From Reactive to Proactive
The deal process between the demo and the signature — what breaks in it, and the four parts that make it a system instead of a scramble.
Start with the part nobody puts in a business case.
Your people are stressed. They are overworked, and a good share of what they are working on is not the work they were hired to do. A rep is chasing a security answer they are not qualified to give. A solutions engineer is rewriting a response they have written four times this year. Somebody senior is being asked to approve a discount at 9pm because the buyer’s board meets Monday.
None of them feel equipped to fix the thing that is making their week hard. Most of them have stopped mentioning it.
That is what a reactive commercial process actually feels like from the inside. Not a dashboard trending down. People absorbing variance with their evenings.
And it reaches the customer. Every disparate system and undefined process shows up in the quality of the interaction a buyer has with you — the answer that took nine days, the second person asking a question they already answered, the quote that does not match the proposal. Your buyer cannot see your internal process. They can see the output of it, and they are drawing conclusions.
The stretch where it happens
Most sales advice points at the front of the deal. Outreach, discovery, the demo. That is not where winnable deals are dying.
They die after the demo. In the RFP that starts from a blank page. In the security questionnaire that sits for three weeks. In the procurement committee you were never invited to. In the approval that took four days while the buyer’s quarter closed on day three.
Won in the demo. Lost in the evaluation.
That stretch is not a gap between stages. It is a system, and at most scaling companies it is the only part of the commercial process that was never designed — it accreted. Which is why it runs reactively: too many variables, no clear plan, and a different answer depending on who happens to be free.
The reactive version has a cost you can count.
Source: Loopio, 2026
Roughly two-fifths of the money moves through the part of the process nobody owns.
The proactive version is not more effort. It is the same effort, spent before the fire instead of during it.
The four parts
Everything between is this worth pursuing? and it is signed and ready to deliver runs on four linked capabilities, with a fifth thing underneath them that is not a stage at all.
What broken looks like, part by part:
- Pursue — you respond to everything. Nobody owns the decision not to bid.
- Respond — every RFP starts from scratch. Answers drift. Nobody owns the questionnaire.
- Document — scope is vague, quotes vary by rep, deals stall after the verbal yes.
- Orchestrate — approvals live in DMs. The forecast is rep optimism with dates on it.
And underneath all four: Measure. Not a fifth stage — the baseline. It is how you know a change worked rather than believing it did. Without it every improvement is an anecdote, and you will spend a year fixing the loudest problem instead of the most expensive one.
A weakness in any one part leaks into the others. That is the argument for treating it as a system rather than four projects.
Pursue — the deal you should never have chased
The most expensive deals are not the ones you lose at the end. They are the ones you should never have started.
Bandwidth is now the number one constraint proposal teams report — for the first time on record, with about half ranking it a top challenge (Loopio, 2026). The instinct is to read that as a staffing problem. It is usually a qualification problem wearing a bandwidth costume.
Run the math. At ~166 RFPs a year against an average win rate near 45% (Loopio, 2026), a typical team loses somewhere around ninety responses annually. The question is not how to answer faster. It is how many of those ninety were knowable at intake.
In my experience, at least a quarter of them. That is my estimate from running and auditing this work, not a research finding — but a quarter of ninety is more than twenty serious responses a year, each consuming senior hours, that a governed decision would have declined.
A real pursuit decision weighs seven things, in two groups.
Can we win it?
Fit
Not could we serve them, but do we win here. Name the last three deals you won that look like this one. If you cannot, you are hopeful rather than qualified.
Access
A named champion who will act, a route to the person who signs, and permission to ask questions before the deadline. Responding into a portal with no human on the other side is not an obstacle to overcome. It is an answer.
Timing
Is something on their side forcing a decision, and does someone own that outcome? Or are you the third quote that makes an incumbent’s renewal look competitive?
Incumbency
If someone is already inside, can you name the specific failure the buyer will say out loud, unprompted? “We’re better” is a hope with a slide deck.
Do we want it?
Cost to compete
What a serious response costs in senior hours, and whose hours they are. Every yes here is a no somewhere else, and the somewhere else is often a deal you would have won.
Complexity
What delivering this will actually take. A deal that wins and then costs double to deliver is a loss with a longer settlement period.
Growth potential
Is this a ceiling or a floor? A smaller deal that expands beats a larger one that terminates, and almost no bid/no-bid process weighs it.
Most frameworks stop at the first group. The second group is where operators separate from templates.
More on this: You’re not bidding. You’re donating.
Respond — surviving someone else’s evaluation
Once a deal is worth pursuing, it stops being about your pitch and becomes about their evaluation. The RFP, the security questionnaire, the vendor-risk review.
Here is the state of play.
Source: Loopio, 2026
The tools are in the building. Research published this year is titled, plainly, “AI Adoption Is No Longer the Advantage — Execution Is” (Responsive, 2026).
That is the whole argument for the layer underneath. Adoption is not a differentiator when four in five of your competitors have the same thing.
What actually breaks:
- No source of truth. Type two words into most response libraries and you get several results that do not agree. Each was written once, under a deadline, for one buyer’s phrasing. Nothing marks which is canonical.
- No owner. The security review sits between sales, security and IT, and each is correct that it is not theirs. It is one of the longest stages in enterprise procurement, and at most companies nobody owns it end to end.
- No expiry. A fact changes on a Tuesday. The answer changes whenever someone notices. The distance between those two dates is your real exposure, and almost nobody measures it.
- No consistency check. Two buyers ask the same question in one quarter and get two different answers. Both sent, both on the record, neither flagged.
Speed is the output of fixing those, not a substitute for it. A team answering faster on an ungoverned library ships unverified answers on a shorter timeline.
More on this: Nobody owns the security questionnaire · Every answer you send has an expiry date · Two buyers. Same question. Two answers.
Document — making the case and getting to signature
This is the least-discussed part of the process and, in my experience, where the most avoidable damage happens. The deal is effectively won. Then it stalls, and the reasons are unglamorous.
What I have watched go wrong, repeatedly:
- Scope described imprecisely. Not wrong — vague. Two readers form two different pictures of what is being bought, and neither finds out until delivery.
- Service specificity missing. What is included, what is not, what is optional, at what point extra work becomes a change. Left implicit at proposal stage, it becomes an argument later.
- Redlined clauses arriving late, on terms nobody on the commercial side had read until after the verbal yes.
- Partnership-level language living only in the master agreement, never carried down into the project-level documents that people actually work from — so what was agreed at the top never reaches the work.
To be exact about my own boundary: contracts and redlines are not what I do. That is a lawyer’s job and I will not pretend otherwise. But three of those four failures are commercial-side failures with commercial-side fixes, and the fourth is a handoff problem dressed as a legal one.
What I have built to close them:
- Templates, so scope and service definition are written the same way every time and the gaps are visible before a buyer finds them.
- Integrated systems, so the proposal, the quote and the scoping document draw from the same source instead of three people’s understanding.
- Communication methods and milestones, agreed while the deal is open — a named cadence, named participants, and dates the buyer can hold you to — so the move from proposal to signature to scoped delivery follows a path rather than someone’s memory.
None of that is exotic. It is the difference between a back half that holds and one that runs on whoever remembers what was promised.
More on this: What the delivery team inherits
Orchestrate — approvals, visibility, and a forecast that is real
Orchestration is the connective tissue running through the other three. It decides whether decisions get made before the fire or during it.
Approvals without a record. A discount or an exception gets approved in a thread, and three months later nobody can reconstruct who approved what or why. In a regulated or audited sale that is a real exposure. And the approval that takes four days because it is bouncing between inboxes is a timing killer — the buyer’s quarter closes whether or not your chain is ready.
The fix is tiered authority, written down: who can approve what, at what threshold, with what recorded reason. Real bands are wider than people expect — in enterprise software I have seen tiers run 10–25%, 25–50%, and above 50%, each with a different owner and a different level of scrutiny. The point is that the routine ones stop being requests at all, and only genuine exceptions reach a senior inbox.
No shared view of the deal. Sales knows one thing, the deal desk another, delivery finds out at kickoff. This gets worse as buying groups grow: healthcare IT purchases commonly run 12–18 months with roughly 8–10 stakeholders (Salesmotion / MarketBetter). At that size, alignment is not a meeting. It is an artifact everyone reads.
A forecast built on optimism. If the forecast is a roll-up of what reps hope will close, it is a wish list with dates. Confidence comes from a consistent read on where each deal sits in a defined process.
More on this: The approval that took four days · You added a human. You didn’t add a standard.
Measure — the baseline underneath
Measure is not a fifth stage. It is the benchmark that runs across all four, and it exists to answer one question: did the change work?
Take a reading before you change anything. Response cycle time. Share of responses assembled from governed content versus written fresh. Time an approval waits on you rather than on the buyer. Share of pursuits declined at intake. Then take it again.
Without a baseline, every improvement is a story someone tells. With one, you find out that the thing everybody complained about was the third most expensive problem, and the quiet one was the first.
The layer that makes it safe
The tools most teams reach for — AI answer generators, response software, CPQ — are not the problem and they are not the fix. Four in five proposal teams already have the AI. Adoption stopped being an advantage.
What is missing is the governed system underneath it: the human-designed standard that decides what good is.
AI does the volume. A person who ran the motion owns the judgment.
Every automated output runs through a rubric, a governance rule, or a review pass, and the regulated, high-stakes decisions stay with a person and a documented standard.
Regulators are converging on AI accountability in adjacent disciplines — the FDA published draft guidance on AI supporting regulatory decision-making in January 2025, and the EMA and FDA issued joint guiding principles in January 2026. The commercial side of the same company usually has none of that attention. Worth saying plainly: this is context for why buyers now ask, not a claim about my own scope. I govern the answer, not the model. No GxP, no computer-system validation — a different discipline and a different vendor.
More on this: Same discipline. Different room.
And a commitment, in both directions. My goal is to install the system and hand over the keys — documentation as a deliverable, a named internal owner on your side, and success measured by how little you need me a year later.
That only works if it runs both ways. You have to commit to the change management and to sticking with the plan. A governed process that gets abandoned in week four when a large deal asks for an exception is not a governed process. It is a document. The install is the easy half; holding the standard when it is inconvenient is the half that decides whether any of this survives.
Where does yours leak?
Most teams have one or two clear leaks, not four. Fixing the worst one first tends to move the most revenue.
A useful self-check, one question per stage:
Could you produce a written reason you chose to pursue your last five deals?
Pick the answer you send most often. When was it last verified, and by whom?
Does your scope language say what is not included?
Pull your largest discount last quarter. Can you reconstruct who approved it and why, from a record rather than a memory?
Do you have a baseline for any of the above?
If those took longer than a minute each, the gap is worth measuring rather than guessing at. The Content Readiness Assessment scores your response process in about five minutes, free, and returns your highest-leverage fixes.
Rob Roberts is the founder of Widenine Solutions. In B2B SaaS and life sciences since 2011, running Deal Desk and Sales Operations since 2018 — including a Deal Desk built from scratch governing a $25M+ pipeline, and a 2,000+ vetted Q&A library behind 5,000+ responses built and audited.