Skip to content
Interface language: English
Imperium / Signal

Who brought the client? B2B attribution and the medal for the last click

B2B attribution organises contact credit but does not reconstruct why someone bought. Separate channel history, allocation rules and evidence of an action’s effect.

Article language: English

Fine white strands cross a wooden piece with gold-coloured teeth on a workbench.
In this article

Imagine an exhibition design studio receiving an enquiry from a museum. The form was submitted after a direct visit to the website. The team therefore assigns the enquiry to "direct traffic" and wonders why it still produces a newsletter. A conversation reveals that one person was referred by a friend, another read an article and a third attended a webinar. The form closed one stage; it did not tell the whole story.

Attribution assigns a share of an outcome to selected contacts or channels. It helps organise data, but requires a rule and a defined scope. When a company asks what brought a client, it may mean first awareness of the brand, help with evaluation or the prompt to send a message. Those are different questions. "Last click" is sometimes convenient chiefly because it fits easily into one cell.

First establish which outcome you are trying to explain

In our hypothetical example an enquiry was sent. No contract has yet been signed and no exhibition delivered. If a report calls that moment a sale, subsequent conclusions begin to concern an outcome that has not been observed. Keep the contact, the conversation about scope and the later purchasing decision separate, in line with how the company actually works.

The channel assignment may look different at each stage. A piece of content helped explain the studio's specialisation, a webinar allowed someone to ask a question, and a separate conversation clarified whether the project was feasible. None of those contributions follows automatically from the time of the last website visit. Establish which event is being reported and what is confirmed about it.

Google's description of conversion attribution models distinguishes data-driven and last-click models. It describes how credit is assigned to channels in a particular report. It does not read every conversation or buying motive in every company. Before relying on it for a decision, a team needs to know which contacts the report covers.

When using a report, check which event was treated as the outcome, which contacts are visible and how the chosen rule works. That information should accompany the figures. Without it, a reader may treat a channel name as a complete explanation, even though it shows only the recorded part of the journey. The word "model" does not remove the limits of the input.

A guide to the B2B contact form (in Polish) helps check whether an enquiry is delivered. That is a necessary operational outcome. Why the visitor decided to send it is a different question and needs other information. A working form can close a journey that began with a contact the website system never recorded.

One organisation can have several contact histories

Consider the fictional museum in more detail. The person developing the idea knows the studio through a referral. A colleague responsible for materials reads its explanation of how it works. Another team member attends an online event. The message is sent by someone who received only the website address internally. This is a possible sequence, not a statistic about every B2B buyer.

A company that reads only the last visit by the person submitting the form sees a small part of the event. It should not turn arbitrary visits by different people into one certain journey. Keep the boundary between tool data, a person's account and the team's assumption. A neat story assembled afterwards can still consist of unconfirmed connections.

In conversation, ask how the organisation learned about the company and what helped it decide to make contact now. Answers have limits of their own: the person knows one part of the history and may not remember every piece of content. They should not be treated as a perfect measurement of all influence. They can, however, reveal a referral or work by other people on the client's side.

In the notes, distinguish what the client said from what the company established in its own system. "The interviewee named the webinar as helpful during evaluation" is narrower than "the webinar caused the sale". In our example, the first statement can be kept without inventing an exact share for every person and publication.

Record a different purpose if the conversation reveals one. The webinar participant may have sought knowledge for their own work and only later passed the studio's name to the team. That does not automatically make them a ready buyer on the day of the event. The company’s own webinar should retain value for its participant regardless of a later order. A guide to planning an expert webinar (in Polish) discusses how to prepare such an event.

Three people do not necessarily mean three clients

In our fictional museum, different people preparing one exhibition registered for the webinar. The studio later received one enquiry about that project. If a report counts participants in one place and cases in another, its reader may see growth caused solely by a change in unit. Define what appears in each column.

The same applies when information moves between teams. One employee of the client may attend an event, another discuss the evaluation and a third act as the contact for preparing an offer. There is no need to assign each a separate sale. Nor should their contribution be erased merely because one person formally sent the enquiry. A report can retain a connection to the case if it has actually been confirmed.

The reverse needs similar care. Imagine two departments of the same organisation enquire about independent exhibitions. A shared client name does not necessarily mean one task or one decision. The report should answer what the team wants to explain: the number of participants, new organisations or projects under consideration. Each unit can be useful if its meaning is preserved.

Before a management meeting, read column names as ordinary sentences. Does "client source" mean how the organisation first learned of the studio, or the last visible visit by the person sending the message? Does the number of contacts mean people, enquiries or conversations? Such questions often reveal a discrepancy before a discussion about a more complicated attribution model.

The allocation rule changes the picture of credit

Now adopt a working rule for the studio: assign all credit to the last recorded contact before the enquiry. We are not describing all actual analytics-platform settings. We are showing the consequence of one rule adopted for this example. Earlier contacts remain in the history but receive no credit in this particular field.

If the team later treats no credit as proof of no value, it draws a conclusion the rule did not test. A newsletter may help someone recognise the studio's specialisation even if they return to the website another way. A referral may start a search without a click on company content. Their invisibility proves neither effectiveness nor ineffectiveness.

Aeneia conceptual photography

Changing the rule to an equal share among visible contacts will not automatically reconstruct buying motives either. It is another way of organising the same recorded history. It can show how the assumption affects the report. It does not establish that an article, meeting and message influenced the decision in those exact proportions.

Ask management's question more precisely. Do we want to see which materials appear before a conversation? Learn how audiences know the company? Test the effect of changing a particular activity? Each purpose may need different evidence. A single "client source" column should not pretend to answer all of them.

The article on measures that produce unwanted behaviour (in Polish) shows the organisational significance of such a rule. If only the last contact is rewarded, people may concentrate on taking over the end of a conversation. Look for that effect in your own company rather than treating a table entry as a neutral medal for the entire contribution.

A contact before a purchase and the cause of a purchase are different questions

Research by Coviello, Gneezy and Goette, published as a working paper in 2017 and revised in 2018, examined an experimental suspension of paid search at Edmunds.com. The authors tested the effect of changing an activity rather than stopping at the fact that a click preceded later behaviour. This concerns one company and channel; it does not justify transferring the result to our museum or every newsletter.

For the studio, it is useful to distinguish the question. Contact history describes what was recorded before the enquiry. A causal assessment asks what would have changed without an activity or after changing it. The first report alone does not answer the second question. That does not mean throwing the history away; it means using it for the right purpose.

Suppose more enquiries arrive after the newsletter changes. At the same time, the studio published a new project and was recommended at an industry event. The sequence of dates alone cannot assign the whole difference to one issue. The team can retain the observation and identify possible explanations, then choose a question that can be tested in a suitably limited trial.

A business experiment plan (in Polish) helps record a hypothesis and the limits of the conclusion before the result. Not every small company can make a strong causal comparison. A more honest result may be a better understanding of how content is used, with its effect on purchasing still open. Do not fill the gap with certainty simply because the report has a deadline.

A decision about your own activity needs its own grounds. A historical study of a particular channel does not set the studio's spending on a newsletter or webinar.

Aeneia Design

Let us organise the questions and contact points on your website before you assign the entire enquiry to one button or channel. We will show how materials guide a visitor toward a clear conversation about the offer.

Discuss the audience's path to B2B contact

Leave room for what is unknown

In our example, part of the history comes from the interviewee, part from company records and part remains unknown. That is useful information. There is no need to force every case into one apparently precise field. Keep the missing information visible and say whether it actually prevents the decision the team wants to make now.

If the company wants to improve its newsletter's editorial promise (in Polish), it can ask readers about the usefulness of particular topics. It need not first prove which text alone caused a contract. If it wants to assess the effect of changing a channel on an outcome, however, it needs an appropriate comparison. The question should guide the choice of evidence, not the other way around.

When collecting more information, restrict the scope to the actual task and appropriate access rules. The company should agree permissible measurement with the appropriate people; reconstructing a history does not justify independently bypassing access or tracking rules. A missing observation remains missing, even when it would make a very attractive chart.

Also distinguish material sent from material used. A salesperson may attach an article to a message without knowing whether the recipient read it. A client may register for a webinar but watch only the recording later. Each confirmed state has value. None should automatically advance to the next stage without appropriate information.

A report should end with a better decision

A review of your contacts can show three things: confirmed events, the allocation rule and questions still unanswered. In the museum example, the team keeps the referral, the identified materials and the enquiry. It does not have to choose one hero to decide what to ask the next person or what gap to fix on the page.

At Aeneia Design, a starting point can be to organise the offer, materials and path to contact. That scope helps establish the page's role in a longer B2B conversation. Measurement configuration and its implementation require separate agreement. A useful site helps the visitor assess the offer and take the next step.

Management can then receive a less spectacular but more useful answer: we know the last recorded contact, the client also named other materials, and each one's effect on the purchase remains unestablished. We know what can be improved and what needs further checking. The last-click medal can stay in the drawer. It need not direct the company's entire plan.

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear