Read website and shop insights

See the widget journeys for examples of the requests, bookings and counter results behind these reports.

Screenshots and counts use fictional example records, not customer results.

For: owners and authorised team members checking website traffic, shop orders, QR codes, visitor agents and widget results.

Where: Customer services > Analytics. You need View analytics. Owners and Admins also have Configure analytics by default.

In a customer workspace, open Analytics from the business menu. Your plan must include analytics. Counter staff cannot open it.

Compare the results from each widget placement. Open a result to see the records you can access. Different widget types have different purposes, so compare placements within the same type of work.

Accepted requests appear even when audience measurement is off. Turn it on to add page counts and permitted browser activity.

Read the answer first

Most sections with data start with one sentence. Choices, Content, Decisions and Audience have none. Pepper builds it from the figures of that section only. For example: "Table 07 leads on Orders paid: 41 of 63. Opens rose by 18 compared with the previous period."

Select a number in the sentence to open its records, its filtered report or its definition. If a figure is missing or withheld, Pepper leaves that part of the sentence out. The sentence never names a cause.

Below the sentence, a section can show one next action, such as Open orders waiting for payment. Pepper shows it only when you may open its destination. Click × beside the action to hide it for the selected period. Pepper stores this choice for your account only. Other people still see the action, and other dates can show it again.

Pepper checks the rules of a section in table order and shows the first that applies.

A website action opens the website editor at that page, language and field. It needs permission to edit website drafts.

Compare with the previous period

Every comparison uses the previous period of equal length. Under a figure, a chip shows the change in percent, such as "↗ 12.5%", and was shows the earlier value, such as "was 38". The chip is green when the figure moved the good way and red when it moved the bad way. More refusals, cancellations, no-shows or refunds count as the bad way. Money shows the direction and the earlier amount in the same currency, without a percent. Charts use one color per kind of figure: blue for views, green for results, violet for bookings, teal for money, orange for offers and pink for visitor agents. The previous period is a lighter shade of the same color behind the current one. A rate names both counts, such as "64% of 50 · previous 58% of 43".

If the earlier value was withheld, the line reads Previous period withheld. If no earlier value exists, no line appears. If Pepper withholds the whole previous period because it was too small, the Overview figures show no line. Pepper never shows a missing value as zero.

Know what kind of figure you read

Select a figure's class to read its definition. How these figures are measured in the report menu lists every definition.

Turn on audience measurement

1. Open Analytics settings.
2. Check Enable audience measurement.
3. Keep Respect Do Not Track and Global Privacy Control checked to exclude those requests.
4. Explain Why you collect these figures.
5. Enter the HTTPS address of Your privacy notice.
6. Click Save settings.

Collection starts off. Known bots, previews and your installation's signed-in staff sessions are excluded. Your server must run the Laravel scheduler.

At a widget's own address, Pepper counts successful page responses on the server. The page asks for separate permission before measuring browser actions. That permission lasts for this page load. Visitors can stop measurement from the same control.

Switching collection off stops new audience activity. Accepted service records still appear. Earlier counts remain until their retention period ends.

Read website traffic

Open Websites in Analytics. The opening sentence describes the selected dates and the largest supported change. Select a phrase or figure to inspect it.

Page views counts successful published page requests, including reloads and cached responses. Arrivals counts requests without a referrer from the same website. Missing referrers can increase arrivals. Neither figure counts people or visits.

Select a website, page or language to narrow the report. Page, source and campaign rows open that same scope. Leads and Orders open permitted business records.

The daily chart compares the previous period of equal length. Publication and campaign marks explain changes in the website's history. Read daily values opens the table behind the chart.

Today's usual range uses recorded activity from the same weekday in the previous 8 weeks. It needs at least 4 recorded weeks. The comparison stops at the same completed quarter-hour each day.

Small groups stay hidden, including additional figures that could reveal them by subtraction. Country, device and arrival-source filters cannot identify the people behind business records.

Website counting requires no snippet or browser consent event. Do Not Track and Global Privacy Control always exclude website and shop request counts. Signed-in staff, known bots, previews, redirects and failed page requests do not count.

When collection is off, the report keeps permitted leads and orders. An owner can enable counting in Analytics settings.

See what a website led to

Each website row shows its leads and orders. With See balances and figures, it also shows invoiced money for each currency. A result belongs to the last website page before it.

When people visit shows page views by weekday and hour in the business timezone.

Page codes ranks the arrivals from each QR code for a page by code name, with the previous period. It shows every count exactly, also 1 to 4, because a code counts scans of a printed code. The placement register shows the same figure. The arrival also carries the code's channel. When the code has a campaign, Campaigns counts the arrival under it.

Select one website to see more. Pages that earn, Found through search and Read by assistants need audience measurement on.

A group of 1 to 4, and the group that would reveal it, shows Withheld. Each page's search arrivals also show the value from the previous period.

Open the same figures elsewhere

In Websites, choose Analytics from a row's menu or select its page-view count. The count covers the last 30 days.

In the website editor, open Insights. The selected page and language narrow the figures. Choose Whole website to widen them. Unsaved edits do not change published-page counts.

See in Analytics opens the same scope in the full report. Campaign links belong to the selected website, even when several websites share the installation's hostname.

Follow shop orders

Open Shop in Analytics. It combines permitted checkout and order records with optional shop request counts.

Shop shown counts successful shop or buy-button frame requests. It does not prove that someone read a product. Additions to a cart belong to a separate, consented measurement lane.

The funnel starts with checkouts created during the selected period. Later steps count those same checkouts before the period ends. Select a step or stopped count to read its daily totals. The report exposes no cart identity.

The headline order figures use placement and payment dates. They can differ from the checkout funnel when a checkout crosses the period boundary.

Unfinished checkouts separates Not placed from Placed, not paid. Open an available order to follow it up. recovered after expiry counts checkouts paid after they expired. It does not claim that a reminder caused payment.

Expired checkouts lose their personal details. If those missing details prevent product filtering, the funnel says it is unavailable. Optional abandoned-product counts survive as small-group-protected totals.

Products show ordered quantities, current stock and permitted revenue. Open a product or its figures to inspect the records. The product page's insight strip uses the same 30-day report.

Harbour Salt's demo stock is 13 prints left from an edition of 49. The shop report counts online orders; the stock balance also includes the studio's other sales.

Money appears only with See balances and figures and access to its source records. Each currency has its own amounts. Invoices and recorded payments supply revenue once; orders add no second ledger total.

The CSV export and the read tool carry average order: net invoiced revenue divided by orders placed in the same period and currency. Credits can reduce it. A product filter shows its saved invoice-line amounts; Pepper does not guess how to split a payment across products.

Refunds and returns use their own recorded dates. Waiting-for-stock figures require owner access. See sell through a shop for fulfilment, downloads and returns.

Read agent carts, customers and fulfilment

The CSV export and the analytics read tool also group shop orders by website page, code, campaign and channel, with invoiced money when you have See balances and figures.

The refund figure names both counts, such as "3 refunded of 41 paid orders". Shipping countries come from order addresses. Pepper hides a country with fewer than 5 orders.

Money compares each currency with the previous period. Pepper computes the difference exactly and shows no comparison for a currency without earlier orders. No cart, address, contact or customer identity appears in the report.

Start with one widget

In Widgets, use search, type and status filters to find a widget. Sort by Page requests, Measured loads or Interactions. The order compares this month’s recorded activity across the whole list. On a phone, use Sort widgets. The date range uses the business timezone.

Each small chart shows measured loads by day. Focus a chart. Use the arrow keys to read exact values. Home and End go to the first and last day. Requests, loads and interactions are separate counts, not people. Interaction types depend on the widget, so a larger count does not mean a better conversion rate.

Choose Analytics from a widget’s overflow menu or right-click context menu. The report opens with that widget selected; Back retains the library’s filters, sort order and page.

The first section states the widget’s purpose, its primary result, the previous period and the next useful action. Its campaign and share-channel rows compare the same measure. Open an exact result to read the permitted records behind it. Review widget and publishing kit returns to that widget’s editor.

Server page requests and consented browser loads are separate populations. Neither counts unique people. A booking confirms a reservation, not payment. The report keeps money in its own currency and uses recorded payments for cash collected.

An empty report invites you to review publication and links. With collection off, exact service records remain available within your access; optional browser counts do not appear as proof of audience. Restricted result figures remain unavailable. Country or device filters suppress identified results; small anonymous groups can also be withheld. Clear those filters to return to permitted service results. Unrelated widgets’ targets stay outside a selected widget report.

Connect an existing website

Approved widget origins appear under Websites when you enable collection. An origin is the scheme, host and port, such as https://studio.example. It has no page path.

1. Open the website's row in Analytics settings.
2. If its pages already load a Pepper widget, keep the existing loader.
3. For a page without a widget, copy Website snippet into that page once.
4. Connect the site's consent control as shown below.
5. Approve the shared page paths and campaign labels you want in reports.

Use Add website only for a website whose pages you control. This does not give that website permission to embed a widget; each widget still has its own approved origins.

After a visitor permits analytics, dispatch this event in the host page:

document.dispatchEvent(new CustomEvent('pepper:consent', {
    detail: { analytics: true }
}));

Send the same event with analytics: false when permission is withdrawn. The loader collects nothing without this signal. Your consent control must make the choice; loading the snippet alone is not permission.

Blocked scripts and withheld permission leave embed and website views uncounted. Standalone page requests still count when audience measurement is enabled. The report cannot reconstruct missing browser events.

Approve shared labels

Open the website row to add a Shared page path, such as /menu or /projects. Use public paths without names, addresses, account details or personal codes. Unapproved paths appear as unknown.

A campaign has a display name and shared labels for source, medium, campaign, content and term. Campaign label is required. Approve the exact combination used in your links; other combinations stay unknown.

For example, approve source neighbourhood, medium print and campaign autumn-window. Use those same utm_source, utm_medium and utm_campaign values in the link. Leave content and term empty unless you approved values for them.

Do not put a visitor's name, email or personal code in a path or campaign label. Sharing keeps an active placement and approved campaign, then records the chosen share channel. It does not identify the sender. The campaign’s medium label stays as approved. The actual delivery website stays separate from the campaign’s originating website. Use the publishing kit to generate links, QR assets and all three embed presentations.

Read the overview

The report starts with one answer, such as the placement that brings the most results. Below it, the sections divide the report by question:

A section appears only when the report has its data.

The address keeps the open section, so a copied link opens the same section. Changing a filter keeps the section open.

Open the date control. Choose From and To, then Apply. The period uses your business timezone and can cover up to 366 days. Its comparison covers the same number of preceding calendar days.

The leading placement is the one with the most main results within its family and measure. A tie stays a tie. The placement register keeps print versions and channels separate, so changing a destination does not move earlier results to the new widget.

Select a widget, family or placement to narrow the report. Open a placement to compare its print runs and channels. Further filters cover website, page, source, referrer group, campaign, country and device. Click Report options to change them. Each active filter appears as a chip beside the dates; click its × to remove it. The URL retains the selection. Save this view keeps it for your account; Remove view deletes that saved selection.

Tap a figure's explanation to read its basis. A served contact file counts as a download; it does not confirm that a contact was saved. Repeated page requests can count again.

Click an exact figure to open its records. Restricted records are absent from both the figure and its list. Results with no known placement keep an Unknown placement row.

Country and device filters apply to counted activity. Pepper does not attach those facts to a person's records, so exact results are unavailable under those filters. Small groups and revealing totals are hidden. Countries remain unknown if the server has no local GeoIP support.

Daily charts compare the selected and previous periods. Arrow keys read each day; Read daily values opens the values as a table. The hour grid uses the business timezone and has an equivalent table. When the whole period is empty, the report says Nothing recorded in this period. The hour grid, countries and devices say it too. Other charts and lists with nothing in the period are left out.

When the period includes today, a dot on the period button shows that the screen refreshes itself with views and results from the last five minutes. Recent accessible results appear as links at the end of the report. On Websites, Right now at the end shows the page views of the last five minutes across all websites. To hold the screen still while you read, choose Stop live refresh in the ⋯ menu. This stops only this screen from refreshing. Pepper continues to count every visit, and the figures catch up when you choose Start live refresh or reload the page. Background tabs follow Pepper's shared polling rules. Rollup charts update through the hourly scheduler; they are not a live visitor trail.

Read content results

For a case, updates widget or guide, open the Content section. Each exact count opens the records you may read.

Tags can overlap. A request with 2 tags appears under both tags but counts once for the case. A visitor who subscribes to 2 topics gives 2 topic consents. An unsubscribe in this period can refer to a subscription from an earlier period.

These figures count events in the selected period. Open the widget's Records and requests for its current active subscriptions. Published steps, tags and versions come from the saved record; later edits do not rewrite them.

Counted case, post and step activity keeps its own population. The last observed step in a measured load is not proof of completion. Questions, acknowledgements and unsubscribes are separate actions, so they do not form a conversion sequence.

Export CSV and the analytics read tool include the same breakdowns. Country and device filters hide exact records, including these breakdowns.

Read menu and choice results

Select a menu, estimator or service-area widget, then open the Choices section.

Requests use their received date. An estimate can change several times within one load without counting as several shown estimates.

Sent per shown mixes exact records with consented counts. It is not a visitor conversion rate and can exceed 100%.

Quote comparisons start with estimator requests received in the selected period. They include linked quotes accepted before the period ends, or now if earlier.

Each comparison uses the saved estimate and accepted quote total, including tax, in the same currency and scale.

The difference is the accepted quote value minus the original estimate. Versions and currencies remain separate; drafts and cancelled quotes do not contribute.

These comparisons require access to each quote and See balances and figures. Business accounts do not see Platform quotes.

Open exact counts and values to read their permitted records. Country and device filters withhold identified results.

The area country describes the code the visitor checked. It is separate from the anonymous country classification used in audience filters.

Anonymous area checks retain no postal code. A sent request keeps it for follow-up. Export CSV and the analytics read tool include these breakdowns.

Read counter results

For offers, start with confirmed redemptions. The overview names the placement with the most redemptions, even when another placement has a higher rate.

For example, Canal Coffee's window brings 71 redemptions from 96 claims. Its WhatsApp link brings 18 from 21 claims.

The window brings more visits. WhatsApp has the higher redemption rate. These figures answer different questions.

Redemption rate follows the claims acquired during the selected period. It counts their own redemptions before the period ends, or now if earlier.

A correction removes that redemption from the rate when it falls before the report cutoff. Later corrections do not rewrite an earlier report.

Daily counts place confirmations and corrections on their own dates. A correction contributes minus one, so a day can have a negative net result.

Open the Offers section for these figures and the stamp-card grid:

A dash means there are no matching records for a time calculation. Open an exact figure to inspect its permitted records.

The stamp-card grid groups cards by their joining month. Each cell counts cards reaching that stamp in their first cycle, after corrections.

Newer cards have had less time. A dash marks a stamp that no card could yet reach under the saved minimum interval.

The pause summary shows where unfinished cards have stopped. It does not guess why someone has not returned.

Trace revenue to its source

Revenue attribution belongs to the platform workspace. Customer workspaces show their own widget outcomes, without the provider's invoices or payment figures.

With See balances and figures, the section starts with the invoiced total, such as "€4,200.00 invoiced in this period. Most of it came from Search engines." Pepper names a source only when it brought more than half of the invoiced money in the first currency. Unknown never leads.

From enquiry to payment shows Last touch first, with First touch beside it. The two models are alternative explanations of the same records. Do not add them together.

Money requires See balances and figures, plus access to each document. Drafts, cancelled documents and reversed payment records do not contribute. A credit note reduces invoiced revenue once. It changes cash only when a refund is recorded. Cash can be negative when refunds exceed payments in the period.

1. Choose your dates and source filters, then open the Revenue section.
2. Use Revenue currency to select a currency, or keep All to show separate amounts. This selection changes money only; lead and quote counts still cover all currencies.
3. Open Revenue by source. Group by placement, widget, page, source, campaign or share channel.
4. Compare last and first touch. Unknown includes records with no eligible known source.
5. Click a count or amount to open the records behind it. Money rows show their signed contribution and reporting date.

Pepper uses the known source on accepted requests explicitly linked to one person. It looks back 30 calendar days from lead creation, quote acceptance or the moment an invoice was issued in Pepper. For an invoice, the attribution window ends at that moment even if its printed issue date differs. The revenue report still uses the printed issue date. Pepper does not link people by a shared email address or pool a company's contacts. Cash, credit notes and refunds keep the source of their original invoice, even when settlement happens later. A later enquiry cannot move an earlier sale to another campaign.

The source filter applies after first or last touch is selected. The two models can therefore show different totals for one placement. Across all sources, including unknown, each model adds up to the same total for each currency. A currency is never converted or added to another currency.

On a lead or deal, open Where this record came from to see known sources from the 30 days before that record was created. These facts remain separate from your editable source tag. Reading them does not grant access to Analytics. Removing a linked record or changing its access can change the reported source; it does not change an issued invoice or recorded payment.

Country and device filters make identified revenue unavailable. Anonymous views are not matched to sales. These figures show recorded work and payments; they do not establish that a placement caused a sale.

Compare QR codes and sets

Open Placements. The register lists every code in your sets, including codes that nobody opened yet. The tab shows how many sets recorded activity in the period. Make sets in QR studio.

1. Compare sets side by side, or use By widget family.
2. Open a set to see Codes in that set.
3. Sort with Most opened or Fewest opened, or narrow with Show codes by tag.
4. Select a code or set to see Opens per day and when it is opened in the week. The week grid appears from 5 opens in the period.
5. Select Opens or Results to open the records behind them.

Codes without opens or results fold into one row, such as Not scanned yet: 3 of 24. Click Show them to list them. Closed marks a closed code. Each set and code also shows its opens, results and paid money from the previous period.

Single codes lists every code outside a set, most opened first: a code from a widget's share kit, a page code of your website and a code that leads outside. Click a name to filter the report to that code. Scans count only while audience measurement is on. A code that opens a widget, a card or a page of your website counts once, at that page. A code that leads outside opens a web page, a file, a place or an app. Pepper counts its scans at the redirect. It shows no result, and records that follow it show source Unknown. Wi-Fi and contact codes hold their content in the code itself. Their scans never reach Pepper, so the register lists them under Codes that cannot be counted.

See what visitor agents did

Open Visitor agents. The tab appears once a visitor's own AI agent has called the visitor door. The answer states the calls, the change from the previous period and the most common refusal.

Records and calls are separate. Pepper never divides records by calls. Agent calls never count as human views or arrivals.

Agent calls are operational counts. Pepper counts them even while audience measurement is off and keeps no IP address, device, token or other identifier. If agents made fewer than 5 calls in the period, Pepper withholds the details. Pepper also withholds a cell below 5 and the cell that would reveal it.

Agent calls carry no page, code, campaign, source, language, country or device. A website appears only on a page read. If you filter by one of these, the tab says so. Remove that filter to see agent calls.

See figures where you work

Some screens show the same Analytics figures for the last 30 days. Each links to the full report with the same filters. They need View analytics.

The Reach instrument compares each figure with the previous 30 days. It adds testimonials and waiting proofs when you may read them. It stays hidden while audience measurement is off. See desks.

Set a target

1. Open Analytics settings.
2. Under Goals and targets, choose a Widget and Measure.
3. Click Add goal.
4. Choose the goal under Set a target.
5. Enter a positive whole Target count, From and To.
6. Click Save target.

The overview shows progress, what remains and the end date. A target counts its widget's results across all placements. The overview shows targets only when no filter or the widget filter is selected. A filter for a code, set, website, campaign or channel hides them. It changes no request, price or offer. Exact progress counts only records you can open.

Remove target removes that planning target. Remove goal takes the goal off the active report; it does not delete service records. Re-adding the same widget and measure restores its goal.

Export and keep the data

Export CSV uses the current report filters and your access. The analytics read tool uses the same report. Both carry the figures above. The CSV export adds the answer sentence for Websites, Shop, Placements and Visitor agents. The read tool returns the answer of each section that has one. This includes website pages, shop sources and fulfilment, QR sets and codes, and visitor agent calls and records. A withheld value exports as an empty cell with its reason. A saved view never grants extra record access. Revenue rows carry the attribution model, currency, scale and exact signed minor units. Keep long minor-unit strings as text when importing a CSV into a spreadsheet so the spreadsheet does not round them.

With the scheduler running, raw audience events and page-load identifiers are removed within 24 hours. Aggregated counts remain for up to 25 months. Full export and import carry those counts, goals, targets, saved views and operational source facts. Short-lived raw events and load grants are omitted.

An identified request's recorded source belongs to that request's personal data. Its access, data-subject report and erasure rules apply. Removing raw audience events does not delete a customer's service record.

Understand the privacy boundary

Anonymous pages set no cookie, use no browser storage and request no third-party assets. The server reads an address only for coarse country classification and short-lived abuse protection, and a user agent for a coarse device class and bot filtering. Neither raw value is stored by public collection.

For consented widget interactions, a random identifier in page memory connects the permitted steps of one load. Website page counts have no load identifier, visitor hash or fingerprint. It expires and cannot recognise a later visit. Pepper therefore does not report unique visitors, visits, anonymous returns, exit pages or time between visits.

The operator chooses the purpose, privacy notice and consent flow. This method does not establish a legal exemption. Your hosting provider or proxy may keep access logs separately; configure their retention for your installation.

If a record panel fails to load, close it and open the figure again. An invalid settings field keeps your input and explains what to correct. An empty first month points back to publishing a widget and giving each placement its own link.

Read responses to listings

Choose Listings as the family, or select one listing widget. Open the Listings section. Listing journey shows counted views and photo views beside exact enquiries and confirmed viewings for the selected dates. These are separate populations; they do not trace an anonymous person through a funnel.

The listing rows show enquiries, viewing requests, confirmed viewings, applications and shortlists. An enquiry rate uses exact enquiries divided by counted widget views. The label names both populations. Photos viewed per measured view uses consented photo views and widget loads.

Days to first enquiry starts at the first publication. Price intervals show each public price, its enquiries, days of publication and enquiries per day. The calculation excludes unpublished time and uses the business timezone, including daylight-saving changes.

A missing count stays unavailable. Exact results obey your record access; a rate stays unavailable when you cannot read its whole enquiry population. Each currency stays separate. CSV and agent reads use the same values. See publish listings and follow up on requests.

Compare bookings with capacity

Select the bookings family to read confirmations, cancellations, no-shows and accepted waitlist offers. Open the Bookings section for capacity. Open the linked records to understand the result.

Bookings and capacity compares each service and schedule using its own unit: capacity-minutes, resource nights or event seats.

Availability is recorded once per schedule and day, or when future time is first booked. Later edits do not rewrite that basis.

Recorded availability excludes closures and outside busy time observed then. Minimum notice and daily cut-off limits do not reduce configured capacity.

Cancelled bookings do not occupy capacity. Completed bookings and no-shows keep their occupied time.

If days were not recorded, the fill rate is unavailable. Restricted roles and source filters also withhold the whole-schedule percentage.

Clear source filters to compare a complete schedule. Shared schedules appear per service; their capacity figures must not be added together.

Average booking notice uses acceptance time and the original start. The report states its population and observation time.

Bookings do not mean revenue, payment or attendance. Read the booking workflow for completion and invoice drafts.

If you can read money totals but cannot read their source widget, Analytics groups that source as Unknown. The permitted financial totals stay unchanged.

Measure customer decisions and job times

Choose the Customer decisions family, or a proof or job widget, then open the Decisions section. The report lists permitted records published in the selected period. Follow a record link to inspect the decisions and history behind its figures.

Proof figures show seconds from publication to the first decision and approval, change requests, time waiting and whether the current unapproved version exceeds its waiting target. Comments alone are not decisions. A new version starts its own figures; an earlier approval does not carry over.

Job figures show seconds in each recorded stage interval, total job seconds, waiting questions and answer times. Closed job intervals do not accumulate time. Corrections add intervals without changing earlier records. Reopened questions keep answers in separate rounds.

The report stops outcomes and elapsed times at the period end or now, whichever is earlier. Clearing placement, campaign, channel and other audience filters makes these operational figures available. They support dates, widget, widget type and family filters. They are unavailable in customer workspaces.

Linked quote attribution counts quotes accepted after a recorded visit from that job page. This sequence does not establish what caused acceptance. A quote must belong to the same customer and remain within your document access. The job page does not accept it.

These figures are exact records and UTC seconds. They do not connect anonymous visitors or prove payment. See proof decisions and job questions for the workflow.