Account Performance
The account performance page shows you how your store is doing overall and how much of that Shuttlebase is responsible for. It rolls up every experiment on your account into a single view.
Annual totals
Section titled “Annual totals”The top of the page shows your store’s totals for the year:
- Visitors
- Purchases
- Revenue
These come from Shuttlebase’s own first-party tracking snippet, not from a third-party analytics tool. See Tracking snippet for how the data is collected.
Revenue uplift
Section titled “Revenue uplift”Revenue uplift is the extra revenue Shuttlebase generated compared to a modeled baseline. The baseline estimates what your revenue would have been without Shuttlebase running. The difference between actual revenue and that baseline is the uplift.
Each deployed experiment contributes two parts to the total:
- Revenue earned while the test ran. This part is measured directly. It is the revenue the winning variant produced minus what the control produced, adjusted for each side’s share of traffic.
- Revenue earned since the winner was deployed. This part is estimated. Shuttlebase takes the revenue-per-visitor difference between the winner and the control, applies the conservative lower bound of its confidence interval, and multiplies by your expected traffic. See Revenue uplift for the full calculation.
Three safeguards keep the number honest:
- An experiment contributes nothing until both its winner and its control have enough purchase data to trust.
- Each win’s contribution fades gradually to zero over time, so an old change does not keep claiming full credit forever as your site, catalog, and market move on.
- A weak result never subtracts from the total. If the data cannot support a positive uplift, the experiment contributes zero.
For the full model, including the confidence bound, the fade schedule, and a worked example, see Revenue uplift.
Rate metrics
Section titled “Rate metrics”Three rate metrics sit alongside the totals:
| Metric | What it measures |
|---|---|
| Session conversion rate | The share of sessions that end in a purchase. |
| Average order value (AOV) | Revenue divided by number of purchases. |
| Revenue per visitor (RPV) | Revenue divided by number of visitors. |
These are the same three views Shuttlebase uses when a currency metric is the target of an experiment. See Metrics and guardrails for how they work inside experiments.
Experiment totals and success rate
Section titled “Experiment totals and success rate”The page counts the experiments Shuttlebase has run on your account and shows your success rate, the share of experiments that produced a winner. Use this to see how testing volume and win rate change over time.
Cumulative uplift chart
Section titled “Cumulative uplift chart”The cumulative uplift chart plots your revenue uplift over a two-year window: the 12 months behind you and the 12 months ahead. The Today line splits the chart into what has happened and what comes next, in three layers.
Earned
Section titled “Earned”The green area shows the uplift your deployed winners have accrued to date. Each deploy appears as a marker on the timeline, so you can see how every win added to the running total. This is the realized half of the chart, built from the two-part calculation described above.
Locked in
Section titled “Locked in”The purple area projects the next 12 months of uplift from experiments that are already deployed. You do not need any new wins for this revenue to arrive, which is why it is called locked in. It uses the same conservative, fading model as the earned total, so the projection tapers off over time instead of growing forever. Earned plus locked in gives the total revenue impact shown at the right edge of the chart.
Future upside
Section titled “Future upside”The blue band estimates what new experiments could add on top, assuming the next 12 months perform like the last 12. Unlike locked-in revenue, this depends on wins that have not happened yet, so read it as potential rather than commitment.
To learn what happens when a winner rolls out, see Experiment terminal states.