A few months ago, we went through a team call coaching exercise. In the exercise, the buyer was a new GTM analytics leader at a large payroll company, seven days into the job and joining a RevOps function that had only been taking shape for about six months.
The conversation started with familiar enterprise language: data quality, reporting, standardization, predictive decision-making.
Then he said, “We have 10 different definitions of customer and five different definitions of opportunity.”
I probably would have framed the problem as building a trusted data foundation and a more predictive decision engine. Accurate, but thinner. His version explained why the sophisticated stuff is hard: before you build a predictive engine, the company has to agree on what a customer is.
Where the language disappears
In the transcript, the buyer says, “We have 10 different definitions of customer and five different definitions of opportunity.” In the meeting summary, it becomes: current data chaos, inconsistent customer definitions, inconsistent opportunity definitions.
The information survived, but it isn’t the buyer talking anymore. It’s a categorized observation about him.
From there, the sentence gets summarized, the summary becomes an account need and the need gets mapped to a capability. By the time someone writes the brief or campaign, they may be working from the fourth draft of what the buyer said. Nobody is deliberately removing the truth. We’re organizing it until the part that made it feel true disappears.
There’s a practical reason. CRMs need structured fields. Call summaries save time. Account plans need common categories. Marketing needs recurring themes. At some point, somebody has to translate what the customer said into what the customer needs. That’s where specificity dies.
“We have 10 different definitions of customer” becomes “needs data standardization.” The second phrase is searchable and maps neatly to a capability. It’s also something almost any enterprise account could say. The first could only come from somebody living inside that particular mess.
The category is usually right
I’ve watched the same compression happen elsewhere.
At one enterprise market-data company, what could have been recorded as “budget constraint” was: “We have 52 cards in our deck. We can distribute them however we want, but we’re not getting a 53rd card.” The money wasn’t tight. It was already spoken for, and anything new had to displace something else.
At another B2B software company, an integration objection was really a trust objection. A previous vendor had promised, “If you sign the contract, we’ll do a direct integration.” Then: “That didn’t happen.” Calling that an “integration requirement” misses the decision: whether to believe another vendor’s promise.
Customer language is not AEO keyword research
Keyword research can tell me which terms people use. It can’t tell me what those terms mean to the person using them.
Take “visibility.” A buyer at a large technology distributor needed to show executives what she had left and what the next six months should look like. The clean internal label would have been “better visibility.” Her actual language was: “Right now I’m just telling them no, which makes me the bad guy, because I don’t know what I have versus what I need.”
She didn’t need a dashboard. She needed enough certainty to make a plan, defend it to leadership and stop saying no without evidence.
A search tool could tell me that buyers use the word “visibility.” That might help me get found. It still wouldn’t tell me whether one buyer means pipeline transparency and another means defending a number to a CFO. Search language tells me what phrases exist in the market. The surrounding sentence tells me what the person is trying to make happen, what keeps going wrong, and what they’re trying to avoid.
Preserve the sentence next to the interpretation
I’ve become more deliberate about going back to transcripts after calls. My records are better at documenting what my team and I were thinking than at preserving the buyer’s voice. That’s my own weakness, not just everyone else’s.
So when a sentence matters, I keep two things next to each other:
Customer said: “We have 10 different definitions of customer.”
We believe that means: the company needs shared definitions and trusted reporting before it can build the predictive GTM engine it wants.
We have to translate customer language into strategy. Sales needs categories, marketing needs themes, product needs patterns. The mistake is keeping only the translation.
Keep the exact words before you turn them into a Salesforce field, slide or messaging brief. Once that language is gone, it’s hard to recover. If we keep translating the customer without preserving their words, we eventually end up talking to ourselves.
The question I hear most often is: How do we optimize this for AI discovery? It’s the wrong question to end on because it assumes the content already exists. The better question comes earlier: Whose words did this content start from?
Opinions expressed by SmartBrief contributors are their own.
____________________________________
If you liked this article, sign up for the SmartBrief Marketing Pulse, a daily look at the latest in shifts and performance data for marketers.