Skip to content

Voice of Customer

Overview

Voice of Customer gathers what customers say across the conversations of a period and splits it by who should act on it: sales, marketing, product and service, site and content. Open it from Analytics → Voice of Customer in the sidebar; in Bitrix24 it is a tab of the app.

Available on Pro and higher plans: on other plans the sidebar item keeps its lock and the plan name. A free organization also gets it during its 14-day Pro trial, which starts with its next analysis; when the trial ends, what was collected stays stored, the section is locked again, and new conversations get no customer questions, decline reasons or themes. While your organization still has demo data (it is added at sign-up and removed once your first real conversations arrive, or by hand), the section opens filled with that demo data under a banner, «This is demo data. On {plan} and higher, your calls will appear here», with a See plans button (or a note to ask the account owner, for roles that cannot open billing; in Bitrix24, always the note). Actions that write or call a paid model stay unavailable and say «Available on {plan} and higher plans»: the FAQ draft and editing the theme set; export still works. Without demo data the screen is, as before, a short preview with a See plans button (or the note). Available to owners, admins and supervisors; a supervisor sees only the projects they are assigned to. Reps do not see the section.

The counting unit is the conversation: a complaint repeated five times in one call counts once.

Tabs

  • Sales — a summary of cards on top: top objection, top decline reason, top unanswered question and top theme. “Top” is the row heard in the most conversations — at least 5, and not “Other” or “No theme”; when no row qualifies, the card says “Not enough data yet”. Next to the number is the row’s mark, the same as on the row (an unanswered question has none). Clicking a card opens its section. Under the summary are the sections Objections (the Objections screen: what stops buyers, in customers’ words, and the best answers of your team), Declines and Questions (customer questions and decline reasons) and Themes (your organization’s conversation themes). One section is open at a time; its lists show the first five rows, the rest behind Show all N. The decline, question and theme sections and cards are there while your organization has those insights turned on; the theme section and card also need at least one theme in your set, and the card a conversation with a theme in the period.
  • Marketing — competitors named in conversations and how they were spoken of; signals about prices, demand and market trends; the needs customers named. Each block shows ten rows a page; Next and Previous turn the pages, and an open row keeps its page.
  • Product and service — product feedback at every level: incidents, major issues, minor issues and praise (material for case studies); signals about supply and availability and regulations. Each block shows ten rows a page; Next and Previous turn the pages, and an open row keeps its page.
  • Site and content — what customers could not find before calling. A summary of cards on top: asked most often, in the first minute and theme with unanswered questions (the theme whose conversations most often left a question without a full answer), chosen by the same rule as on Sales. Under it, the sections Frequent questions, First minute (first-minute questions), Themes (your organization’s conversation themes) and FAQ draft (a draft FAQ for your website); lists show the first five rows, the rest behind Show all N.

Voice of Customer tabs: Sales, Marketing, Product and service, Site and content

What changed is a view of the tab: on Marketing and Product and service the All blocks | What changed switch above the blocks; on Sales the last section of the switcher, and on Site and content while customer questions are on.

Summary cards of the Sales tab

Reading a row

On the Marketing and Product and service tabs. The Sales tab’s Objections screen reads as described there; the question and decline blocks read the same way as below, each over its own N (see their sections).

  • “In 12 of 400 conversations · 3 %” — how many conversations of the period mention it, out of the conversations whose analysis collects these details. By default personal and internal conversations do not collect them, so they are not in the 400; customer needs are collected only in sales conversations (inquiries, purchases, follow-ups, scheduling) and are counted out of those. Rows with no conversations stay on the list with their zero.
  • Change — a mark against earlier periods of the same length, by one rule for the screen, the printout and the monthly email. ▲/▼ N pp — the share of conversations moved against the previous period, and the move holds: at least 30 conversations in both periods, at least 5 with the row in the larger one, and a move of at least 3 percentage points and at least half the previous share. The share is compared rather than the count, so a quieter month does not look like a drop in everything. New — the row was not heard in any earlier period counted in full, and is heard in at least 3 conversations now; gone — it was heard in at least 3 conversations of every such period and in none now. A row without a mark keeps its numbers — that is not “no change”. Earlier periods count only within your data retention, and those before the previous period within a year too: when the previous period falls outside your retention, the rows get no mark at all. Until the periods before it are counted in full, rows get only ▲/▼, not new or gone.
  • Fewer than 5 conversations — the row carries a Small sample mark, and an opened row says it in words: for now it is a guess, not a pattern.
  • Open a row to see the conversations behind it — up to the 20 most recent, one wording from each, with a link to the conversation.

Wordings are the analysis retelling what the customer said, not verbatim quotes. Rows are grouped by the fixed fields of the analysis (feedback level, signal type, sentiment); free-text product names are shown under a row, not grouped.

Product feedback block: each row with its conversations, share and line

Over time

Next to a row’s numbers there is a line: the row’s share of conversations across the period — by day for a period up to 31 days, by week for a longer one. A gap in the line is a day or a week without data, not a zero. The rows and their marks do not wait for the lines: the lines load separately, and when they fail the block says so with Retry while the numbers stay. When a row’s card is narrow, as in the Bitrix24 app, the line hides before the numbers.

Lines are drawn for objections, decline reasons, frequent questions, themes, competitors, market signals and product feedback. Questions without an answer, first-minute questions and needs have no line. Objections have a line in the Objections section of the Sales tab, in its market view; the separate Objections screen and the team view have no line. For objections the line is the share among all analysed conversations, not only those with a recognised objection, so its percentages differ from the card’s. A summary card shows its row’s line too, taken from a block of the tab that has already loaded it for the period — the card’s section or What changed, with the row on screen there: the card loads no line of its own. The team view loads no lines; when the card’s numbers come back changed while that block is closed, the line goes until the block is opened again. The cards of questions without an answer, the first minute and the theme with unanswered questions have none.

Open a row to see the same as bars, with a Period | 12 months switch:

  • Period — the bars of the period. Click a bar to keep in the row’s list only the conversations of that day or week — “Listed: X of K”, since the list holds the period’s latest conversations; Whole period brings the list back. In an opened objection a bar keeps only the customer quotes from that day’s or week’s conversations; the objection keeps a few of the freshest quotes, one rep each first, so a day may show fewer quotes than its K.
  • 12 months — the row by month over the last 12 months up to the end of the period, no further back than your data retention; the last month is partial when the period ends inside it. Click a month to see its “K of N” and Show this month, which sets the page’s period to that month.

An opened competitor row: bars over the period, Period | 12 months, customer wordings

A note under the block says when a line is incomplete:

  • There is no trend for this period — the trend has no data for its days; the rows’ numbers and marks are exact. If the block was counted over the most recent conversations only (When numbers are partial), the note says its numbers are a lower bound.
  • Part of the period has no line — there is no data for those days, or it is still being collected.
  • The code set changed during this period — the line has a break.
  • Theme assignment was updated during this period — the line has a break: on a theme’s line and on the lines under a theme filter. A demo theme’s line is hand-made and is not cut.
  • How decline reasons are read was updated during this period — the line has a break: on the lines of decline reasons.
  • How customer questions are read was updated during this period — the line has a break: on the lines of question subjects.
  • Some competitor names were merged by another version of the glossary, or the glossary did not load — the line may be split, and a name the facts do not know yet has no line. Conversations counted before a glossary edit keep their earlier merge, so for days before the edit a competitor’s line may stay split.
  • Some objections were grouped by an earlier label map — the line may be split.
  • The trend is unavailable — the step it is counted from is switched off for the organization, or the plan does not include it; the rows’ numbers do not depend on it.
  • A theme has no line before the date it is counted from — when it was created, returned from the archive or its description was edited: it was not counted that way before.

The lines are counted from a table of the codes and normalised names that each analysis already holds — no conversation text, and it is deleted together with the record. For an organization that had Voice of Customer in October 2026 the table was filled for its whole data retention; for one that gets it later, it starts about a week before that, so earlier days have no line and the 12 months view starts there.

What changed

This view gathers the rows of all the tab’s blocks that carry a mark — ▲/▼, new or gone (see Change under Reading a row) — each with its block’s name, strongest shift first: by how many percentage points the share moved (for “new” — its current share, for “gone” — its previous one), then by the number of conversations. A row opens as on its block, with its line and bars where that block has them. Objection reasons are listed under Customer objections by the same rule: their N is the conversations with a recognized objection — the share a reason shows in its section — while their line is the share among all analysed conversations. A reason gets ▲/▼ and new here, never gone: only the reasons heard in the period are listed. Its history counts the same N — conversations with a recognized objection: an earlier period without one says nothing about a reason, and a period, the previous one included, holding an objection that could not be sorted into a reason leaves “new” unsaid — the reason may have been heard there in that wording. The reason’s row in the Objections section is marked new by the same rule. A block counted from the period’s latest conversations (its count reached the limit) is named under the list: its changes are not marked. With no marked rows the section says so and repeats the rule; under a theme filter without a comparable previous period it says that changes are not marked.

What changed block: rows that shifted against the previous period, here none

Competitors

Competitor names are merged through your competitor glossary, the same way as on the dashboard’s Voice of Customer tab (its Competitors card and the competitor named in Highlights of the period on Summary) — for the same period the counts match. A name missing from the glossary is marked not in glossary, and the share of such mentions is shown under the list with examples. Add those spellings to the glossary to merge them.

For each competitor you see in how many conversations it was spoken of well, neutrally or badly. A conversation where it was praised and criticised counts in both.

Customer questions and decline reasons (beta)

Aelo finds in each newly analysed conversation the questions the customer asked and, for a sales conversation — an inquiry, a purchase or scheduling — how it ended and, when the customer declined, why. It does this in a separate step after the analysis; the step charges no minutes. It happens only while customer questions and decline reasons are on for your organization: on Pro and higher plans they are on by default; switched off, nothing is collected and the blocks are not shown. The blocks are in beta.

  • Data since. Questions and decline reasons are collected only from conversations analysed after collection started; earlier conversations are not processed. Above the blocks the screen says “Data since …” — the earliest conversation of your organization that has them. A conversation you re-analyse is processed like a new one.
  • Collected in X of Y. Above the blocks the screen also says in how many of the period’s analysed conversations questions and declines were collected, and in how many they were not — for example, a call where the transcript does not show which speaker is the customer. A conversation that was not collected is counted nowhere: neither as “no questions” nor in any N.

Frequent questions by subject

On the Sales and Site and content tabs. Each question gets one subject from a fixed list: price and what it includes, payment and financing, delivery and timing, availability, specifications and configuration, warranty and service, documents, comparison, how to buy, and other questions.

A row reads “in K of N conversations”, where N is the conversations of the period whose questions were collected, with or without questions. Two questions on one subject in one conversation count once. The row shows its mark (see Change above) and the three latest wordings; open it for up to the 20 most recent conversations. A change of the model that reads the conversations silences the marks — see Customer questions and a model change under Why customers decline.

Frequent questions by subject block

Questions without an answer

On the Sales tab: the subjects where questions most often did not get a full answer — answered partly or not at all — as a share of the conversations where that subject came up. This block shows no change against the previous period.

Questions without an answer: subjects where reps did not fully answer, with the share

First-minute questions

On the Site and content tab: what customers had to ask straight away, which is what they did not find before calling.

  • Call — a question asked within the first 60 seconds. Aelo times a question by where the customer’s own words appear in the transcript; a question it cannot place that way is not counted here, though it still counts among the frequent questions.
  • Chat — a question the customer asked before the rep’s first message; if no rep wrote, every question that can be placed in one of the customer’s messages. A Bitrix24 open-line chatbot that answers first counts as the company’s first message.
  • Left out — chats where this cannot be told are counted in neither K nor N: chats collected before this rule was introduced, chats without separate messages, and chats where a message whose author’s role is unknown comes before the rep’s first message.

N for this block is the calls and chats of the period where the first minute can be measured.

First-minute questions: what customers ask within the first 60 seconds

Why customers decline

On the Sales tab, over the sales conversations of the period that ended in a decline: “in K of N declines” for each reason. The reasons are a fixed list that shares its labels with the Objections screen where they overlap — price, budget, financing, payment terms, not now, delivery and availability, product does not fit, quality and reliability, trust in the company and service, documents and contract, does not want calls — plus chose another supplier, no need now, not the decision maker, and other.

Why customers decline block

What counts as a decline. A sales conversation ends in a decline when the customer said no to the offer — even only for now or with the door left open (“no money this season”, “not now, call me next year”): the reason, such as “not now” or “budget”, carries the “for now”. A no for now stays a decline even when a next step is accepted along with it — a call agreed for later, a visit or a meeting, an offer to send. What counts is how the conversation ends: a no the rep turned around — the customer then agreed or ordered — is not a decline. When the customer did not say no and left the decision open (“I’ll think about it”), it is not a decline. When the person on the line says someone else decides and no next step is agreed — neither with the one who decides nor through the person on the line — it is a decline with the reason “not the decision maker” once the rep has made the offer (a call cut off before any offer is a conversation without a decision); a promise to consult and call back without an agreed time is not a next step, and someone who only talks it over with another person but decides themselves does not fall under “someone else decides”. This boundary applies to conversations analysed after the update that introduced it (see the changelog); earlier conversations keep the earlier reading, under which a “not now” was mostly not counted as a decline. When you compare a period with an earlier one, part of the growth in declines — most of all “not now” and “budget” — comes from this change, not from your customers. A conversation you re-analyse is read by the current boundary. A next step through the person on the line counts only when no no stands at the end — a no the rep turned around does not count: if they turned the offer down, or pass on the no of the one who decides, it is a decline with the reason given (“not the decision maker” when no reason is given beyond the decision not being theirs — a bare no, “not interested”, “we don’t need it” while sending the rep to the one who decides), even when they agree to pass the offer on to the one who decides; if no no stands, agreeing to pass the offer on is a next step, and so is a visit, a meeting or a call at an agreed time with the person on the line themselves. This reading applies to conversations analysed after the update that introduced it (see the changelog).

Which reason a no for now gets. When the customer names a reason together with a later time, the reason decides: “no money now, call me in spring” is “budget”. “We don’t need it now” with no word that it will be needed later is “no need now”. “Not now” is when the customer says they need it or will need it and will come back to buying later (“we’ll need it, but in autumn”). A later time that only the rep offered and the customer merely accepted does not change the reason. This reading applies to conversations analysed after the update that introduced it (see the changelog): compared with earlier periods, part of the drop in “not now” and of the growth in “budget” and “no need now” comes from this change, not from your customers.

Comparing across a reading update. Every conversation remembers which version of the rules and which model read its outcome and decline reason. If the previous period’s sales conversations were read by one version and the current period’s by another, the decline reasons carry no shift, new or gone marks, and the monthly email says under the reasons: “No comparison with last month: the way we read decline reasons has been updated.” The screen names the reason too: the “Why customers decline” block, the “What changed” view and the “Why customers decline” card on the dashboards say “How decline reasons are read was updated — there is no comparison with the previous period”, and the export and the printable report say the same and leave out the reasons’ previous numbers. A reason’s line breaks where the reading changed, and new and gone are checked only against periods of the same reading — even when the previous period had no declines. The marks return once both periods lie after the update. Conversations analysed before the update that introduced this check (see the changelog) count as one earlier version, so the first periods after it also go without marks on the reasons.

Customer questions and a model change. For the question subjects and the first-minute questions, a change of the model that reads the conversations breaks the comparison — in addition to a change in the set of subjects — but a new prompt revision does not. If the previous period’s conversations were read by one model and the current period’s by another (or one of the periods by two), these rows carry no marks, and the subjects’ new and gone are checked only against periods read by the same model. The shift marks return once both periods are read by one model; new and gone also need the periods before them (up to a year back) to be read by it. When the questions of both periods were read, but by different models, the screen names the reason: the “Frequent questions by subject” and “First-minute questions” blocks, the “What changed” view and the questions card on the dashboards say “How customer questions are read was updated — there is no comparison with the previous period”, the export and the printable report say the same and leave out these rows’ previous numbers, and the monthly email says under the questions: “No comparison with last month: the way we read customer questions has been updated.” A subject’s line breaks where the model changed, with a note under the chart; a new prompt revision does not break it.

Open a row for up to the 20 most recent conversations, each with the reason as a phrase in the customer’s words. When no sales conversation of the period had its outcome collected, the block says that declines were not measured — not that there were none.

Declines broken down

In the Declines section, under Why customers decline, there is one breakdown block with a Declines by switch: call number, rep or stage (stage only when there is one). By call number every row is shown; by rep and by stage, the first five by number of declines.

  • By call number in the series — the 1st, 2nd or 3rd call, the 4th and later, or outside a series (a chat or a call without a number). Missed and short calls take no number, and the number counts across the whole series, so a series that started before the period keeps its real numbers.
  • By rep — the rep responsible for the conversation. A rep gets a row only with at least 3 sales conversations in the period. The declines of reps below that are summed in one line under the list, and declines without an assigned rep in another, so that the rows and the two lines add up to all declines.
  • By funnel stage — the CRM stage of the deal at the time of the conversation; shown only when stage-based scorecard selection is on.

Each row lists its reasons and, separately, no reason named — the declines where no reason was named. Every number opens its conversations: up to the 5 most recent for a number in a breakdown, up to 20 for a reason row, with “Showing the latest X of Y” when there are more.

Declines broken down by the call's number in the series

Who sees rep names and customer wordings

The same people as the section: owners and admins for the whole organization, supervisors for their own projects. Reps do not see the section, the breakdown by rep or the wordings.

A question’s wording is the analysis restating the customer’s question; a decline reason is a short phrase in the customer’s words. Neither is a verbatim quote. Phone numbers, emails, card numbers and IBANs in them are masked, and a wording that may contain an identification number (such as a contract or order number) is not kept — the screen says wording hidden.

FAQ draft for your website

On the Site and content tab, in the FAQ draft section, Build a draft writes questions and answer templates from the frequent customer questions of the period and projects you selected.

  • Only when you press the button. Nothing is built in the background.
  • Not saved. The draft is shown to you and stays while you switch the tab’s sections; leaving the screen or building it again replaces it, and a draft built under a theme is no longer shown once that theme’s definition changes — it was built from the theme’s earlier conversations. Use Copy to keep it.
  • Answers are templates. Fill in the placeholders in square brackets and check every answer before publishing: nothing in the draft is a fact about your business.
  • In your organization’s language.
  • What is sent to the AI model: the question subjects, how many conversations each came up in, and up to 40 of the one-sentence question summaries, masked again for phone numbers, emails, card numbers and IBANs. No transcript, quote, conversation or person is sent.
  • Limits: up to 10 drafts in 24 hours per organization and 5 per person. Building a draft charges no minutes.
  • Pro and higher plans, like the section.

Website FAQ draft block with the Build a draft button

Conversation themes (beta)

A theme is what a conversation is about: a machine model, leasing, delivery to regions. Every organization has its own set of themes, and you keep it. A theme is not the call category (purchase, complaint, support — Aelo determines it) and not the subject of a customer’s question (price, timing): those are counted separately. That is why one theme can show, for example, that the same model is both bought and complained about.

Themes are available on Pro and higher plans, like the section, and work for an organization that has customer questions and decline reasons switched on: the same step assigns them. The feature is in beta.

The theme set

The Themes button next to the section’s tabs opens the set. While the set has no theme in use, the button reads Set themes, and the tabs show no themes section or card. Demo data is the exception: with it shown, the tabs count the demo conversations by a demo set of five themes. That set is not part of yours — it is not in the list, not counted toward the 30, not assigned to your conversations — and it goes away when the demo data is cleared.

  • Up to 30 themes in use. A name is up to 60 characters, the “what it covers” description up to 80.
  • Archive instead of delete. A theme cannot be deleted: it is archived and can be returned. Archived themes do not count toward the 30 and are not assigned to new conversations.
  • Renaming does not affect the counts. Editing the description starts the comparison for the theme over — see reading the themes block.
  • A theme is not a category or a question subject. If a name repeats a call category or a question subject (“complaints”, “price”), the screen warns that this is already counted separately; you can still save it.
  • Who changes the set: the account owner and administrators with access to the whole organization. Everyone else who can open the section sees the set but cannot change it.

The theme set: a starter set from your conversations, suggested themes, adding a theme and the themes in use

A starter set from your conversations

So that you do not start from a blank page, Aelo can propose a set itself — the Suggest themes from history button in the themes panel.

  • What is sent to the AI model: the short conversation topics from the analysis (the “Topic” field) and the call categories of the last 90 days — no transcript, quote, conversation or person; before they are sent they are checked again for phone numbers, emails, card numbers and IBANs, and a phrase in which the check finds one, or that may contain an identification number, is not sent. Demo conversations are not used.
  • What comes back: up to 25 themes with descriptions and the share of conversations each came up in. You edit names and descriptions, untick what you do not need and approve — only approved themes enter the set.
  • No more than once in 24 hours can you ask for a new proposal.
  • No history, or the “Topic” analysis field is switched off — add themes by hand.
  • A proposal charges no minutes.

How themes are assigned

In a newly analysed conversation the same step that finds customer questions picks one to three themes from your set — or marks that none fits (No theme). A conversation in which the step cannot tell which side is the customer gets neither questions nor themes and is not counted anywhere. The names and descriptions of your themes are sent to the AI model together with the conversation.

  • New conversations only. A theme is counted from the moment it entered the set: earlier conversations are not labelled again. A conversation you re-analyse is processed like a new one.
  • Only together with customer questions. While customer questions and decline reasons are switched off, new conversations get no themes — the screen shows the themes of the conversations already processed.
  • With a Uspacy connection. Themes are a Uspacy portal’s to receive only while its processing is active, and they are decided per portal, not for the whole organization. A conversation from a portal whose processing has never been authorized, is paused or is withdrawn gets no themes; conversations from your other sources still get them. The starter set and the weekly suggested themes are prepared from your other conversations: those of a portal never authorized, paused or withdrawn are left out, and the Themes panel names such a portal while it is connected. To bring such a portal back in, an owner or administrator reauthorizes processing under Settings → Consent Management, in the Uspacy organization processing card, with Reauthorize processing; the card shows the version the portal accepted, linked to the text of exactly that version.
  • No minutes are charged. Assigning themes, preparing the starter set and collecting the suggested themes charge no minutes.

Reading the themes block

The Conversation themes block is in the Themes section of the Sales and Site and content tabs.

  • Coverage. Above the rows the screen says in how many of the period’s analysed conversations themes were assigned and, separately, in how many they could not be determined: those conversations are counted nowhere.
  • “in K of N conversations” — a theme’s N is the conversations where themes were assigned while this theme was already in the set. A conversation has up to three themes, so the shares do not add up to 100 %.
  • The change against the previous period is shown only when both periods were counted by the same theme and assigned by the same version of the theme rules — after Aelo updates how themes are assigned, the row says “theme assignment updated”, and the change returns once both periods fall after the update. The new and gone marks are not shown while the update stays in their history, for up to 12 months. A new theme shows “since ” instead, and after its description was edited — “definition changed ”.
  • No theme — the conversations that no theme in the set fits. The larger the share, the worse the set describes your conversations. Its change is shown only if the set has not changed since the start of the previous period and both periods were assigned by the same version of the theme rules.
  • Open a row to see the theme’s description, the breakdown by call category and up to 20 of the most recent conversations — with the call category in place of a wording. On the Sales tab it also shows the average quality score of the theme’s conversations next to the average across all conversations with themes, and the share of declines among its sales conversations; on the Site and content tab — in how many of the theme’s conversations a question was left without a full answer.

Conversation themes block: each theme with its conversations, share and line

Declines and questions without an answer per theme are shown while customer questions and decline reasons are on. The themes block is exported with its tab (see Export).

An opened theme: the line over the period, the split by category, quality and declines, conversations

Theme filter

The All themes selector in the section header, next to the Themes button, narrows the whole page to the conversations of one theme. It appears when themes are switched on in the organization and the list has at least one theme (see What it offers below) — or a theme is already selected.

The theme selector opened: All themes and the picked theme

  • What it offers. The themes of the set that have, in the period, at least 10 conversations and at least 3 % of the theme’s N (the same N as in its row of the themes block). The selected theme always stays in the list, even if the period has moved and it no longer passes.
  • What it narrows. Every block of every tab, in the web app and in Bitrix24: on Sales the objections, the declines and their breakdowns, and the questions; on Site and content the frequent and first-minute questions and the FAQ draft; on Marketing and Product and service all their blocks. The FAQ draft is built from the questions of that theme’s conversations; the daily draft limit is shared, not split by theme.
  • The themes block is not narrowed. It keeps showing the whole set; the selected theme’s row is highlighted, and the theme card of the summary shows that theme’s row.
  • The line above the summary. “Every block of the tab counts conversations with the theme “X”: K of N conversations in the period”, and the Clear button. When the count hit its cap, the line says “at least K”. Blocks with few conversations carry their usual small-sample mark.
  • The theme is in the page address. It stays when you switch tabs and when you come back from a conversation.
  • No comparison with the previous period. If the theme was set — activated, returned from the archive or its definition changed — inside the previous period, the page shows no arrows or changes, and the line says “The theme was set N days ago — there is no comparison with the previous period.” If Aelo updated how themes are assigned and the two periods’ conversations were labelled differently, there are no changes either, and the line says “Theme assignment was updated — there is no comparison with the previous period.” The comparison returns once both periods fall after the update.
  • When the filter is cleared. If themes are switched off in the organization, or the theme is archived or no longer in the set, the filter is cleared with a message. If the server cannot confirm the filter, the tab shows no numbers and says that the theme filter is not available: the whole stream is never passed off as the conversations of the theme. The themes block, which the filter does not narrow, stays on screen.

A theme selected: the line above the summary with the count and Clear

Suggested themes

Once a week Aelo collects the subjects that came up in your conversations of the last 30 days and are not in the set yet, and groups them into at most five candidates. A candidate becomes a suggestion when it reaches at least 10 conversations and 3 % of the period’s conversations, does not repeat a call category or a question subject and is not already in the set.

  • What a suggestion shows: a name, a description, the number and share of conversations and up to three example conversations.
  • Accept — the theme enters the set and starts being assigned to new conversations. Reject — it is not suggested again for 90 days.
  • Who sees and decides: the account owner and administrators. A line about new suggestions appears for them above the themes block.
  • What is sent to the AI model: the short subject phrases the step recorded for conversations — at most 300 distinct ones, checked again the same way (a phrase the check flags is not sent); no transcript, quote, conversation or person. The summary charges no minutes.

Filters

  • Period — today, week, month or quarter, with arrows to step back.
  • Call category — for example, only purchase calls.
  • Projects — one or several.

Filters bar: period, categories, projects and export

The filter is shared by all tabs: switching tabs keeps the period and projects. With conversation themes on, the theme filter joins them.

Export

Every tab exports for marketers, product managers, editors and sales managers who do not use Aelo. The exports are in the Export menu on the right of the filter bar:

  • Table (CSV) — a CSV of the tab: every row with its numbers, the previous period and one wording. The «What N counts» column names the conversations each share is counted over: every block has its own N.
  • Printable report — a printable report of the tab, the same rows as on screen.
  • Customer wordings (CSV) — a CSV with every wording shown on the tab, for ads and copy.

Export menu

The Sales tab exports its Objections section — the objection reasons of the market view, in the screen’s order, each with its conversations, its share of the conversations with a recognised objection (N, as on the screen) and the previous period; the print report adds the lines the screen writes under the list: how many conversations had objections, what part of them was not recognised and what part matched no reason, and what N counts. While customer questions and decline reasons are on, the Declines and Questions sections are added: why customers decline, questions without an answer and frequent questions by subject. Questions without an answer have no previous period, so the previous-period columns of the table are empty for them. The customer wordings include the customer quotes of the objections.

While conversation themes are on, the Sales and Site and content tabs also export the Conversation themes block: “No theme” first, then every theme of the set, each with its conversations, its own N, its share and the previous period. A row without a comparable previous period has empty previous-period cells in the table; in the printed report such a theme shows «since » or «definition changed » in place of the change, as on the screen. A theme with no conversation in its N yet has an empty share. Under a theme filter the themes block is not exported: the filter does not narrow it. On the Site and content tab without customer questions, the export is the themes block — the whole set even with a theme selected, and the files carry no theme.

While the tab loads, the menu is unavailable. When there is nothing to export — the tab’s data did not load, or the tab has no numbers under the theme filter — the menu opens and says why.

Under a theme filter the files carry it: the names of the CSV and print files contain -theme-<theme code>, and the printed report says “Conversations with the theme “X””. Without a comparable previous period the export has no previous-period columns and no trends. The monthly Voice of Customer email is not narrowed by a theme.

Exports contain only the analysis text, which is built from transcripts after personal data is masked, and no rep names. The decline breakdowns of the Sales tab (by call number, rep and stage) and the team view of objections (handling by rep) are not exported.

Monthly email

On the 1st of each month at 08:00 in your organization’s time zone, owners and admins receive a summary of the previous calendar month — competitors, product feedback, market signals, customer needs, and the most frequent question subjects and decline reasons — and supervisors receive the same for their own projects. Questions and decline reasons are counts only, with no customer wording. The numbers are the ones the section shows for that month, and the button opens the section on it. Nothing is sent for a month without analysed conversations. A person with a narrow access that opens Voice of Customer aggregates receives it for the area of their access.

The email follows the weekly report subscription: whoever receives weekly reports receives it, and unsubscribing turns off weekly reports together with this email. A person who holds a narrow analytics access (Settings → Team) and Voice of Customer aggregates receives it instead of the letter of their role, over the area of the access (the whole organization, its projects, or the people of its departments); the questions and decline reasons are included only with access to the text of records, so a Group lead has them and an Analyst does not. The email is sent on Pro and up by the organization’s plan: below Pro it stops even while the access is active, while the daily and weekly digests continue. During the Pro trial of a free organization the email is sent as well.

When numbers are partial

  • Most recent conversations only — when a period has more than 10,000 conversations with customer signals, part of the section is counted over the most recent 10,000, and the screen says which part. If the screen says the wordings (on Marketing, also competitor sentiment) come from the latest 10,000, the rows’ “K of N” cover the whole period. Otherwise the block’s numbers are a lower bound: narrow the period for exact numbers. The lines over time are not limited to 10,000 conversations.
  • Glossary not read — if your glossary could not be read at that moment, spellings of one competitor are shown separately and the screen says so.

See also: Objections, Dashboard.