How impact is estimated

Where the euro figure on an idea comes from, what it assumes, and how much weight to put on it.

Every idea carries an estimated impact in money per month. This page explains the arithmetic, so you can judge the number instead of trusting it.

The shape of the estimate

The estimate is always the same four terms:

affected sessions per month
  × the share of those sessions the change plausibly recovers
  × your conversion rate on that surface
  × your average order value
= estimated revenue per month

Each term comes from your own data:

Term	Source
Affected sessions	GA4 sessions on the pages the change touches
Recovered share	A conservative assumption, stated on the idea
Conversion rate	Your rate for that surface and segment, not a benchmark
Average order value	Shopify orders over the same period

A worked example

An idea proposes fixing a variant selector that fails on mobile.

Term	Value	Where it came from
Mobile sessions on product pages, per month	24,000	GA4
Share showing the failed-tap pattern	6% → 1,440 sessions	Click Context
Share plausibly recovered	20% → 288 sessions	Stated assumption
Mobile product-page conversion rate	1.8%	GA4
Average order value	€78	Shopify

288 × 1.8% × €78 ≈ €404 per month.

The recovered share is the only assumption. HI Engine keeps it conservative on purpose, and it always states the figure it used.

Confidence

An estimate is only as good as its inputs. HI Engine labels each idea's confidence, and the label is part of the TRICE score.

Confidence	What it means
High	Large sample, sources agree, effect visible across segments
Medium	Adequate sample, or one source only
Low	Small sample, or a difference not distinguishable from noise

Two groups of visitors rarely convert at the same rate, even when nothing distinguishes them. The smaller the sample, the more of that everyday variation shows up as an apparent difference. On a low-traffic page, a handful of orders either way can double an apparent conversion rate.

So HI Engine only states a conversion-rate comparison as fact when the difference is statistically significant — large enough, over enough visitors, that chance is an unlikely explanation. Below that, it writes the idea as a hypothesis and ranks it accordingly, and it says which one it is on the idea itself.

Read this as a ranking device

The euro figure is best used to compare two ideas, not to forecast your P&L. It tells you which change is worth doing first. It does not promise that revenue will rise by that amount.

TRICE

The estimate feeds one of five dimensions in the ranking score.

Letter	Dimension	Weight	Scored on
T	Traffic affected	0.20	Sessions the change reaches
R	Revenue potential	0.30	The estimate above
I	Implementation ease	0.15	How much theme work the change needs
C	Confidence	0.20	Data quality behind the claim
E	Expected time to a result	0.15	How quickly a change would show up in the numbers

Each dimension scores 1 to 5. The weighted composite, out of 5.0, sets queue rank. Revenue carries the most weight; confidence carries as much as traffic, so a large but poorly evidenced idea does not outrank a smaller certain one.

Calibration from your own tests

When you run A/B tests, the results feed back. If sticky add-to-cart changes reliably win on your store, HI Engine raises the confidence it assigns to that category of idea for your store specifically. If they reliably do nothing, it lowers it.

The ranking gets more accurate for your store the longer you run it.

What the estimate does not include

Margin. The figure is revenue, not profit.
Seasonality. It uses a recent period and does not project a peak.
Interaction between changes. Two ideas that both fix the same drop-off do not add up.