Which Interactions to Include in an Attribution Report

Choosing which interactions to include in an attribution report is the decision that shapes every number the report produces.

Include too many interaction types and high-volume, low-signal events drown out the moments that actually moved a deal.

Include too few and you lose the context that explains why a contact converted in the first place.

This guide walks through how to make that selection deliberately, using the way attribution reporting platforms handle interaction history as the practical frame.

Why the interaction list matters more than the model

Most teams spend their energy picking an attribution model — first touch, last touch, linear, time decay — and treat the list of included interactions as a checkbox.

That is backwards.

The model only distributes credit across the interactions you have already decided to count.

If the wrong interactions are in the set, every model distributes credit to the wrong things.

HubSpot's documentation describes this explicitly: each attribution report uses a defined set of interaction types to determine credit eligibility.

You can include up to 100 event input types per report HubSpot Knowledge Base.

That limit is generous, but the practical question is not how many types you can add — it is which types carry decision-relevant signal for the conversion you are measuring.

Start from the conversion, not the interaction list

Before touching the event inputs, be clear about what conversion the report measures.

Attribution platforms typically let you choose a data source that defines the conversion event.

In HubSpot, the three report types are Contact Create Attribution, Deal Create Attribution, and Deal Revenue Attribution.

The documentation suggests thinking of them as the top, middle, and bottom of a funnel respectively HubSpot Knowledge Base.

The conversion you pick changes which interactions belong in the report:

• Contact creation (top of funnel): early discovery interactions matter most — first page views, ad clicks, organic search arrivals, and the form submission that created the contact.
• Deal creation (middle of funnel): engagement that signals buying intent becomes relevant — pricing page visits, content downloads, email engagement, meetings booked.
• Revenue attribution (bottom of funnel): sales-side interactions join the picture — calls, meetings, and proposal-related activity alongside the marketing touchpoints that set them up.

A common mistake is building one interaction list and reusing it across all three report types.

The interactions that explain contact creation are not the same set that explains revenue, so a single list serves none of the three questions well.

Separate high-volume from high-signal interactions

Not all interaction types carry the same evidentiary weight.

Page views and email opens happen constantly; a meeting request or a form submission is comparatively rare and far more deliberate.

When both kinds sit in the same report, the high-volume types can dominate the credit distribution simply because they occur more often.

Platforms handle this in different ways.

HubSpot samples high-frequency interaction types, such as email interactions and page views, when total volume would exceed the report's interaction limit.

Lower-volume types like form submissions, calls, and meetings are not sampled HubSpot Knowledge Base.

Sampling preserves the report's date range and prevents high-volume types from receiving a disproportionate share of credit.

But it is a mitigation, not a substitute for your own judgment about what belongs in the set.

A useful rule of thumb: include an interaction type if, when you see it in a contact's timeline, it changes your read on that contact's journey.

If a generic page view tells you nothing you did not already know, it is a candidate to exclude.

If a pricing page visit or a webinar attendance meaningfully shifts the story, keep it.

Practical selection criteria for event inputs

When you review the available interaction types for a report, work through these questions for each candidate:

1. Does it reflect a deliberate action? Form submissions, meeting bookings, and chat starts require effort. Passive impressions require none. Deliberate actions usually deserve inclusion.
2. Can your team act on it? If an interaction type appears in reports but no one can change how often it happens — for example, a third-party directory listing you do not control — its inclusion adds noise without a lever.
3. Is it captured reliably? An interaction type that only records intermittently, because of tracking gaps or sync failures, will understate its own channel. Including it can be worse than excluding it, because the report then looks authoritative while missing data. This is where cleanup work upstream pays off; if syncs are unreliable, fix that first — see our guide on why solving sync failures is the fastest path to attribution cleanup.
4. Does it duplicate another type? If two interaction types describe essentially the same behavior, including both splits credit artificially. Pick the one your team reports on.

How the model interacts with your selection

Once the interaction set is chosen, the model determines how credit flows across it.

HubSpot supports First Touch, Last Touch, Linear, Time Decay, and an Empirical model that weighs interaction types by how often each occurs across conversion paths.

Rarer types receive more weight because they are more distinctive HubSpot Knowledge Base.

Notice the interaction between model and selection.

The Empirical model rewards you for including distinctive, lower-frequency interactions, because they receive more weight.

If your set is dominated by page views, the model has little distinctive signal to work with.

This is a concrete reason to curate the list rather than accept a default: the composition of the set changes what each model can express.

For most operators, a sensible pairing is a curated set of deliberate interactions with a time decay or empirical model for revenue questions.

Use a broader set including early discovery interactions with a first touch or linear model for contact creation questions.

Match the model to the question you are asking — we cover that pairing in more depth in which attribution report type answers your revenue question.

Scope the report to a journey segment when the full picture is too noisy

Another lever besides the interaction list is the population the report covers.

If you sell into multiple segments with very different journeys, a single report mixing them all will blur the interaction patterns.

Scoping the report to one journey segment lets you tune the interaction set to that segment's actual behavior.

For example, scope by contacts from a specific source, lifecycle stage, or industry.

This matters because interaction availability differs by segment.

Enterprise buyers may show long sequences of email engagement and meetings, while self-serve buyers convert with almost nothing but page views and a signup form.

A report scoped to each segment can include the interaction types that actually occur there, instead of a compromise list that fits nobody.

Watch for what the report cannot see

Even a well-curated interaction set only includes what your systems captured.

Offline interactions, conversations that happened before tracking was in place, and events trapped in a tool that never synced to your CRM are invisible to the report.

Before drawing conclusions from an attribution report, confirm that the systems holding those interactions are actually connected.

If you run marketing and sales across separate platforms, see our comparison of where operators still lose execution across HubSpot and Pipedrive.

It covers the integration gaps that create these blind spots.

It is also worth distinguishing editorial confidence from platform behavior.

The interaction types available, the event input limit, and the sampling behavior are documented platform facts.

Which types deserve inclusion for your business is an editorial judgment that should be reviewed by someone who knows the funnel.

An automated or default configuration does not by itself prove the selection is right for your data.

A review process that keeps the set honest

Interaction selection is not a decision you make once and file away.

Campaigns change, new channels appear, and tracking implementations drift.

Treat the event input list as a maintained artifact:

• Document why each interaction type is included, so future reviewers can challenge the list rather than inherit it silently.
• Revisit the list when you launch a new channel or notice a report behaving oddly — for instance, when one interaction type's share of credit shifts without a corresponding change in activity.
• Check capture reliability periodically. An interaction type that stops recording will quietly distort the report long before anyone notices.

Teams that coordinate marketing and sales across several tools often find this maintenance is where attribution efforts stall — the report gets built once, then drifts.

Putting it together

A defensible interaction selection follows a short sequence.

First, pick the conversion the report measures.

Then list the interaction types that genuinely changed outcomes for that conversion, and exclude high-volume types that add noise.

Verify the remaining types are captured reliably, and only then choose the model that distributes credit across the set.

Keep the list documented and reviewed, and scope reports to journey segments when behavior differs across your audience.

The payoff is reports your team actually trusts — where a shift in credited interactions reflects a real change in buyer behavior, not a change in what the report happens to count.

A report is only as honest as the interactions you let into it. Curate the set deliberately, and the model finally has something worth distributing.

For a broader view of attribution decisions in an organic growth workflow, see marketing attribution automation for organic growth decisions.

How Meshline can help. Connect automation, Organic Marketing (demand generation), and customer lifecycle management (Revenue Intelligence).

Bring topic planning, content publishing and performance feedback into the conversation about your workflow. Book a Meshline demo.