UTM Attribution ROI Calculator

Start with the four numbers your analytics can actually give you. Last-touch conversions and first-touch conversions come straight from a UTM-segmented conversion report. Journeys touched is the count of converting journeys where this campaign appeared anywhere, the "assisted conversions" view in most tools. Average touchpoints is the mean number of marketing touches across converting journeys, typically between two and five.

For the campaign above, the linear credit is 260 journeys divided by 3.2 touchpoints, or 81.25 conversions: the campaign gets an equal share of every journey it appeared in. Position-based gives 40% of the 96 first touches, 40% of the 140 last touches, and splits the remaining 20% across the 24 journeys where the campaign was neither first nor last: 38.4 + 56.0 + 4.0 = 98.4 conversions.

Then the correction almost nobody makes. Set untracked share to the proportion of real conversions your UTM tagging never sees at all: dark social, copied links stripped of parameters, cross-device journeys, direct return visits. At 18%, the 140 tracked conversions imply 140 ÷ 0.82 = 170.7 real ones, adding 30.7 conversions the report never showed. Revenue rises from $11,900 tracked to $14,512 attributed, and cost per acquisition falls from $34.29 to $28.11.

Finally, judge the result against gross profit rather than revenue. At 62% margin the campaign produces $8,998 of gross profit on $4,800 of spend ($4,198 net, an 87% ROI). The maximum you could pay per acquisition and still break even is $85 × 62% = $52.70, so at $28.11 there is real headroom to bid harder. The marketing ROI calculator runs the same profit logic across a whole marketing budget rather than one campaign.

The models diverge differently when journeys are short. Take a niche campaign with $650 of spend, 1,900 sessions, 22 last-touch and 14 first-touch conversions, 41 journeys touched, only 2.4 average touchpoints, a $240 average order value, a 48% margin and a heavier 25% untracked share. Read it position-based.

Credit comes out at 0.4 × 14 + 0.4 × 22 + the middle share, giving 15.4 conversions, which is less than last click's 22 and less than linear's 17.1. Grossed up for untracked traffic that becomes 20.5 conversions, worth $4,928 in revenue and $2,365 in gross profit against $650 of spend. That is 7.58x ROAS, a 264% ROI, and a $31.66 cost per acquisition against a $115.20 ceiling.

Two things are worth noticing. The model spread here is 57% rather than 72%, because with only 2.4 touchpoints per journey there is less middle to share out. Short journeys make attribution arguments smaller. And the untracked correction matters far more at 25% than at 18%: it adds a fifth of the campaign's measured performance back.

For campaigns where the spend decision is a daily budget rather than a retrospective review, the PPC daily ad budget calculator works backwards from a target cost per acquisition, and the email marketing ROI calculator handles the channel where last-click attribution is least defensible of all.

There is no correct model, only models that are wrong in known directions. Last click overstates bottom-funnel channels (branded search, retargeting, email) because they are near the purchase by construction. First click overstates discovery channels for the same reason in reverse. Linear understates everything if journeys are long, since credit is spread thin. Position-based is a compromise that hard-codes an opinion: that the first and last touches matter twice as much as the middle ones.

The practical answer is to pick one model as the reporting standard, keep it fixed, and look at the spread as a confidence interval. A campaign that clears break-even under every model is genuinely working. A campaign that clears it only under last click is a candidate for a holdout test, not a budget increase.

The tagging itself decides whether any of this is measurable. Three rules cover most of the damage: use a documented, lowercase naming convention so Facebook, facebook and FB do not split into three sources; never tag internal links, which overwrites the original source and hands your own site the credit; and check that redirects preserve query strings, since a single redirect that drops parameters silently converts a tracked campaign into direct traffic.

The untracked share is worth estimating rather than ignoring. A rough method: compare direct-traffic conversions in a period with heavy campaign activity against a quiet baseline period. The lift in "direct" is largely campaign traffic that lost its tags, and expressing it as a share of total conversions gives a defensible input. For the broader vocabulary around all of this, the paid media metrics guide covers how CPC, CPM, CTR, CPA, ROAS and ROI relate to each other.

Every model here is a rule for dividing credit, not a measurement of causation. None of them can tell you whether a conversion would have happened without the campaign. The only method that answers that question is a holdout or geo-lift experiment, where a comparable audience is deliberately not exposed and the difference is measured. Attribution allocates observed conversions; incrementality testing establishes which ones the marketing caused.

The linear and position-based figures use an average number of touchpoints across all journeys rather than journey-level data. That is the input most people can actually get, but it flattens real variation: a campaign that appears in short journeys and one that appears in long journeys are treated identically. If your analytics can export per-journey path data, a path-level model will be more accurate than this approximation.

The untracked-share correction assumes the conversions UTM tagging misses look like the ones it captures: same order value, same distribution across campaigns. In practice untracked traffic skews toward brand-aware buyers and repeat customers, so grossing up a prospecting campaign by the site-wide untracked rate probably flatters it. Use a channel-specific estimate where you have one.

Finally, the margin input carries a lot of weight and is easy to get wrong. Gross margin here means revenue minus cost of goods, shipping and payment processing, not net margin after overheads, and not the headline markup. A campaign judged against an overstated margin will look profitable at a ROAS that is actually losing money on every order.

The Marketing ROI Calculator applies the same gross-profit logic across a whole marketing budget rather than a single campaign. The PPC Daily Ad Budget Calculator works backwards from a target cost per acquisition to the daily spend that supports it. The Email Marketing ROI Calculator covers the channel where last-click attribution is least defensible, since email usually arrives late in a journey it did not start. And the Cost Per Lead (CPL) Calculator handles the top of the funnel, where conversions are leads rather than orders and the attribution question starts all over again.