom high enough to see a light mark.
3. Test "check one box only" groups. Exactly one should be marked. Zero or two is a finding to report, not to resolve.
4. Follow each checked box to its consequence. A checked financing addendum means that addendum should be in the set. A checked "specific repairs" box means a list should follow.
5. Look for initials beside any box that was changed.

Do not take checkbox states from converted text. A box may come through as a symbol, a stray letter or nothing, and the result can vary from page to page. Even an unscanned PDF can show the problem. In one extraction of the commission's own PDF of the form with the pdftotext utility, each empty box came back as a lowercase "q", and another tool may return something else. How recognition deals with boxes in general is covered under whether OCR can tell a checked box from an empty one.

Which choice the parties were entitled to make, and what a wrongly checked box means, are questions for the lawyer. The handler's job is an accurate list of what the page shows.

Counteroffers and addenda that override the printed purchase form

The printed form is the base, and the papers signed with it or after it can change it. Put every paper in the order it was signed, then read each one for its own statement of what it changes and what it leaves in place. The papers' own words, not their position in the scan, tell you how they relate.

A purchase file can hold several layers:

The Texas form shows the base rule in the form's own words. Paragraph 22 begins: "This contract contains the entire agreement of the parties and cannot be changed except by their written agreement. Addenda and notices which are a part of this contract are (check all applicable boxes):". On that form the boxes checked in paragraph 22 are therefore your index of attachments. Read the equivalent clause on yours.

Checks that the layers call for:

• Every addendum listed is present, and every addendum present is listed. A loose addendum that no paper refers to goes on the open items list.
• Each paper names the same property and parties. A paper from another deal can end up in the same scan, and the address is how you catch it.
• Each paper is signed by both sides. A counteroffer that shows only its maker's signature goes on the open items list.
• Dates and times are legible. Offers and counteroffers may carry a time of day and an expiry. Read them from the image.
• Numbering has no gaps. Counteroffer No. 1 and No. 3 with no No. 2 needs a question to the sender.

Whether a later paper prevails over the printed form, and whether a counteroffer was accepted in time, are legal conclusions. Present the papers in order with the controlling sentences quoted and page references, and leave the conclusion to the lawyer. The general method for sequencing changes to a contract is in the post on putting contract amendments in the order they took effect.

Converting a purchase agreement and its addenda into one working set

Split the scan into its separate papers, convert each one, then build a single index that lists them in signing order with the terms each one sets. The working set is that index plus a searchable copy of every paper and one deal sheet. The scan as received stays untouched beside it.

The example is fictional: Renata Sandoval agrees to buy a house from Clifford and Mae Whitlow, and the broker sends one 31-page PDF.

1. Save the PDF as received and record its page count and sender.
2. Page through the images and mark boundaries. Note where each paper starts and ends: the form, each counteroffer, each addendum, any disclosures and receipts.
3. Record the form's identity. Publisher, form number and revision date from the footer of the main form and of each addendum.
4. Check completeness per paper. A form that prints "Page 4 of 12" in its header makes this quick. The wider tests are in the post on missing pages in a scanned contract.
5. Split into one file per paper, named so they sort in signing order.
6. Convert each file. Export a searchable PDF for the record copy and Word or plain text for reading. In Schema Law each file is uploaded from the browser as a PDF, PNG or JPEG of up to 20 MB, and you can check each parsed page against the original before exporting; the steps are shown under how it works.
7. Read the blanks and boxes against the image, using the lists you made from a blank copy of the form.
8. Build the index.
9. Fill the deal sheet from the index, not from memory of the first form.

The index for the Sandoval purchase:

Row 3 shows why order matters. The sellers signed the counteroffer on 4 September, before they signed the form on 5 September, and the buyer accepted both that day. Sorting by the last signature would hide the sequence, so record every signature date.

A purchase funded by a loan produces a second set of papers at closing: the note, the security instrument and the lender's forms. Those are covered in the post on OCR for loan documents.

Pulling the price, deposit and contingency dates into a deal sheet

A deal sheet is one page that states the current terms of the purchase with the paper and page each term comes from. Build it from the index, taking each term from the latest signed paper that sets it, and show the earlier value when a term was changed so the lawyer can see every change.

The deal sheet for the Sandoval purchase:

Checks before anyone relies on it:

1. Add the price. The cash portion and the financed portion must equal the sales price: $82,500 plus $330,000 is $412,500. A sum that fails means a digit was misread or one paper changed a figure and another did not.
2. Match the loan amount in the financing addendum to the financed portion.
3. Read every amount and date from the image, not from the converted text.
4. Record intervals as written. "7 days after the effective date" goes on the sheet in those words. Converting intervals to calendar dates has rules about weekends and counting, and the form may state its own. That step belongs to the lawyer or to whoever keeps the calendar, using the method in the post on how to turn a contract notice period into a calendar date.
5. Leave unclear entries marked unclear. "Earnest money: $6,000 or $8,000, handwritten, see page 2" is a better entry than a guess.

Head the sheet with the date it was prepared, the name of the scan it was built from, and a line stating that it is a summary and the signed papers control. When another amendment or a notice arrives, add a row to the index first, then change the deal sheet and move the old value to the last column.