The moment where deals quietly die
Listen to the middle third of any recommendation. That's where you find out if your rep is selling or reciting.
Here's what reciting sounds like on a roofing call:
"So the package includes synthetic underlayment, ice and water shield in the valleys, ridge vent, a lifetime workmanship warranty, and we're a certified installer so you get the manufacturer's 50-year non-prorated coverage."
Nothing in there is wrong. It's all true and it's all valuable. But the homeowner told this rep, twelve minutes earlier, that her ceiling stained after the February thaw and her husband is convinced they need a whole new roof while she thinks it's one spot. She said the word "thaw." She said "argue about it every winter."
The rep never touched either one.
That's feature dumping. Not a wall of jargon — that's the cartoon version. Real feature dumping is a competent, accurate, well-organized recommendation that happens to be identical to the one the rep gave the last four people. It's a monologue with the buyer's name on the outside of it.
Why "did they sell value?" is unscoreable
Most managers already know feature dumping is bad. The problem is that "value selling" as a coaching note goes nowhere. A rep hears "connect it to value" and thinks I did — I said the warranty saves them money.
So the note has to get mechanical. Here's the definition I'd put on a sales call scorecard:
A recommendation is connected when the rep names something the buyer said, in the buyer's own words or close to them, and links a specific part of the offer to it.
Two parts. Buyer's stated thing. Specific part of the offer. Both, in proximity, out loud.
That's it. Now it's binary enough to score and specific enough to coach.
What it sounds like when it's there:
"You mentioned the stain showed up after the February thaw and that's actually the tell — that's ice damming, not roof failure. So the piece I'd point you at first is the ice and water shield in the valleys. That's the part that ends the argument you two have been having every winter, and it doesn't require replacing the whole roof to do."
Same product. Same features. She now knows he was listening, and she has a sentence she can repeat to her husband tonight.
The three-level rubric
Don't score this pass/fail. Reps land in the middle constantly and a binary score hides the coaching.
Level 0 — Generic. The recommendation would be word-for-word identical for any buyer. No reference to anything discovered on the call.
Level 1 — Labeled. The rep references the buyer's situation generically or restates a category. "Since you're worried about cost..." or "For a busy family like yours..." This is the most common score and the most misleading one, because it feels personalized to the rep. But "you're worried about cost" is what everyone says. It's a label, not a quote.
Level 2 — Connected. The rep uses the buyer's specific stated priority — the actual detail, the actual phrase — and maps a specific element of the recommendation to it. Bonus if the rep also says which parts don't matter for this buyer.
That last bit is the real tell. A Level 2 recommendation is usually shorter than a Level 0 one, because the rep is subtracting. When a rep says "honestly, the extended labor coverage probably isn't the piece to focus on for you — you said you're selling in three years," they've proved they were listening in a way no amount of adding can.
Scoring it in practice
Pull the recommendation segment. Ask three questions:
- What did the buyer say they cared about? Write it down in their words.
- Did the rep say that thing back, specifically, during the recommendation?
- Did the rep attach a named part of the offer to it?
If you can't answer question 1 from the recording, that's not a recommendation problem. That's a discovery problem, and you're coaching the wrong half of the call.
That happens more than people expect. Maybe a third of the feature dumps I've seen aren't recommendation failures at all — the rep never got anything to connect to. They asked "what are you looking for today" and got "just pricing it out" and moved on. You cannot connect a recommendation to nothing.
The two-sentence drill
Here's the coaching rep. It takes four minutes and it works on the next call.
Before the rep gives any recommendation, they have to be able to fill in this out loud:
"You said ______. So the part of this that matters most for you is ______."
That's it. Two blanks. First blank must be a near-quote. Second blank must be a specific named element, not "our quality" or "our team."
Run it as a live drill in your one-on-one. You play the buyer, you give them three stated priorities in a two-minute roleplay, then you stop and make them fill in the sentence. Most reps get the second blank instantly and fumble the first. That tells you they're pattern-matching to a product pitch instead of retrieving what the buyer said.
For insurance:
"You said the reason you're shopping is the renewal jumped $340 and nothing changed on your end. So the part of this that matters most is the multi-policy structure — that's where the $340 actually comes back, not in the coverage tweaks."
For B2B inside sales:
"You said your ops lead is spending Friday afternoons rebuilding the same report by hand. So the piece to look at first is the scheduled export — that's the four hours."
For a dealership:
"You said your last lease you went over mileage twice and got hit for it. So the number that matters here isn't the monthly, it's the mileage allowance, and that's what I'd change."
Notice all three name a number or a concrete detail the buyer supplied. That's what makes it un-fakeable.
What to do with the scores
Two patterns show up when you start tracking this across a team.
The rep who scores Level 1 constantly. They're not lazy, they're fast. They hear a category ("price-sensitive," "wants it done before the holidays") and jump to the pitch that fits the category. Coach the retrieval, not the empathy. Make them write down three verbatim buyer phrases during discovery — literally on paper — and reference at least one in the recommendation. Physical note-taking fixes this faster than any script.
The rep who scores Level 2 on discovery-heavy calls and Level 0 on quick ones. They can do it, they just skip it under time pressure. That's a habit problem, and the two-sentence drill handles it because it's short enough to survive a rushed call.
One more thing worth watching: Level 2 recommendations tend to produce different objections. Instead of "let me think about it," you get "okay but does the ice shield cover the back valley too?" That's a better objection. It means they're evaluating the thing you actually recommended rather than politely exiting a presentation.
If you're reviewing calls by hand, tag the recommendation segment and score just those ninety seconds — you don't need the whole call. If you're using something that scores calls against a rubric automatically, this is a clean item to add, because the two-part definition holds up: buyer's stated words, named offer element, in the same breath.
Either way, the point isn't to grade your reps into compliance. It's that once you name the specific mechanic, "sell value, don't feature dump" turns from a slogan into something a rep can actually go do on their eleven o'clock.