We don't play favorites with channels. We show their role in sales.
Flowpeek combines its own visit tracking with your store's orders and attributes sales to channels using the model you choose. Compare six models and see which sources start, support and close a purchase.
30 days for PLN 0 · first payment after the trial · cancel anytime. We set it up for you within 48 hours.
With this module
6 models with clear crediting rules
Purchase paths and assisted conversions shown separately
Attributed and unattributed revenue in one context
Attribution in practice
Six ways to assign sales to sources.
Flowpeek supports First Touch, Last Touch, Last Non-Direct, Linear, Position Based and Time Decay. Start by comparing the first and last touch in the report below, then choose the rule that fits your question about a channel's role.
First Touch and Last Touch
Who starts the journey, and who closes it?
The screenshot shows two of the available models: First Touch and Last Touch. The same orders can give different channel shares because the assignment rule changes. You choose the other rules in the attribution settings.
First Touch and Last Touch for the same period. This is an excerpt of the attribution reports, not the full list of available models.
From the report view to your next question
01
Start with the channel whose result changes.
Google Ads has 47 conversions under First Touch and 39 under Last Touch here. Review Top paths and assisted conversions to see where the source appeared earlier. The difference doesn't mean eight orders were lost.
02
See what a Direct return changes.
Direct has 74 conversions under First Touch and 83 under Last Touch here. Last Non-Direct lets you assess the last earlier touchpoint other than Direct, if one exists in a qualifying journey.
03
Keep unattributed sales in your assessment.
Both tables show 136 unattributed purchases. Changing the model won't fill in their history by itself. Check the links, sources and window before you treat a channel's result as the complete picture of sales.
From report to decision
See more. Choose your next step.
6 models with clear crediting rules
Purchase paths and assisted conversions shown separately
Attributed and unattributed revenue in one context
01
The same rules for every channel
You assess ads, the newsletter and other sources based on the same set of orders, the same model and the same lookback before the purchase. A channel gets credit according to its role in the recorded path.
02
Six rules to choose from
You can attribute a sale to the first or last source, skip the final Direct visit or split credit between touchpoints: equally, by position or by time before purchase. Choose the rule that fits your question and stick with it in later comparisons.
03
Earlier touchpoints in view
Top paths show the order of sources before an order. Assisted conversions, or assists, point to the sources the customer came from before the last visit that led to the purchase.
04
A clear scope for conclusions
Check how far before the purchase the report looks for visits and how much of your sales could be linked to a source. Without a confirmed source, a sale stays unattributed; it isn't automatically assigned to Direct visits.
01
six models
Choose the rule that answers your question.
The first touchpoint describes the start of the observed path, the last one its end. Multi-touch models split credit between touchpoints. Switching perspectives helps you notice a channel's role that a single model might miss.
First Touch: 100% credit to the first qualifying touchpoint
Last Touch: 100% credit to the last touchpoint, including Direct
Last Non-Direct: the last touchpoint other than Direct; if there are none, the last Direct
Linear: equal credit for every qualifying touchpoint
Position Based: 40% to the start, 20% to the middle, 40% to the end
Time Decay: more weight for newer touchpoints, halving every 7 days
02
splitting a single transaction
See how the model changes credit without changing sales.
A worked example: before a PLN 700 order, the customer comes from Meta Ads, Google Ads and the newsletter, with one touchpoint each on days 1, 8 and 15. First Touch will attribute PLN 700 to Meta Ads, and Last Touch to the newsletter. Google Ads doesn't take over the purchase just because it appeared on the path. The order is still worth PLN 700.
Linear: each of the three touchpoints gets one third of the credit
Position Based: Meta Ads PLN 280, Google Ads PLN 140, newsletter PLN 280
Time Decay with 7-day gaps: Meta Ads PLN 100, Google Ads PLN 200, newsletter PLN 400
When the same channel appears several times, its shares add up in the sources table
03
paths and assists
See what happened before the touchpoint that closed the sale.
Top conversion paths show the order of observed sources, the number of orders and revenue. A separate assists view helps you find channels that appear earlier, even when they don't get full credit in a last-touch model.
Trace which combinations of sources appear before a purchase
Compare a channel's presence in assists and at the last touchpoint
Don't add assists to orders: several sources can support the same purchase
04
a shared reporting rule
Agree on the model and window before you start talking budget.
You can set one sales attribution rule for the whole app or choose it separately in supported reports: the main dashboard, conversions, marketing costs and multi-channel attribution. When comparing results, check that the reports use the same rule.
Separate the dates of analyzed orders from the window for earlier touchpoints
Check the active model and window in every report you compare
Keep the same rules, periods and currency when analyzing trends
05
coverage and unattributed sales
Know what share of orders your channel assessment describes.
The report also shows unattributed purchases and revenue. The reasons for missing attribution help you choose a specific check: identifiers, timing, the window or source recognition. Changing the model doesn't fix missing input data.
A missing link to a visit stays visible
Visits after a purchase aren't proof of its source
We don't replace an ambiguous order of events with a made-up path
Read any conclusion about a channel together with the scope of recognized sales
Getting started
Start with your own store.
Go from setup to a specific question. During the trial, you'll see how this module helps in your day-to-day work.
01
We link visits with orders
We set up tracking and the store integration, and check the identifiers needed to link a purchase to an observed path.
02
Choose the period, window and model
Decide which dates you're analyzing orders from and how far back you look for touchpoints. Then compare models on the same data set.
03
Note the channel's role and the next step
Set the share of sales against paths and assists. Base any budget conclusion on costs and the scale of the data, too.
Google Ads has a bigger share at the first touchpoint than at the last.
Compare First Touch, Last Non-Direct and assists for the same period and window.
What you gain
You establish whether, before cutting the budget, it's worth checking the ad's role in starting the customer's path.
Advertising agency
A client is comparing the ad platform's results with the store's results.
Write down the sales definition, model, window and unattributed revenue, then review sample paths.
What you gain
You explain the cause of the difference and set a consistent basis for the next report.
Analyst
The share of unattributed orders is growing.
Check visit links, qualifying sources and event timing before changing the crediting rule.
What you gain
You separate a change in tracking quality from a change in marketing results and point to the part of the integration that needs fixing.
Store owner
A customer buys after coming back several times.
Compare the linear, position-based and Time Decay models, and set the sources' roles against acquisition costs.
What you gain
The budget discussion covers different moments in the purchase decision, not just the last visit.
Your plan for the first 30 days
Check the value. On your own data.
Explain how the chosen model changes the assessment of one important channel.
01
Link orders and visits
We set up Flowpeek within 48 hours and check the links needed for attribution reporting.
02
Record a reference point
Set the period, currency, window and model. Note the number of orders and the unattributed share of sales.
03
Compare credit and paths
Check at least two crediting rules, then assists and Top paths. We'll discuss the differences in the consultation during your first month.
04
Prepare a decision to test
Write down the channel's role, its cost and the next action. Keep a description of its share in the path separate from proof of the ad's incremental impact.
How will you know it's worth it?
You can show why a channel's result depends on the model and explain what share of the data your conclusion covers.
Does not playing favorites mean every channel gets the same share?
No. Neutrality means an ad platform's name doesn't earn it any extra credit. What counts are the observed touchpoints and the chosen model: their order, number or time before purchase. The Last Non-Direct model has an explicit rule for skipping a final direct visit if another source appeared earlier. So channel results can differ a lot.
Which attribution models does Flowpeek support?
Six: First Touch (the first touchpoint); Last Touch (the last one); Last Non-Direct (the last one other than Direct); Linear (an equal split); Position Based (a split by position, 40/20/40); and Time Decay (more credit for touchpoints closer to the purchase, with weight halving every seven days). You choose models in the attribution settings. The screenshot shown compares only First Touch and Last Touch; it doesn't present the full list of options. Assisted conversions are a separate breakdown, not a seventh model for splitting revenue.
What does Last Non-Direct do when a customer comes back directly?
It gives credit to the last qualifying touchpoint other than Direct. If the entire available path contains only Direct touchpoints, the last of them gets the credit. Observed data and a reliable link to the order are still required.
How do the position-based and Time Decay models work?
With at least three touchpoints, the position-based model gives 40% to the first, 40% to the last and splits 20% among the rest. For two touchpoints it uses 50/50, and for one, 100%. Time Decay gives newer touchpoints more weight: a touchpoint seven days older has half the weight of a newer one. The final shares are recalculated so that together they add up to 100%.
Why can conversions in the table be fractions?
In models that split credit, a single purchase can be partly assigned to several touchpoints. For three touchpoints, the linear model gives each one a third of the conversion and of the order value. The table adds up these shares for each source; it doesn't mean there's a fractional order in your store.
How do assisted conversions differ from attributed sales?
An assist means a source was present before the last touchpoint on the observed purchase path. Within a single purchase, a given source is counted once, even if it appeared several times earlier. Several sources can assist the same purchase, which is why assist totals aren't added to the number of orders or to revenue.
How is the report period different from the attribution window?
The report period selects the orders to analyze. The attribution window defines how far before each order to look for qualifying touchpoints. By default, it's 30 days. The model and window can be set globally and separately for supported reports, so always check both settings when comparing results.
Why does some revenue stay unattributed?
The link between an order and a visit, the source name or a touchpoint within the selected window may be missing. The accuracy of event timing and order matters too. Importing historical orders alone doesn't recreate past clicks. Unattributed doesn't automatically mean Direct, declined consent or an ad blocker.
Does the model show how many sales an ad caused?
It shows how credit is split according to the selected rule and the observed path. It doesn't prove how many purchases wouldn't have happened without the ad. Assessing incremental impact requires a separate test. During the 30-day trial, we'll help you understand the report and prepare a comparable assessment of your channels.
