Insight Selling: Evidence, Tests and First-Message Examples

Insight selling means opening a sale with a specific, checkable idea about the buyer's business, one that changes how they see a problem they had missed or underrated. You earn the meeting by telling buyers something true about their own company or market before you say a word about your product. The search "insights sales" also catches a second, unrelated thing: sales insights software inside a CRM.

I cover both, but the method gets most of the space, because you can use it tomorrow without buying a license.

First, where I stand. My team works mainly by cold email and LinkedIn. We book the meetings, and the client's own salespeople run those calls and close the deals. For us insight selling lives in one place: the first message to a stranger. Every buyer statistic below comes from a named study, linked where it first appears.

What is insight selling, and is it the same as sales insights?

Sales insights is a different thing. The term covers software that analyzes your own sales data: CRM records, emails and recorded calls. Microsoft's Dynamics 365 is a clear example. Its Sales Insights package has standard features (an assistant, auto capture and email engagement) and premium ones, including conversation intelligence, predictive lead and opportunity scoring, relationship analytics and premium forecasting (Microsoft Learn).

The same page says the premium features are not switched on by default; an admin has to enable them. Conversation intelligence gives managers the sentiment, keywords and competitor mentions from calls, so they can coach their sellers.

The query mixes the two because the words overlap and both promise smarter selling. The software can feed the method. For example, your call notes might show which problem keeps coming up in deals you won. But a dashboard never puts an idea in front of a buyer. Someone still has to write the sentence.

Why did sellers move from solutions to insights?

In the July-August 2012 issue of Harvard Business Review, three authors, Adamson, Dixon and Toman, argued that solution selling was losing its grip (HBR). The old model worked because customers often understood their problems but did not know how to solve them. A rep who diagnosed needs and bundled products and services into a "solution" was useful.

By 2012 that had changed. Procurement teams and purchasing consultants had so much data that companies could define the solution themselves, and they pushed suppliers into a contest decided mainly on price.

The article's abstract describes what the best reps did instead (Harvard Business Publishing):

• They went after organizations where demand was still forming.
• They arrived early with bold suggestions for the customer's next move.
• They looked for skeptical change agents inside the customer.
• They taught those skeptics how to go about buying.

The three authors were directors at Corporate Executive Board. Their full data sits behind a paywall, so none of its numbers appear here.

The link to today is short. If buyers can write their own spec, a seller who shows up asking what they need adds very little. The opening has to bring something the buyer does not already have.

What does the evidence for insight selling actually show?

Two sources carry most of the weight here. Here is what each one is, and what it can and cannot carry.

The Gartner numbers are the more interesting ones. According to the survey, 69% of B2B buyers would rather have sales reps confirm what AI told them (Gartner). Across one recent purchase, they had drawn on seven information sources on average. 45% used GenAI along the way.

The same survey also found that 67% of buyers prefer a buying experience without a rep, and 70% prefer one that is completely digital and self-service. Asked where they were more likely to run into misleading information, 51% said GenAI and 49% said a sales rep. That is close to a coin flip.

Read those figures as stated preferences: the release reports what buyers say they want, not what they went on to do. Taken together, they describe someone who does most of the research alone and does not fully trust either the chatbot or the salesperson.

Gartner's own advice is for reps to move from being the main source of information to being a source of validation and confidence. In practice, the seller worth a meeting checks and sharpens what the buyer already found, then adds one thing the other seven sources missed.

A seller who repeats what a chatbot would have said hands the buyer a reason to pick the rep-free route. Why take the call?

One gap matters before you trust any claim. The public pages of these two sources do not show what insight selling does to win rates. Treat any percentage lift quoted for insight selling as unproven unless it names its study and sample.

How do you know if what you have is really an insight?

Run every candidate sentence through these five questions before it goes into a message.

1. Is it specific to this buyer's type of company? Swap in a company from an unrelated industry. If the sentence still works unchanged, it fails.
2. Can the buyer check it? They should be able to count it, ask their team, or look it up in their own records.
3. Is it likely news to them? If anyone in their role would already know it, it fails.
4. Does it carry a cost of doing nothing? The buyer should see what staying put is costing them in time, money or risk.
5. Does it stay clear of your product? The moment it names what you sell, it has become a pitch.

A sentence that fails two or more of these is a statistic or a pitch, so cut it or rework it.

Three fakes turn up often:

• The generic industry statistic. "Most companies in your sector struggle with reporting." It fails questions 1 and 3, and the buyer has seen it in other inboxes already.
• The compliment. "Loved your post about the new office." It is checkable, but it is not news and carries no cost, so it fails questions 3 and 4.
• The disguised pitch. "Firms like yours lose hours on admin that our platform removes." It fails question 5 outright, and usually question 2.

You do not need a research department to find real insights. They come from your own customers. When I start a campaign, I look at the client's current and past customers first, and only then search for companies that resemble them.

I also go back through past projects to see who really signed off on each one, because that person's problem is what the opening line should describe. Then target that exact title in the list, not a generic "manager".

A pattern you have seen at three companies of the same type is specific and checkable, and the fourth company may not know it applies to them.

Insight, solution or consultative selling: which fits your deal?

All three methods overlap in practice. The table shows which one should lead, depending on where the buyer is right now.

The rule that falls out of the table: if the buyer has already written the spec, an insight arrives too late. Compete on fit and proof instead. The solution selling guide covers that case, and consultative selling covers the buyer who knows something is off.

If the buyer has not noticed the problem, discovery questions alone will stall, because nobody answers questions about a problem they cannot see.

The opening carries extra weight because the decision mostly happens before a call is booked. 6sense's 2025 Buyer Experience Report found that buyers now contact sellers about 61% of the way through their journey, down from about 69% in 2024. It also reports that when buyers finish selecting, the vendor they rank first wins roughly 80% of the time.

Nobody buys from an email. They buy after they have checked who sent it, and the first line decides whether they bother to check.

What does an insight look like in a first message to a stranger?

In the first messages my team writes, we sell a conversation and leave the service for later. Conversations come first. Calls come out of conversations. Clients come out of calls. Pitch in message one and the stranger stops reading.

Proof goes inside the message itself: past clients and what changed for them, written into the lines where a credentials paragraph would usually sit.

The rewrites below are invented for illustration. The sender is a software development company that builds field tools for civil engineering contractors. No reply rates are claimed for any of these lines.

A researched line has nothing in common with a filled-in merge field. Dropping a first name and a company name into a template proves nothing about the buyer, and the buyer can tell within a sentence.

Here is a full first message built from those lines. It is invented for illustration too.

Subject: site diaries typed twice

Hi Martin, the last two civil contractors we built field tools for found the same thing. Site diaries were written up on site, then typed again in the office by someone else. Neither company had counted the office hours the second round took. If your project managers spend Friday evenings closing out the week, the same double entry may be sitting there. Worth a 15-minute call to compare notes with what those two found? No demo, no deck.

Run it through the test. It is specific to civil contractors, Martin can check it with one question to his office, it is probably news to him, it points at a cost (the office hours spent on the second round), and it never names a product.

Where does insight selling stop working?

There are four places where no opening line can save you.

The wrong list. When a campaign goes quiet, I ask to see the list before I look at the email. The message gets blamed because it is the part everyone can see and edit, while the list usually causes the silence. Good copy cannot rescue a poor list or an unreliable sender.

The same message can pull 2% replies for one company and 7.5% for another. A sharp line about site diaries sent to office coordinators at companies that never run site work is still a wasted send. For more on this, read why the problem is usually the list.

Weak foundations. Carefully personalized messages still get ignored when the foundations are broken: a sending domain that lands in spam, or a profile that looks empty when the buyer checks it. A so-so line from a profile full of proof will outpull a brilliant line from a ghost. Fix the foundations before you polish any sentence. In one of our projects, doing only that lifted the reply rate by 571% in 60 days, with no change to the message.

A closing problem after the call. Booking a meeting and winning the deal are separate jobs. Whoever takes the call needs to know the pitfalls of that conversation, how this buyer makes decisions, and where selling ends and closing begins. If closing is already weak, adding more calls only makes it weaker.

An insight-led opening needs an insight-led call behind it, run by the client's own team. For why, read why a booked call still needs a closer.

Buyers who are past the point of being taught. Some buyers are deep into their own research, with a shortlist already drawn up. Others are making small, routine purchases where the research would cost more than the deal is worth. In both cases, fit, proof and a fast answer serve them better than a provocative idea.

Before your next sequence goes out, take its first line and run the five questions on it. If it fails two, replace it with one pattern from your last three clients. If you would rather have a team run the outbound side, here is how our cold email agency work is set up.

Want to turn one real insight about your buyers into a first message? Book a call and bring a draft. I will tell you whether it reads as an insight or a pitch.