maitiq guide
Google Ads software or managed service: what really takes the load off your team?
maitiq · Published
The decisive question is therefore not «Which AI is better?», but: who configures goals, monitors data quality, checks platform defaults, assesses alerts, approves changes, documents decisions and responds to errors? A fair comparison covers the licence or service fee, internal operating time, setup, integrations, cover during absences, governance and exit. Anyone who compares only the visible price is measuring different scopes of service.
What is Google Ads management software?
Management software connects to one or more advertising accounts and supports monitoring, rules, budgets, audits, recommendations, reporting or approved changes. The spectrum ranges from a specialised dashboard to a comprehensive platform for the ongoing operation of several accounts.
The licence, however, does not introduce an operating process. The company needs at least one person who:
- defines the business goal and the primary conversions;
- monitors data quality and integrations;
- understands rules, warnings and recommendation criteria;
- links alerts with price, margin, availability and brand rules;
- approves changes and tracks results;
- handles false alarms, exceptions and product changes;
- organises documentation, cover and the provider exit.
Software can reduce the amount of work and make more account data visible. It cannot take over an unstaffed role. A dashboard without a responsible person creates a second inbox. In maitiq’s managed service, this role is staffed: a Client Success Manager leads the operation. Actionable alerts arrive as a proposal with the action, the rationale and supporting data; diagnostic alerts and findings about missing data remain visible as such – not every alert becomes a proposal.
What maitiq takes over as a managed service
A managed service combines technology with a named operating process. At maitiq, recurring workflows review account and performance data at a configured cadence, daily by default – budget pacing and budget reallocation, target CPA adjustments, search term and exclusion maintenance, and ad review – and prepare proposals with the action, the rationale and supporting data; old and new values are stated where they are meaningful. Reporting runs as a separate evaluation and does not produce proposals. A Client Success Manager supports setup, priorities and reviews; your team provides business context and decides. A limited autopilot is switched off by default; when it is explicitly configured, that standing authority can accept eligible proposals within the limits you set and implement them through the controlled, logged implementation process – anything beyond those limits waits for a decision.
What automation does Google Ads already deliver?
Google automates central platform tasks before you buy anything on top. Smart Bidding uses Google AI to align bids in every auction with conversion or value goals. Performance Max is a goal-based campaign type across several Google inventories. For Search campaigns, Google additionally documents AI Max: according to Google’s setup help, this is not a new campaign type but an optimisation layer in existing Search campaigns with expanded search term matching, text customisation and URL expansion. In new Search campaigns, AI Max is enabled by default; existing Search campaigns are not automatically upgraded by it. It can be adjusted or switched off individually in the AI Max settings and retains negative keywords, brand, URL and location controls as well as its own reporting columns. Also documented: selected recommendations can be applied automatically at account level.
The observation that matters for this decision follows from this: part of the automation is already in the account, in some cases as a default. The operating effort does not disappear as a result; it shifts to checking defaults, reporting columns and automatically applied changes. That is exactly the work that a transfer of operating responsibility has to cover. How an automation register and an auto-apply check are built in practice is covered in the guide on AI automation and control.
Additional software or a service should therefore not repackage Google functions. The additional operating task must be named: cross-account monitoring, evidence, internal rules, a decision or an explicitly configured standing authority, documentation, escalation or managed responsibility. At maitiq, this task is named: recurring account analysis by workflows at a configured cadence, daily by default, proposals with the action, the rationale and supporting data, and a controlled, logged implementation process – managed by a Client Success Manager.
When is a software licence worthwhile?
A licence fits when an experienced team wants to run operations itself. Good prerequisites:
- a named internal person with operational responsibility and an available time budget;
- stable conversion measurement and clarified data responsibility;
- clear budget, approval and change rights;
- the ability to maintain API, tracking, default and tool changes;
- documented cover and a procedure for false alerts;
- an export and exit plan.
Software is less suitable when the purchase is meant to conceal a missing specialist or decision-making role. In that case the number of alerts and integrations rises, but not necessarily the number of good decisions.
When is a managed service worthwhile?
A managed service fits when business context and control stay in-house but recurring account management is not fully organised internally: with many accounts, campaigns or conversion paths, with competition between routine and strategic projects, or with no cover during absences arranged.
Check which tasks the provider actually performs. «Managed» should include at least a named contact, a working rhythm, an evidence format, an escalation path, cover during absences, approval limits and exit. Software access plus occasional support is not an operating process that has been taken over. At maitiq, the tasks actually delivered are named: a Client Success Manager as the contact, recurring workflows at a configured cadence, daily by default, as the rhythm, proposals with the action, the rationale and, where meaningful, the old and new value as the evidence format, and a quote per account instead of a percentage of the media budget. You set the approval limits yourself; cover during absences, the escalation path and exit terms belong in the agreement as separate service items and are not already evidenced by a named contact. Creative work, landing pages and further channels remain internal tasks or tasks agreed separately.
How do a licence, a managed service and a classic agency differ?
| Operating question | Software licence | Managed service | Classic agency |
|---|---|---|---|
| Goal configuration | internal team | provider with the client team | agency with the client team |
| Checking the Google defaults | internal, ongoing | to be agreed within the scope of services | to be checked in the contract |
| Alert review | internal | named service process | agency team, depending on the scope of services |
| Internal review effort | depends on the scope and the tools; measure it | depends on the scope taken over; measure it | depends on the model |
| Business context | fully internal | decisive internally | shared, depending on the model |
| Approval of consequential changes | internal | limited by contract | depending on the process |
| Cover during absences | organise internally | check the service process | check the team process |
| Documentation and audit trail | configure internally | part of the agreed operation | to be checked in the contract |
| Exit | data export and separation of integrations | separate data, rules and service | account and team handover |
The table does not assess quality. An internal team of experts can work very efficiently with software, a clear managed service can take responsibility off a tight team, and an agency can provide additional disciplines. The scope of services must be identical before the price comparison; where it differs, name the differences instead of treating the comparison as inadmissible. The same review effort on both sides does not prove that no service process was taken over: scope, difficult individual cases and review quality can differ. The columns compare arrangements, not three strictly separate provider types; agencies can also offer managed services.
This is how maitiq is positioned: goal configuration together with your team; alert review through recurring workflows at a configured cadence, daily by default, and a named Client Success Manager; reporting as a separate evaluation without proposals; approval of consequential changes with you, within the limits you set – manually or through an explicitly enabled standing authority; documentation as a decision log per proposal; implementation through the controlled, logged implementation process, without every acceptance requiring a fresh grant.
How do you calculate the total operating costs of both models?
Compare both models on the same scope of services and the same period – exclusively with your own quotes and fully loaded internal costs, without market assumptions. This worksheet is a manual comparison template, not an automatic calculator or a scorecard: the formulas below describe the total cost of ownership (TCO) per model for the same period, and you enter the figures yourself. For the complete annual calculation including the media budget, use Calculate total Google Ads costs: an annual planning worksheet.
For each model, record the amount, currency, unit, period, source and status (planned or measured):
| Line item | Unit | Type | Applies to |
|---|---|---|---|
| Licence or service fee | CHF per month | recurring | both |
| Setup and onboarding | CHF one-off | one-off when it falls in the comparison period | both |
| Integration and tracking maintenance | CHF per year | recurring | both |
| Internal operating time excluding alert review | hours per week | recurring | both |
| Internal context and decision time | hours per month | recurring; without overlap with approval time | both |
| Governance and approval time | hours per month | recurring; without overlap with decision time | both |
| Alert volume | alerts per week | basis for the review hours | both |
| Check time per alert | minutes per alert | basis for the review hours | both |
| Fully loaded internal cost rate | CHF per hour | basis | both |
| Training and induction | CHF one-off or per year | one-off when it occurs; an annually recurring training pro rata | both |
| Stand-in and absence cover | CHF per year | recurring | both |
| Additional services outside the scope of services | CHF per year | recurring | both |
| Exit and handover | CHF one-off | one-off when it falls in the comparison period | both |
You make the review effort visible separately because it drives the internal effort. Calculate the review hours from the alert volume; record the internal operating time without this alert review so that the same work does not enter the total twice. The same applies to internal context and decision time and to governance and approval time: record both without overlap; an hour already recorded as decision time does not also count as approval time:
Review hours in the period = alerts per week × minutes per alert × weeks in the period ÷ 60; only use 52 weeks for an explicitly full active year.
Internal costs in the period = (review hours in the period + operating time per week × weeks in the period + context and decision time per month × months in the period + governance and approval time per month × months in the period) × fully loaded internal cost rate
Software TCO in the period = (licence fee per month × months in the period) + setup and onboarding and exit and handover where they fall in the comparison period + (integration and tracking maintenance + stand-in and absence cover + additional services, each per year, pro rata for the period) + training and induction (one-off in the period when it occurs; an annually recurring training pro rata) + internal costs in the period
Service TCO in the period = (service fee per month × months in the period) + setup and onboarding and exit and handover where they fall in the comparison period + (integration and tracking maintenance + stand-in and absence cover + additional services, each per year, pro rata for the period) + training and induction (one-off in the period when it occurs; an annually recurring training pro rata) + internal costs in the period
Unknown is a status of its own and must not enter the total as zero; an empty field prevents a defensible total, because this worksheet calculates nothing by itself. Document items that are included in the agreed scope of services as «included» with no extra amount; «included» is not the same as «unknown» and does not prevent the total. One-off items such as setup and exit are only included when they occur in the chosen comparison period, and are shown separately from recurring items; planned estimates and measured values remain distinguishable. The Google media budget remains separate. Hoped-for performance improvements are not guaranteed savings. For the «licence or service fee» line, the following applies at maitiq: the fee is not linked to the media budget – maitiq does not earn more when your advertising spend rises.
Who is responsible for which stage from signal to change?
A controlled process has seven visible stages. What matters is not that they exist, but who is responsible for them in which model. The right-hand column shows the process at maitiq:
| Stage | Software licence | Managed service (maitiq) |
|---|---|---|
| 1 Signal: data movement or a rule triggers a review | tool | maitiq workflow, configured cadence, daily by default |
| 2 Validation: source, period, coverage, gaps | internal | maitiq, documented |
| 3 Context: goal, meaning of the conversion, budget, exceptions | internal | client team provides, maitiq records |
| 4 Proposal: action with evidence and limit | to be formulated internally | maitiq formulates: the action, the rationale, the old and the new value where meaningful |
| 5 Decision: accept, reject, clarify | internal | named person in the client team or an explicitly configured standing authority, within set limits |
| 6 Authorised implementation: only the agreed path changes | internal | controlled, logged implementation process; where the standing authority covers the proposal, no fresh grant is needed |
| 7 Post-check: attempt, result, next review date | internal | part of operations, with the Client Success Manager |
Compare providers precisely on that basis. If a signal leads directly to a change, the rule limit must be narrow and verifiable. If everything ends in one inbox, the internal operating effort has not been solved, only shifted.
Which questions belong in data protection, access and exit?
Clarify which account and performance data is processed, in which systems it resides, which subprocessors are involved and how long data is retained. Check separately: read-only analysis access, user and manager roles, technical write permission and human authorisation. Google distinguishes its own access levels for this, including a read-only level; however, the access level is a platform permission, not a contractual approval. At maitiq (Noomera GmbH, Switzerland), hosting and central processing are located in Switzerland; for the AI-powered analysis, account-derived data can reach a configured AI provider outside Switzerland – both statements belong in the same answer.
The exit needs an export of the relevant rules, decisions and documents, removal of access that is no longer required, handling of open changes and a clear deletion or retention rule. Do not adopt an architectural claim from a sales conversation; have the current agreement and the production configuration reviewed.
What should a read-only pilot look like?
A pilot first tests evidence quality and operational relief, not short-term advertising performance. Define in advance an account, a time window, permitted data sources, the workflows reviewed, internal reviewers, blocked changes, the escalation path and – for the review effort dimension – how the actual minutes are recorded. Then complete the blank evaluation table using your observations:
| Dimension | What is recorded | Permitted states |
|---|---|---|
| Coverage | what share of the agreed scope of services was reviewed | met / partly met / not met / cannot be assessed |
| Traceability | whether every alert shows source, period and limit | the same four states |
| Relevance | whether the alert matches the goal and the business context | the same four states |
| False alerts | number and type of alerts that were refuted | count with reasoning |
| Review effort | actual minutes per alert and week | measured time |
| Escalation | whether the agreed path worked in a real case | met / partly met / not met / not tested |
| Handover and exit | whether access removal, export and deletion work | met / partly met / not met / not tested |
The account assessment remains read-only and produces no provider score; you complete the evaluation table with observed results. CPC, CTR, conversion rate and ROI are not assumed pilot values. Write permissions are only reviewed after a separate decision. The read-only audit from maitiq evidences the account- and report-based part: agreed account access or separately supplied native reports, coverage per check as a visible state, no change in the account. It cannot measure the internal review effort from ads data: minutes per alert require supplied time records. Escalation, handover and exit also cannot be certified from the account; that requires the exercises and documents actually agreed.
The decision rule in four sentences
If the operating role is staffed internally and backed by time, buy a tool. If it is not staffed, buy an operation, not a licence. If the review effort remains equally high in both models, first check whether scope, difficult individual cases and review quality are comparable; the same review effort alone does not prove that no service process was taken over. And if both quotes do not point to the same scope of services, the same period and the same approval limit, the price comparison is not meaningful: name the differences before you set prices against each other. The operation you buy in the second case is what maitiq delivers: recurring workflows at a configured cadence, daily by default, a named Client Success Manager, proposals with the rationale, your decisions – manually or through an explicitly configured standing authority – and a controlled, logged implementation process, quoted per account.
Frequently asked questions
Does Google automation replace additional management software?
Not necessarily. Google automates important platform tasks and enables individual features in new Search campaigns by default. Additional software is only worthwhile if it solves a clear additional operating job that the internal team actually owns. maitiq does not replace Google automation; it complements it with recurring account analysis at a configured cadence, proposals with the rationale, decisions and a controlled, logged implementation process.
Is a managed service simply software with support?
No. A robust managed service takes over defined operating tasks with a named rhythm, responsibility, escalation, documentation and exit. Check the concrete contract. At maitiq, the rhythm, the Client Success Manager, the decision log and the exit are part of the managed service, quoted per account; cover during absences, the escalation path and exit terms belong in the agreement as separate service items and are not already evidenced by a named contact.
Is software cheaper than a service?
That cannot be answered without your own quotes and fully loaded internal costs. Licence, review effort, integration, cover during absences, governance and exit belong in the same calculation.
Who decides in AI-powered management?
The business goal, the meaning of the conversion, the budget and consequential changes need named people. Software can structure evidence and prepare work, but cannot take on corporate responsibility. At maitiq, named people in your company therefore decide; a limited autopilot is switched off by default and is deliberately activated per workflow and per account. When it is explicitly configured, that standing authority can accept eligible proposals within the limits you set and implement them through the controlled, logged implementation process; anything beyond those limits waits for a decision. Responsibility for objectives and authority stays with you.
Why should a pilot start read-only?
The pilot assesses coverage, evidence quality and internal effort without mixing up analysis and account changes; you complete the evaluation table with observed results. The account assessment remains read-only and the account itself unchanged; the audit from maitiq is built in this way.
Sources and how to interpret them
All sources below are Google documentation: Smart Bidding, Performance Max, AI Max for Search campaigns, applying recommendations automatically and access levels in Google Ads accounts. They describe platform features and limits, not a general market price or proof of success. Information on how maitiq works is available at maitiq.com.
To compare fees and services, we additionally include your existing offer or your invoice; this information is not held in the Google Ads account.