Read Analytics: sales, goals and customer help

Analytics shows how customer conversations lead to sales and other goals. It also shows how well customers were helped, even when there was no purchase. Use it to spot patterns, not as proof that your bot caused a sale or that every customer was satisfied.

How it works

The page combines recorded results with AI estimates from finished conversations. Recorded results include paid orders, appointment bookings, group class bookings and entries added to Lists. AI estimates describe realistic opportunities, who created them and signs of customer satisfaction.

For these measures, “conversations” means the customer interactions tracked as tickets, not individual messages or a customer’s entire chat history. One customer can contribute several interactions. Rehearsals and interactions classified as having no customer need are excluded from the outcome rates.

Two different dates matter

Outcome rates use interactions that started in the selected period and have finished. They need not have finished within that period. An interaction that finishes later can change the figures for earlier dates.
Revenue and Orders use the date of the order, even if the interaction started earlier or is still open.
The Conversations trend counts interactions that started in the period, including ones still open. It excludes interactions already classified as having no customer need.

The outcome cards show how many interactions from the selected period are still open and how many are waiting on you. Those open interactions are not in the finished-conversation rates. See tickets and outstanding promises for the underlying customer work.

Where to find it

Open Analytics in the sidebar, after Inbox. When available, See analytics on Home also opens it. Your plan must include Analytics and your permissions must allow viewing it. Changing the outcome selection requires separate update permission.

Set it up

Open the page.
Select a time range.
Open Filter conversations if you want one channel or line.
Open Choose outcomes.
In Outcomes to show, check the results you want to see.
Wait for the save status in the top bar before leaving.

The checkboxes save automatically; there is no separate Save button. The choice is shared across your business, not just your account. Dates and conversation filters change the view, not this shared selection.

You do not start measurement by checking an outcome. Measurement continues regardless of what you show. Hiding an outcome does not disable its integration, stop bookings, change billing or delete results.

Settings

Dates and conversation filters

Control	What it does	Default or starting state
7d	Today and the six preceding calendar days.	Not initially selected.
30d	Today and the 29 preceding calendar days.	Used when no valid range is supplied.
All	Activity since the business was created in omotto.	No previous-period comparison.
Custom / Custom range	Choose inclusive start and end dates.	No custom dates until you select them.
Filter conversations	Select a channel or a named line from the menu. When a channel has several lines, the menu lists its lines separately rather than offering the whole channel as well.	All conversations.
All conversations	Remove the channel or line restriction.	No channel or line restriction.

Analytics has no outbound-only switch or label filter. A channel filter includes interactions with messages on that channel; the By channel table assigns each included interaction to one starting channel, which is a different rule.

All date boundaries use the business’s time zone, not your device’s time zone. Today is counted only up to the time you load the figures.

To choose custom dates:

Select Custom.
Use Previous month or Next month to reach the dates you need.
Pick the start date.
Pick the end date. The range applies immediately after this second pick.

Both dates are included. Pick the same date twice for a single day. Picking an earlier date while choosing the end starts a new selection. Future dates cannot be picked. Dates before the business’s creation are brought forward to its creation date.

Closing the calendar before finishing discards the unfinished choice. Reopen the displayed date span to change it. Clear dates returns an applied custom range to 30d. If you open an invalid custom-date link, a missing or invalid start, or a start after the end, falls back to 30d. A missing, invalid or future end becomes today.

Outcomes to show

Choice	What it measures	Default
Sales	Finished interactions with a recorded paid order linked to them. See Shopify connection.	All available outcomes are shown until a selection is saved.
Meetings booked	Interactions with a recorded meeting booking. See meeting types.	Same as above.
Class places booked	Interactions with a recorded place booked in a class. See group classes.	Same as above.
A List’s name	Interactions with an entry recorded as a goal for that particular List. Each qualifying List has its own card.	Same as above.

The available choices depend on your active connections, enabled Lists and plan. Historical outcomes can remain available for the selected period after they stop being active. They are marked no longer active. A deleted List can appear as (deleted list).

After you have saved a selection, a newly available outcome starts unchecked and is marked new until a later selection save records that it is known. An outcome you deliberately unchecked is not marked new just because it is hidden. Changing one checkbox preserves the other choices, including choices outside the current date range.

Hiding Sales removes the sales card, sales headline and sales trends. It does not remove available sales columns from By channel. Hiding every outcome leaves How well customers were helped, the conversation trend and the channel table visible. With view-only permission, the outcome checkboxes are disabled.

Trend display controls

Select a tile under Trends to expand its detail. Only one detail is expanded at a time; select it again to collapse it. Table adds the date-by-date values below the chart. Chart hides that table again. The table starts closed. These controls do not change measurement or the shared outcome selection.

What happens when…

You read Conversion, Had a real chance and Closed when possible

These three measures answer different questions:

Measure	How to read it
Conversion	Of all eligible finished interactions that started in the period, how many achieved this outcome? It uses recorded results, not the AI opportunity estimate.
Had a real chance · AI-judged	Among interactions with a usable assessment for this outcome, how many had a realistic opportunity? A recorded achievement also establishes an opportunity within this assessed group.
Closed when possible · AI-judged	Among those assessed interactions with an opportunity, how many actually achieved the outcome? It is not a percentage of all interactions.

An interaction counts once for an outcome rate, even if it produced several results of that kind. The smaller line shows the numerator, denominator and distinct customers behind each rate.

Judged on … of … conversations reports coverage for this particular outcome. … so far means coverage is incomplete. Achievements outside the usable assessed group still count toward Conversion, but not the two AI-judged rates. The extra count beside coverage counts interactions with the outcome, not a second total of all orders or entries.

The interface describes these as results on conversations without a current assessment. This does not promise that they will all be assessed later. Older interactions, unavailable assessments or removed assessment evidence can leave results without a usable assessment.

Percentages or comparisons are missing

Too few to compare is not zero. A rate needs at least 20 observations in its denominator and 20 distinct customers before a percentage appears. The denominator depends on the measure: finished interactions for Conversion, assessed interactions for Had a real chance, and assessed opportunities for Closed when possible. Needs answered uses individual needs instead of interactions.

The Who raised the chance breakdown needs 20 assessed opportunities from 20 customers. The satisfaction breakdown needs 20 interactions with a clear satisfaction signal from 20 customers. Below these minimums, the page gives a too-few message rather than a percentage breakdown.

Outcome-rate tiles and Needs answered also show an approximate range around the percentage when enough evidence exists. This is an uncertainty estimate, not a guarantee about the next customer. Repeat interactions and AI judgement can affect interpretation.

You compare trends and settled periods

The trend tiles show:

Revenue: linked order revenue, separately for each currency, using order dates.
Sales conversion, 7-day rolling: each chart point uses the seven calendar days ending on that point’s date. It is sales-achieving finished interactions divided by all eligible finished interactions that started in those seven days, not a simple average of daily percentages. The tile’s main percentage covers the whole selected period.
Conversations: eligible interactions started in each interval, open and finished together.

Up to 62 selected days, charts use daily points. Longer custom ranges and All use seven-day intervals from the range start, with a shorter final interval if needed. Conversion still uses a rolling seven-day calculation at each point.

This period is compared with Previous period, the immediately preceding span with the same number of calendar dates. The previous chart line is dashed. Revenue and conversation changes are relative percentages. Conversion changes are percentage points: 20% to 25% is up 5 points. A rise from zero says new this period, rather than giving an impossible percentage. All has no previous line, column or comparison.

Custom ranges do support these trend comparisons. Separately, the comparison lines inside the outcome-rate cards appear only for 7d and 30d. They compare the two most recent settled periods, not today’s unfinished period, and need enough evidence in both periods. Their change indicator is a relative percentage change, not percentage points.

still settling and the shaded conversion dates mean recent interactions can still finish and alter results. The page waits before treating dates as settled; there is no setting here for shortening that wait. A recent custom range can also be settling.

The Conversations total can appear before its chart: that chart needs at least 20 eligible finished interactions in the selected period. Conversion charts need the rate minimums; comparing conversion also needs enough evidence in the previous period. Revenue totals do not require 20 customers.

You read Who raised the chance

On sales and booking cards, this is an AI estimate of who first made the outcome realistically possible:

The customer — the customer raised the opportunity.
Your assistant — the bot raised it.
Your team — a staff member raised it.
Both — the customer and the business both contributed.
Unclear — the origin could not be determined.

The breakdown uses opportunities explicitly identified in the assessment. It does not invent an origin just because an achievement was recorded later. It is not a measure of who deserves payment or credit, and List cards do not have this breakdown.

You read sales attribution and money

Sales from conversations includes orders omotto could link to customer interactions. It is not your entire store turnover.

Linked through the conversation: the link comes from a checkout link sent by the bot or a website cart linked to the interaction.
Linked by timing and customer: the order is linked through the customer and its timing, without that direct checkout or cart link.

Neither category proves the conversation caused the order. Unlinked orders are excluded. A notice about double-checking order links means the displayed attribution figures may still change.

Money figure	Meaning
Revenue	Total recorded linked order value after refunds, for qualifying orders dated in the selected period. Unpaid, partially paid, cancelled and fully refunded orders are excluded.
Orders	Number of those recorded orders, not the number of interactions that bought. One interaction can produce several orders.
Average order	Period revenue divided by period orders.
Revenue per conversation	Linked order value after refunds on the eligible finished interactions that started in the selected period, divided by all those interactions. It includes their qualifying recorded orders up to the current reporting time, not only orders dated in the selected period.

Currencies stay separate. omotto does not convert them into one combined total. Order counts and averages follow the same per-currency grouping. Revenue per conversation appears only when that currency has supporting sales on at least 20 interactions from 20 customers; otherwise it shows a dash. Period revenue and Average order do not use that minimum.

An order is unpaid, cancelled or refunded

Unpaid or partially paid: the order does not count toward Revenue, Orders or Sales conversion. A deposit alone is not enough.
Partially refunded: the order still counts if it is not cancelled and has a positive amount left after refunds. Revenue is reduced by the refund amount. That reduction belongs to the original order date, not the refund date.
Cancelled or fully refunded: the order no longer counts toward Revenue or Orders. Sales conversion can also fall if the interaction has no other qualifying order.

Earlier periods can therefore decrease after an order update, even if the conversation has already finished. Check the order’s payment, cancellation and refund status in Shopify before treating a missing order or lower total as a reporting error.

You read From need to order

Needed something, Saw products, Got a checkout link and Bought are independent counts of eligible finished interactions. They are not steps every customer must pass through. Product guidance is not a purchase, and a checkout link is not proof of payment. A customer can buy without receiving a checkout link.

The bars are each relative to Needed something, not to the bar before them. The checkout-link note, when shown, gives the date tracking began; earlier activity is not a reliable zero for that measure.

You read How well customers were helped

This section remains available regardless of which outcomes you show.

Needs answered: answered needs divided by all assessed needs on the eligible finished interactions. One interaction can contain several needs. An answer may come from your bot or your team.
Dropped promises: interactions recorded as closed with an unkept promise. It is a closing-result count, not the number of unchecked items in today’s promise checklist. Checking off a promise does not rewrite the closing result. Historical results can therefore remain in this count.
Satisfaction, as judged by AI from the conversation: Satisfied, Neutral and Unhappy, based on the customer’s own messages. Customers are not asked for a rating. These are estimates, not survey results.

No clear signal is not Neutral or Unhappy. Unknown signals are excluded from the satisfaction bar and shown separately when the breakdown has enough evidence. The known categories add up within the clear-signal group, not across all interactions.

You read By channel

Each included interaction is assigned once, using its first available real incoming customer message. With several lines, the line name can appear beside the channel. Unknown means a usable starting message is no longer available. The conversation counts across these rows add up to the eligible finished total.

Sales conversion and Closed when possible use that row’s own interactions and evidence minimums. Revenue uses order dates, so a row can have revenue even when no interactions in that row started and finished in your selected period. Revenue across rows adds up to the sales total, separately per currency.

A top-level channel filter can include an interaction that touched that channel but started on another one. The table still assigns it to its starting channel; this is not duplicate counting.

Home shows different numbers

For equivalent dates and the same channel or line filter, Analytics period sales revenue and order count should match the sales result on Home. Make sure Home is not still showing its shorter default period.

Other totals can differ legitimately. Home includes activity-based views; Analytics rates use finished interactions by start date, and its Conversations trend also includes open interactions. Money follows order dates. Do not divide a Home activity count into Analytics revenue and expect Revenue per conversation.

Troubleshooting

Analytics is missing: ask the business owner to check plan access and your viewing permission. If both should be available, contact support.
A goal is missing: check the relevant connection or enabled List, then the selected dates and Choose outcomes. A hidden card is not an inactive integration.
No outcomes selected: check at least one result. If the checkboxes are disabled, you need update permission.
There are achievements but no AI percentages: inspect coverage and the customer counts. Widen the period or remove a channel restriction. Some older or unavailable assessments will not become usable just by waiting.
A filter change does not go through after editing outcomes: resolve the save error beside the outcome choices first. The page waits for pending selection changes to save before changing dates or channel.
A newly recorded result is missing: reload after about a minute; reporting figures can be briefly delayed. Also check the two date rules and attribution. If it still looks wrong, contact support with the range, channel or line, currency and the relevant interaction or order reference through a private support channel.