The half of the deal that happens after you send
Document tools assume you draft and send. Real agreements come back marked up in Word, and that round trip used to happen entirely outside the product.
The XDocs team4 min read
Every document product has a clean story about drafting. You write the agreement, you send it, it comes back signed. The demo is lovely.
What actually happens is that it comes back as Acme_NDA_v3_REVISED_jl_clean.docx attached to an email that says "minor comments, nothing controversial". Somebody opens it in Word, squints at the tracked changes, and forms an opinion. That opinion is where the risk lives, and for most teams it is formed in a tool the agreement does not live in, with no record of how it was reached.
This was the biggest hole in XDocs for a long time, and it was a hole of exactly that shape: we had the drafting and we had the signing, and the part where the deal is actually decided happened in the gap between them.
Start by reconciling, not by reading
Attach the file they returned and it is reconciled against what we sent — not against the current draft, which may have moved on, and not against nothing at all. Their version is merged into the document as tracked suggestions in their name, so what you review is a redline in the editor you already use, with their edits attributed and coloured, acceptable or rejectable one at a time.
Both bodies are snapshotted to version history before the merge runs. A merge you cannot back out of is not a merge, it is a replacement with extra steps.
Then turn the differences into rows
Every difference becomes a record: the clause, the text before, the text after. That shape is what makes the rest possible, because a list of rows can be sorted, triaged, counted across rounds and reported on. A page of strikethroughs can only be read.
| Triage band | What it means | What happens next |
|---|---|---|
| Cosmetic | Wording, formatting, defined-term tidying. No change to who owes what. | Accept in a batch and stop thinking about it. |
| Material | A real change to an obligation, a cap, a term length, a notice period. | Read it against the playbook, decide, and counter if it strays. |
| Dangerous | Exposure that was not there before — a cap removed, an indemnity widened, a liability carve-out deleted. | Surfaced first, with the exposure named and a reason attached. |
The band that earns its keep is the third one, and the reason is the covering note. Each deletion is read against what the other side *said* they changed, so the clause that was quietly removed and not mentioned is flagged rather than buried among the ones that were. The most expensive edit in a returned draft is routinely the one nobody drew attention to.
"Minor comments" is a claim about a document, and it is checkable.
Counter in clauses, not in notes
When a change has to be pushed back on, the counter-language goes into the document as a suggestion of our own: the clause in full, ready to accept, rather than a comment saying what the clause ought to say. A note describing a redraft leaves the redraft still to be written, usually by whoever is most tired.
Positions are written down once
A workspace keeps a playbook: standing positions, each with the fallback you can live with. *We never accept uncapped liability; our fallback is 12 months of fees.* Those positions are quoted into every turn, so a rule stated once is held to without being restated in each conversation — and more importantly, so a rule the team agreed in March is still being applied in September by somebody who was not in the room.
From that, two reports fall out. The deviation report ranks a returned draft by what each change would actually cost, rather than by where it happens to appear in the document — page 14 is not less important than page 2. And the concession ledger records what each side has given, round by round, which is the only reliable defence against paying twice for the same point in a long negotiation.
The reply is assembled, not remembered
The response that goes back is built from the change records: what we accepted, what we rejected and why, what we have countered with. Not from anyone's recollection of a thread that is now sixty messages long. Our current position travels with it as a Word file, because that is what the other side will open.
Sending it is the one action in this whole flow that needs a click. Everything before it — merging, triaging, drafting counter-language, updating the ledger — happens inside the workspace where it can be reviewed or undone.
And then a clean copy
When everything is settled, the agreed redline becomes a clean, signature-ready draft, with a change log filed against it as a comment. The marked-up original is left exactly as it is: it is the record of how you got here, and overwriting it to tidy up would destroy the only account of the negotiation at the moment it finally became valuable.
None of this makes the other side agree with you. It does mean that by the time you form an opinion about their draft, you are reading a list of what changed and what it costs — in the document itself, in the tool that will eventually collect the signature.
Read next
Why we freeze the document at send
The moment an agreement goes out for signature it stops being one document and becomes two. Most tools hide that. We would rather say it out loud.
Read itAn assistant you can hold to account
Letting software act on your agreements is a trust problem, not a model problem. Nine things we built before we let it send anything.
Read itWhat is a master services agreement, and when do you need an SOW?
An MSA settles the legal terms once; a statement of work describes each job. Splitting them is how agencies and their clients stop renegotiating liability every time the scope changes.
Read it
One place for every important document.
Create, sign, share and understand your documents with XDocs.

