From FR Advisory.
Every sales methodology ever written starts in the same place. Find the pain.
Discovery questions, pain funnels, the probe, the double click. Thirty years of training material built on one assumption, which is that the seller arrives before the problem has been named, and that the skill is in naming it with them.
That assumption is now wrong, and it is wrong in a way your pipeline cannot see.
This is the first of eight stages, and it is the one you have the least visibility into and the least control over. It is also the one that decides what every later stage is negotiating about. Get skipped here and everything downstream is an argument inside a frame somebody else built.
The stage, defined
Before anything else, the stage needs boundaries, otherwise it cannot be managed or handed over.
What starts it. A constraint becomes conscious inside the customer's business. A cost that has drifted. A team that cannot keep up. A board question nobody can answer. A regulation with a date on it. A person leaving who turned out to be the process.
What happens in it. The problem gets a name. Naming it assigns it a category, and the category decides which kinds of company are candidates at all.
What ends it. The account has a category and a rough candidate set. That is the handover into Stage Two, where the business case gets written.
Who owns it today. Nobody. That is the finding, not an aside. Marketing owns demand capture, sales owns the conversation after capture, and the stage before capture belongs to no function, carries no number, and appears in no review.
What done looks like from your side. You are present, specific and legible at the moment the problem is named. Not persuasive. Present.
How long it lasts. Unobservable from outside. It can be a day or it can be two quarters. You will not know until you ask, which is the point of most of what follows.
The evidence
Semrush surveyed 643 United States business professionals in March and April of this year, 622 valid responses, with 54% of them the final decision maker in their buying group and another 41% part of the deciding group. These are not interns doing desk research.
Of those who use AI at work, 72% use it for early research and category scoping. That is the single highest-usage stage in the entire buying journey. Higher than comparing vendors at 62%, higher than narrowing the shortlist at 48%, higher than supporting the final decision at 45%.
The moment the problem gets a name is now the most AI-assisted moment in the whole process.
And by the time it has a name, it has a category. 92% say AI has shaped their vendor shortlist, 45% of those calling the effect significant. 97% say it helped them discover a vendor they did not previously know.
Then the number that reframes your reporting. After encountering a vendor through AI, 71% then visit that vendor's website. 63% then run a search. 38% then check review sites.
Read that from inside your own systems. Your form captures the website visit. Your reporting calls that visit the start of the relationship and attributes it to organic, or direct, or paid. For most of those buyers the website visit is the second step. The decision about which vendors were worth visiting was made somewhere your systems cannot observe, by something that will never fill in a form.
Your first-touch attribution is measuring the second touch and calling it the beginning.
What the seller actually does
This is where the stage becomes a sales problem rather than a marketing one.
A seller arrives at a qualified lead believing the discovery conversation is where the problem gets defined. It is not. It is where the problem gets confirmed. The buyer has already decided what sort of thing they are buying, roughly what it should cost, and which three names are credible.
So the job changes. It is no longer to uncover the pain. It is to find out how the pain was framed, by whom, and whether the framing favours you.
Stop asking these. What keeps you up at night. What are your biggest challenges. What is your most pressing pain point. Every one of them assumes you are present at naming, and every one of them asks the buyer to repeat work they have already done, in less detail than they did it.
Ask these instead, in roughly this order.
"What happened, and when, that made this a priority now?"
"Who raised it first, and what were they trying to fix?"
"When you started looking into it, what did you call the problem?"
"What did you read, or ask, before you got to us?"
"What other approaches came up, and where did that list come from?"
"Is there anyone in the business who thinks this is not the problem?"
Six questions. None of them clever. All of them aimed at the same thing, which is reconstructing a stage you were not present for.
What to do with the answers. The framing tells you which deal you are actually in.
If the category was named in your language, you are the default and the motion is defending a position. Expect a shorter cycle and a higher win rate, and expect the competition to attack the frame rather than the product.
If it was named in a competitor's language, you are re-framing. That is a different, longer, lower-win-rate motion that most teams run without noticing they are in it. It deserves its own qualification treatment and its own forecast weighting, and treating it as a normal deal is how a quarter goes missing.
If the buyer cannot say who named it, the frame came from a model, and the honest answer is that you are on the shortlist because of what you published, not because of anything anyone did.
The diagnostic
Six questions. Score each one 0, 1 or 2. This takes ten minutes and needs no tool.
1. Can you name the trigger event in your last ten closed-won deals? Not the campaign. The thing that happened inside the customer's business. 0 none, 1 some, 2 all ten.
2. Does your system of record hold the date the problem was first noticed, separately from the date the lead was created? 0 no, 1 sometimes, in notes, 2 a field with a date in it.
3. Do you know what proportion of your pipeline arrives with the category already chosen? 0 never measured, 1 anecdotally, 2 a number you could state out loud.
4. Have you tested what a model returns for your three strongest use cases, described as a customer would describe them rather than as you describe them? 0 no, 1 once, informally, 2 on a schedule, with the results written down.
5. Is anyone's objective tied to being present when the problem is named, rather than to the volume of captured leads? 0 no, 1 someone cares but is measured on something else, 2 yes, and it is written down.
6. Can your sellers tell you, in a live deal, who framed the problem? 0 the question would confuse them, 1 the good ones could, 2 it is in the qualification criteria.
0 to 3. Blind. You are managing the part of the stage you can see and assuming it is the whole thing. Every pipeline conversation you have is about the second half of a process.
4 to 7. Partially sighted. Individual sellers are compensating for a system that does not ask the right questions. That knowledge leaves when they do, and around seventy per cent of software companies lost more go-to-market people last year than the year before.
8 to 12. Instrumented. You can see the stage. The remaining question is whether anyone is accountable for it.
What to instrument
Two new fields, two derived numbers, one renaming, one recurring test.
Retire the idea that first-touch source is the origin of the deal. You do not have to delete the field, and deleting it starts a fight you do not need. Rename it to first observed channel, which is what it has always actually been. That renaming does more work than it looks like, because it ends a whole category of wrong conversation in pipeline reviews.
Add Trigger Date. The date the constraint fired inside the customer's business, captured by the seller in discovery, from the answer to the first question above. It will be approximate. Approximate is fine. It is the only date in your system that belongs to the customer rather than to you.
Add Framing Origin. One field, five values: us, a competitor, an analyst or publication, a model, internal. Captured in discovery, from the third and fifth questions.
Derive Framing Lag. Lead created date minus trigger date. This is the number almost nobody has, and it is the real measure of your top of funnel. It tells you how long you were invisible while the deal was being decided. Track the median, by segment. A lag that grows is a presence problem, and no campaign fixes it.
Derive Default Rate. The share of deals where framing origin is us. That is what a top-of-funnel team should be judged on, because it measures whether you were in the answer, not whether somebody filled in a form afterwards.
Run a presence test every quarter. Describe the problem your product solves exactly as a customer would describe it, not as your positioning describes it, and see what comes back. Do it for your three strongest use cases. Record whether you appear and roughly where. This is measurement, not a growth tactic, and treating it as a growth tactic is how it becomes somebody's side project and then quietly stops.
Stop treating captured-lead volume as a quality signal. It measures the second step. It will keep rising while your framing lag gets worse, and those two numbers moving in opposite directions is the exact shape of this problem.
What changes underneath
The sales consequence is the headline. Four layers have to move before any of it holds, and each one has a specific change rather than a direction of travel.
Go to market. Demand generation is built for a search box and a form. The material that wins the framing moment is published, structured and specific, with the numbers in the body text rather than behind a gate. A gated asset is invisible to the thing now doing the shortlisting. The change: take the three assets that describe your strongest use cases, ungate them, and rewrite their headings as the questions a customer would actually type. Not your product names. The problem, in their words.
Commercial systems. The two fields, the two derived numbers and the renaming, as above. Then the harder part, which is that the opening question of every pipeline review changes from where did this come from to when did this start and who named it. The field changes take a fortnight. The review habit takes a quarter.
Sales strategy and enablement. The discovery script is replaced, not supplemented. Then qualification gains a framing test, so re-framing deals carry a different stage gate and a different expected win rate from defending deals. Most forecasts today average those two together and produce a number that is wrong in both directions.
Compensation. Lightly, because this stage is not where compensation does its heaviest work. If the top of the funnel is paid on captured leads, it will manufacture captured leads. Being present when the problem is named is harder to count than a conversion, which is why it is not paid for, which is why nobody owns it. You do not have to solve that this quarter. You do have to stop calling the absence of ownership an execution problem.
None of those four are technology decisions. All four are commercial operating model decisions, which is the argument this entire series rests on.
The first ninety days
Sequenced, because doing these in the wrong order is how the whole thing stalls.
Week one. Run the presence test on three use cases. It costs nothing, takes an afternoon, and it is the only item on this list that tells you where you stand before you have changed anything.
Weeks two to four. Add the two fields. Brief the sellers on the six questions and have them use them in every discovery call. Change no compensation and buy no software.
Weeks five to eight. Backfill trigger date on the last twenty closed deals from the sellers' memory. It will be crude. It is still the first framing lag number your business has ever had, and a crude number you can act on beats a precise one you do not have.
Weeks nine to twelve. Ungate and rewrite the three assets. Run the first pipeline review that opens on trigger and framing instead of source. Expect it to be uncomfortable, because the first honest answer to who named this is usually that nobody knows.
What not to do in the first ninety days. Do not rebuild attribution. Do not buy a tool for this. Do not change compensation. All three are the expensive version of a problem you have not measured yet, and all three are how this becomes a project instead of a change.
The position
The first stage of the customer's process is the one you have the least visibility into and the least control over, and it is now the stage most shaped by AI.
Seventy two per cent scope the category with a model. Ninety two per cent let it shape the shortlist. Seventy one per cent only reach your website afterwards, which means the moment your systems call the beginning is already the middle.
Absence is not neutral. It generates no record, no rejection and no feedback. You will never see the deals you were not in.
You cannot generate the trigger. You can only be the answer when it fires.
Everything after this stage is a negotiation inside a frame somebody else built.
Next: Stage Two, Frames. The business case gets written in a room you are not in, by someone whose name is not on your opportunity record.
The Agentic Commercial Model. agenticcommercialmodel.com
Sources
- Semrush, "How AI tools shape the B2B buying process", survey of 643 United States B2B professionals in March and April 2026, 622 valid responses, findings based on the 519 who use AI at work: 72% use AI for early research and category scoping, 62% for comparing vendors, 48% for narrowing the shortlist, 45% for supporting the final decision; 92% say AI has shaped their vendor shortlist with 45% calling the impact significant; 97% say AI helped them discover new vendors; 83% say it influenced their final decision; 66% have noticed vendors missing from AI results; after an AI mention 71% visit the vendor's website, 63% search, 38% check review sites; 54% of respondents are final decision makers, 41% sit in a decision-making group
- Teneo, 2026 B2B Software Vendor Survey, 300 United States software vendors, fieldwork 23 December 2025 to 22 January 2026: around 70% reported higher go-to-market turnover year on year, with four in five saying it limited new-logo attainment. The report does not state its sampling method, weighting or margin of error
- A separate G2 figure putting AI-first vendor research at 51% was reviewed and excluded from this piece, as the only available reporting of it was secondhand with no stated sample size or publication date