What is multi-touch attribution?
Multi-touch attribution splits the credit for a conversion across the ads and visits that came before it, using a formula such as linear or time-decay. It aims to be fairer than last-click and is harder to check against anything real.
Updated
Multi-touch attribution for gyms
The pitch for multi-touch is fairness. A member saw four things before joining, so why should one ad take all the credit? Linear gives each touch an equal share. Time-decay gives more to the touches nearest the sale. Position-based gives most to the first and last and splits the rest. Data-driven, the platforms’ favorite, uses a model they do not show you.
For a large ecommerce brand with thousands of sales a day, those formulas smooth out into something useful. For a gym signing a few dozen members a month, they produce fractions: a quarter of a member credited to a Meta ad, a third to a Google search, and no way to check any of it, because nobody at the front desk ever signed up a third of a person.
The stakes are budget. If a formula hands a prospecting campaign partial credit for every member, that campaign looks essential and its spend never gets questioned. If the formula is wrong, you will not find out, because the report is the only evidence. Whatever model you use, apply it to payments rather than forms, and read the attribution model entry for how the rule and the attribution window work together.
How Adsu uses multi-touch attribution
Adsu does not weight touches. Every visit in the 30 days before a payment is recorded on the member’s journey, and one of them gets the credit. Adsu credits the last eligible touch inside the 30 days before the payment. Adsu counts verified stages only: if a stage cannot be verified, it is not counted. Every report is one line per campaign per location.
That is a deliberate trade. A whole member credited to one ad is a number you can argue with and check; a member split four ways by formula is not. Renewals are carried as recurring value for that member, credited to the click that started the relationship. Booked, showed and paid go back to Google as offline conversions and to Meta through the Conversions API.
Questions, answered.
No. Adsu records every touch in a member’s journey and credits the last eligible one inside 30 days of the payment, without splitting or weighting. You can see the whole journey for any member, so the earlier touches are not lost; they just do not get a fraction of the revenue.
It is more elaborate, which is not the same thing. Multi-touch formulas are assumptions about how much each touch mattered, and there is no front-desk record to check them against. Last-click applied to a verified payment is a smaller claim that can be checked. For a gym, the checkable number is the accurate one.
A multi-touch model where the platform’s own machine learning decides how to split credit, using its data and a method it does not publish. It is applied to the conversion you gave the platform, usually a form fill, and it credits touches on that platform. It cannot see a payment at your front desk unless something sends it there.
Related
An attribution model is the rule that decides which ad, click or call gets credit when someone joins after several touches. Four common models, and Adsu’s.
Last-click attributionLast-click attribution gives all the credit for a conversion to the last ad click or touch before it happened. What it misses, and how Adsu applies it.
Attribution windowAn attribution window is the number of days after an ad click or call during which a conversion can still be credited to that ad. Adsu uses 30 days.
Lead trackingLead tracking is recording where each enquiry came from and what happened to it next. What to track, and how Adsu follows a lead to the payment.
The proof
- Net members
- +21
- Lifetime value added
- $67,000+
Bobby & Maria Gasdia, Big Day Fitness, 2 locations
Stop paying for leads. Start paying for members.
Early access is opening in small groups. Applying takes two minutes, and we’ll tell you straight whether your stack is ready to connect.