By the time someone fills in a form, the shortlist usually exists. Most of the work that built it happened somewhere your reporting cannot see.

In short

• Considered purchases are researched by a group, mostly independently, long before any supplier is contacted.

• Only a small fraction of the journey involves suppliers at all, and that fraction is split between everyone shortlisted.

• Much of the research happens on surfaces that pass no data back: communities, peer conversations, messaging apps and summaries.

• The response is not more tracking. It is being present and accurate where the research happens, and asking buyers directly how they found you.

There is a moment in most sales conversations that should worry marketing more than it does. The prospect already knows the pricing model, has opinions about two competitors, and mentions a comparison they read somewhere they cannot quite remember.

Everything in that sentence happened before the first contact, and almost none of it appears in any marketing report.

What the research says about the journey

Gartner's work on complex purchasing has produced a set of figures that are worth keeping visible in any planning session.

Divide that 17% across a shortlist of three and a single supplier gets somewhere around 5% of the buyer's total attention. Everything else is research, internal discussion and the slow work of building agreement among people who do not all want the same thing.

The third figure is the one most teams skip past. Buyers do not describe these purchases as exciting. They describe them as difficult. That is a significant clue about what content actually helps, and it is not more persuasion.

Where the research actually happens

Ask any buyer to reconstruct their process and the same set of surfaces appears, in no consistent order.

People they already trust

The first move in almost every considered purchase is asking someone who has done it before. A former colleague, a peer in another company, a group chat, an industry community. This is the single most influential input and the one marketing has least control over, which is precisely why it is usually ignored in planning.

Public discussion by strangers

When there is nobody to ask, buyers look for people arguing about it in public. Community threads, forums, question sites, comment sections under reviews. The appeal is not accuracy. It is that the brand is not in the room, so the assessment feels unmanaged.

This matters more than it used to for a measurable reason. When Pew analysed which sources AI-generated summaries most frequently cited, community and reference sites featured prominently alongside video. What a public thread says about a category increasingly becomes what an automated answer says about it too.

Independent comparison

Review platforms, category listings, roundups and "alternatives to" pages. These often rank for a company's own brand name, which means a buyer searching for you may read a third party's summary of you before reaching anything you wrote.

Video and long-form demonstration

For anything with a learning curve, buyers want to watch somebody use it before they will believe a description of it. A walkthrough recorded by a customer frequently outperforms anything produced by the company, because it carries no incentive to flatter.

The supplier's own site, late

Your website is usually consulted to confirm specifics rather than to be persuaded: does it integrate, what does it cost, is it available in this market, how long does implementation take. By this point the buyer is checking, not exploring.

Most marketing budgets are concentrated on the last surface in that list, which is the one the buyer visits last and trusts least.

Why this is invisible in reporting

Almost every surface above passes no useful data back. A link shared in a private message arrives with no referrer. A recommendation given verbally leaves no trace at all. A community thread read on a phone and acted on a week later from a work laptop appears as two unrelated sessions, if it appears at all.

The result is a reporting picture that systematically over-credits whatever happened last and under-credits everything that created the interest. Teams then optimise towards the over-credited channels, which usually means bidding harder on branded search terms and taking credit for demand that something else created.

What to do about it

The instinct is to track harder. That instinct is wrong, because the surfaces are structurally untrackable. The productive response is different.

1. Ask buyers directly

Add one optional open question to your enquiry form: how did you first hear about us? Then read the answers rather than charting them.

Within a quarter you will have a list of podcasts, communities, individuals, events and publications that your attribution model has never once mentioned. That list is a media plan written by your own customers.

2. Be present where the research happens, honestly

Being present in a community means answering questions as a named person who discloses where they work, not posting recommendations under a pseudonym. The second approach is discovered eventually, and the damage lasts longer than the benefit.

The realistic goals are narrower and more achievable: know what is being said about you, correct factual errors politely, and make sure someone knowledgeable from your company is genuinely useful in the places your buyers gather.

3. Fix what third parties say before writing more of your own content

Your profile on a review platform, your entry in a category listing, your description in an industry directory. These rank, they get cited, and they are frequently years out of date. Updating them is unglamorous, takes a day or two, and often shifts more than a month of new content production.

4. Build for the difficult part, not the exciting part

Buyers said these purchases are difficult. The content that helps with difficulty is not aspirational. It is specific.

• What it costs, and what changes the cost.

• What it does not do, stated plainly.

• How long implementation really takes and what goes wrong.

• How it compares to the two alternatives the buyer is genuinely considering, including the option of doing nothing.

• What the internal business case looks like, because somebody has to defend this to a committee.

  
  

That last one is consistently the most valuable and the least produced. In a purchase involving six to ten stakeholders, the person you are talking to has to sell your solution internally without you in the room. Giving them the material to do that is one of the highest-return assets a marketing team can build.

A useful exercise

The ten-call audit: Listen to ten recent sales calls, or read the notes. Write down every question asked, in the exact words the buyer used. Then check whether your website answers each one, clearly, on a page you could send without embarrassment.

Most teams find between five and fifteen questions with no good answer anywhere. That list, prioritised by how often each question came up, is a content plan grounded in evidence rather than in keyword volume.

What changes when you get this right

Three things tend to happen in order. Sales calls start shorter because more has been settled in advance. The proportion of enquiries that sales accepts as qualified rises, because better-informed buyers self-select. And the mix of self-reported sources widens, which is the clearest sign that you are reaching people earlier than you used to.

None of those show up as a traffic increase. All of them show up in the pipeline, which is the only place it matters.

Frequently asked questions