peaksly explains · 10

Services

Monitoring or Deep Audit: which one when?

Both services count mention, observed position and source with a denominator. What differs is place, cadence and depth: the Deep Audit reads the visible interface by hand, once and broadly across six answer surfaces; monitoring measures continuously via API paths, many search prompts, every day. The question that is open right now decides the tool.

6-minute readUpdated 5 Sep 2026with a fictional example

The switch · seven situations, three tools · Brückner LaborgeräteFictional teaching example, not a measurement

1First assessment, no actions yetWhere does the brand stand today, on which surfaces, with which sources?Deep Audit

2Relaunch of the product pages in Octobera baseline on the visible interface beforehandDeep Audit

3Keeping seven competitors in view continuouslywho stands ahead of the brand, and whether that stays soMonitoring

4Several markets, many prompts, a report every montha scope that cannot be read by handMonitoring

5Checking the effect of the changes from Novemberthe same prompts, day by day, with control measurementsMonitoring

6Understanding the spread of the same question across six answer surfacesorder and sources per surface, read by handDeep Audit

7Looking once during ongoing operation at what customers read on the interfaceone pre-defined run, not a second Deep AuditSpot check

Deep Audit 3Monitoring 3Spot check 1of 7 situations

Decision · Brückner Laborgeräte

October: Deep Audit as the baseline before the relaunch
from November: Monitoring, 50 prompts, one market, a report every month
February: Spot check, one run on the interface

Both together, one after the other: first the baseline, then the time series.

Brückner Laborgeräte, the situations and all figures are fictional. Scope, measurement path and cadence of a real service are set out exclusively in the accepted written proposal.

In short

Two tools, two questions.

The Deep Audit answers where a brand stands today on the visible interface: twelve prompts on six answer surfaces, read twice by hand, 144 documented observations. Monitoring answers what changes: 50, 150 or 350 prompts via API paths, daily with two control measurements a month, one report per month. In the example, Brückner Laborgeräte uses the Deep Audit before the relaunch as a baseline and monitoring afterwards for the time series.

Step 1 · Two toolsFictional example: Brückner Laborgeräte

Same values, different place, different cadence.

Both services read the same three values from every answer: mention, observed position, source of the mention, each with a denominator, point in time and measurement path. What sets them apart is in the table: the Deep Audit reads the visible interface, where customers read and click, on six answer surfaces and by hand. Monitoring queries three AI search engines in the agreed core via their technical interface, every day.

An API measurement path is not a visible consumer interface. The same question can deliver different orders and sources on the same day via the two paths. That is why peaksly keeps both as separate time series and never offsets them; the article on answer surfaces shows which surface has which measurement path.

No substitute for each other. Whoever knows only the interface sees no course; whoever measures only the interface via API does not see what customers read.

Deep Audit and monitoring · six rows, two columnsScope as on the service pages

Deep Auditone-off single serviceMonitoringmanaged plan

Question: Where does the brand stand today?; What changes over time?
Place: visible interface, six answer surfaces; API measurement path, three AI search engines in the agreed core
Scope: 12 × 6 × 2 = 144 observations; 50, 150 or 350 prompts × 3 per measurement round
Cadence: once, two runs on documented dates; daily, plus two control measurements a month
Hand: read by hand, every row; measured by machine, anomalies checked by hand
Result: report with warning cases, source picture, comparison of the six surfaces, at most five actions; monthly report with values and development, competition and source picture, at most five actions, raw-data annex

What stays the same

the same three values, each with a denominator
point in time and measurement path on every value
interface and API never offset
at most five actions per report

Step 2 · SnapshotFictional example: Brückner Laborgeräte

The Deep Audit is a baseline, not a course.

Brückner plans the relaunch of its product pages for October. Beforehand, the Deep Audit reads twelve prompts on six answer surfaces twice by hand: 144 observations, each with prompt, surface, run, point in time and review status. Mention in 31 of 144, three warning cases opened by hand.

That is a state for one day, broad and deep, but without a course. How such an audit comes about is shown in the Deep Audit example; every surface needs a passed approval gate beforehand, otherwise the scope is agreed with its own denominator.

Baseline · October · visible interfacefictional

Scope: 12 prompts × 6 answer surfaces × 2 runs = 144 observations
Mention: 31 of 144, per surface 3 to 7 of 24
Warning cases: 3 opened by hand: location, source, identity
Read: every row by hand, on two documented dates

Source of the mention · in the 31

own page: 9 of 31
trade media: 11 of 31
industry profiles: 8 of 31
forums: 3 of 31

Step 3 · Time seriesFictional example: Brückner Laborgeräte

Monitoring is a course, not an interface.

From November, 50 prompts run via the API paths of three AI search engines, every day at a confirmed daily cadence: 150 answers per measurement round. On two days a month, at least seven days apart, the run is repeated as a control measurement. In the example month that makes 32 measurement rounds and 4,800 answers.

The strip shows a single value from that: prompt 7 on AI search engine 1, one day one cell, a mention on 19 of 30 days. The monthly report summarises all 50 prompts; how fluctuation is separated from trend is explained in the article on denominator and points in time.

Control measurements are not extra days. They repeat a run on the same day so that fluctuations and measurement errors become visible. Cadence and control measurements are binding in the accepted proposal.

November · prompt 7 on AI search engine 1 · API path · one day = one cellfictional

no mentionmentioncontrol measurement, daily value confirmedmention on 19 of 30 days

Month · 50 prompts × 3 AI search engines

Rounds: 30 days + 2 control measurements = 32
Answers: 32 × 150 = 4,800 in the month
Prompt 7: mention on 19 of 30 days, AI search engine 1
Control: 2 of 2 confirm the daily value

One time series per measurement path, one denominator per day; no value from the interface.

Step 4 · SequenceFictional example: Brückner Laborgeräte

First the baseline, then the time series, then the spot check.

Brückner uses both tools one after the other: in October the Deep Audit as a baseline before the relaunch, from November monitoring for the time series, in February a pre-defined spot check on the visible interface, in April the second Deep Audit against the baseline.

Only like is compared with like: Deep Audit against Deep Audit with the same twelve prompts on the same six surfaces, monitoring against monitoring with the same 50 prompts on the same measurement path. The spot check is an add-on subject to capacity, described on the pricing page, and not a second Deep Audit.

One year, four stations · Brückner Laborgerätefictional

OctDeep Audit · baseline
NovMonitoring · 3 monthly reports until January
FebSpot check · one run on the interface
AprDeep Audit · comparison with the baseline

Every station carries place, measurement path, denominator and point in time; the sequence is a decision, not part of a contract.

Like is compared with like

Deep Audit against Deep Auditthe same 12 prompts on the same 6 surfaces, interface
Monitoring against monitoringthe same 50 prompts on the same API path
Spot check against Deep Auditonly on the interface and with its own denominator
Interface against APInever; two time series, kept separate

Honest limitFictional example: Brückner Laborgeräte

What no tool promises.

Neither the Deep Audit nor monitoring promises a mention, a position or a trend. The Deep Audit measures only on surfaces whose approval gate has been passed; monitoring measures only via the AI search engines, the measurement path and the cadence set out in the accepted proposal.

What stays checkable is the same for both: place and measurement path on every value, one denominator per tool, one point in time per observation and warning cases that a person has opened. The values themselves are explained in the fundamentals article.

Not promised

that one tool replaces the other
that API values show what customers read on the interface
mention, position or trend as a result
a Deep Audit on a surface without a passed approval gate

What stays checkable

place and measurement path on every value
one denominator per tool
one point in time per observation
warning cases opened by hand

First the question, then the tool.

The Deep Audit reads the interface once by hand. Managed monitoring measures the same prompts continuously via API paths.

View the Deep Audit Plans and pricing →Look up the terms →