Designing a signing order that does not stall
Sequential, parallel, or hybrid — how to route a document through approvers and signers without creating a queue that dies when one person goes on holiday.
The most common reason a document sits unsigned for two weeks is not that someone refused. It is that the routing put a single person in the critical path at a moment when they were unavailable, and nothing about the process noticed.
Three patterns
Sequential routes to one recipient at a time, in order. Parallel sends to everyone simultaneously. Hybrid groups recipients into stages: everyone within a stage signs in parallel, stages proceed in sequence.
Hybrid is the right answer far more often than people realise, and it is the one most teams never configure. Internal approvers in stage one, all in parallel; counterparty signatories in stage two, also in parallel. Two stages instead of five sequential steps.
Only sequence what genuinely depends on order
Ask of each step: does this person need to see another person’s input before they can act? If not, they do not need to wait.
- A manager approving before it reaches a customer — genuinely sequential
- Two co-founders both signing a resolution — parallel
- Finance reviewing terms a salesperson entered — sequential
- A witness who must observe a signature — sequential by legal necessity
- Two unrelated departments acknowledging a policy — parallel
Every unnecessary sequential step adds its recipient’s full response time to the total. Four unnecessary steps at two days each is a week and a half of pure queueing.
Approve-only and CC roles
Not everyone in a workflow needs to sign. An approve-only recipient reviews and approves or rejects without applying a signature — right for internal sign-off where the approver is not a party to the agreement.
A CC recipient receives a copy at a chosen point and takes no action. Useful for keeping finance or a shared archive informed without inserting them into the path.
Plan for absence
The failure case is always the same: someone in the chain is on leave and the document is invisible until they return.
- 1Use a shared role address for approval steps where any of several people can act
- 2Set delegation so a recipient can formally reassign — this is recorded in the audit trail
- 3Configure escalation so a step outstanding beyond a threshold notifies a backup
- 4Watch the dashboard for age, not just status — "sent" for eleven days is a problem "sent" does not convey
Delegation is better than sharing an inbox. It preserves attribution: the certificate records that A delegated to B and that B signed, rather than leaving ambiguity about who actually acted.
Handling rejections
Decide in advance what a rejection means. By default, a decline voids the envelope for everyone — correct for a counterparty refusing terms, wrong for an internal approver wanting a small change.
For internal approval steps, enable return-for-changes instead. The document goes back to the sender with a comment, the sender amends, and it re-enters the flow without anyone who already signed having to sign again.
Witnesses and countersignatures
Some documents require a witness, and the requirement is more specific than people assume. In several jurisdictions a witness must observe the signature being made, not merely attest afterwards that it happened. A witness step placed after the signer completes does not satisfy that in every case.
Where witnessing is legally required, check the local rule before designing the flow. Some jurisdictions now expressly permit remote witnessing over video; others do not. This is one of the few areas where getting the routing wrong invalidates the document rather than just slowing it down.
Do not route around a decision that has not been made
A pattern worth naming: teams sometimes add approval steps as a substitute for deciding something. Nobody is sure whether marketing should sign off on customer contracts, so marketing gets added to the chain and the question is never resolved.
The result is a step where the approver does not know what they are approving or what would cause them to reject. Those steps are pure latency. If you cannot state what an approver is checking for and what would make them say no, remove them and have the conversation instead.
Instrument the queue
- Track median time per stage, not just end to end — one stage is usually most of the delay
- Look at the 90th percentile as well as the median; the tail is where deals die
- Count how often documents are voided and resent, which usually signals a preparation problem
- Review the slowest stage quarterly and ask whether it needs to exist
Keep it short
If a workflow has more than four sequential stages, the process usually has a problem that no amount of routing configuration will fix. Every stage is a place a document can stop.
The teams with the fastest turnaround are not the ones with the cleverest workflows. They are the ones who removed approval steps that existed because somebody once asked to be included and nobody revisited it.
David leads operations and customer success. He has onboarded several thousand teams and has opinions about templates.