AI that works for your media budget

Your campaign is on track. Impressions are growing, CPM looks healthy and the report is showing green across the board. 

Should you increase the budget? And if results fall short, do you need to change the strategy, adjust the creative, or investigate a setup error?

In both cases, you need to understand what is behind the results. Are you reaching new audiences or the same people repeatedly? 

• Does the creative capture attention and communicate your message? 

• Does the campaign setup match the brief? 

• And what do the results tell you about the contribution to your brand and how to use the remaining budget?

We want to devote more attention to answering these questions. That is why we are bringing AI into our daily work at Proficio: connecting it with client materials and advertising platforms, building our own skills and knowledge base, and using vibe coding to create applications for specific team needs.

For clients, these technologies matter when they improve how we manage their investment.

AI needs more than the latest report

Campaign results make sense in the context of the brief. We need to understand what the campaign was intended to change, the role of each channel and any adjustments made along the way.

This information is often spread across emails, tasks, shared documents and advertising platforms. At Proficio, we use MCP connections provided by Databy to give AI access to available information from tools such as Asana, Slack, email, shared drives and selected advertising platforms.

When analysing performance, we can refer to the brief, previous recommendations and changes agreed with the client alongside the results. This helps us assess why campaigns perform differently and whether each is fulfilling its purpose.

For clients, this means recommendations grounded in their situation and objectives. A specialist verifies the source material, assesses AI’s conclusions and decides what they mean for the next steps in campaign management.

Knowledge base and skills: turning experience into a working process

Access to data is a starting point. To be useful, AI also needs to understand how we work with it.

We are building a knowledge base of our standards and experience. It includes guidance on reading media plans, interpreting and evaluating metrics, using consistent naming conventions and configuring campaigns across advertising platforms. It also covers measurement requirements, connections between systems and brand safety.

We continuously expand this knowledge base with lessons from audits and analysis: recurring errors, situations that need closer assessment and examples that help interpret results correctly. 

Capturing the reasoning behind our practices matters too: why we use a particular setup and when a different approach is needed.

The knowledge base supports skills: defined procedures for recurring tasks. They specify the inputs AI needs, what it should check, the steps to follow and the expected output. We do not have to explain our experience from scratch with every request.

We use AI for reach analysis and the evaluation of Brand Lift studies, ongoing performance evaluation and optimization recommendations. 

It also helps us prepare assets for ad formats that combine images and text, check advertising materials supplied by clients and build results presentations. We are progressively turning these activities into shared team skills.

AI audits: from checking the setup to taking action

AI audits help us systematically examine advertising accounts and campaign settings. We compare the data available from platforms against our standards and the client brief. 

Checks cover the account’s basic setup, connections to other platforms and measurement systems, campaign settings, and ad and creative assignments. They also flag inconsistencies in destination URLs, UTM parameters and naming conventions, or whether the correct creatives are active.

• Before launch, we check that everything is ready according to the media plan: budgets, dates, targeting, frequency caps, measurement and brand safety. We also review creative assignments and destination links. The aim is to identify inconsistencies before the campaign starts spending.

• For live campaigns, we check whether the current setup still matches the plan and approved changes. This can reveal outdated links, creatives that should no longer be active, or inconsistencies in the settings of related campaign elements or across platforms. The audit complements ongoing performance evaluation and optimisation.

AI helps us review campaign settings quickly and find inconsistencies that would otherwise require scanning hundreds of rows manually. Connections to advertising platforms have limits, however. The audit therefore clearly states which checkpoints it has verified and which require a specialist’s review. This makes checks faster and leaves more time to assess and resolve the issues found.

The audit summary shows which checks passed, where a setup mismatch was detected, which AI findings need specialist assessment and what must be checked manually.

Vibe coding: when a tool is missing, we build our own

Campaign management involves recurring tasks that could be simpler, but a suitable tool is not always available. We use vibe coding — building applications with AI assistance — to create our own. 

These applications help with tasks such as checking creative assets in bulk, preparing destination URLs and UTM parameters, and analysing frequency.

One example is our Reach Analyzer. It turns advertising platform reports into charts showing the distribution of ad exposure at different campaign levels in moments. This gives us clear inputs for assessing frequency, planning further optimisation or reallocating budget.

A healthy average frequency does not necessarily mean exposure is well distributed. It can conceal a large audience reached only once and a smaller group that has seen the ad many times. 

We therefore look beyond the average to understand the distribution and its implications for campaign management.

From overview to detail: the comparison reveals differences in ad exposure. The breakdown by insertion order helps us identify which parts of the media buy drive those differences and where to investigate opportunities for optimisation.

Broader analysis. More specific recommendations.

For larger analyses, we combine AI with scripts and calculations. We process exports, compare periods and investigate patterns across channels, devices and creative variants.

In reach analysis, we can examine: audience overlap, the audience reached exclusively by each channel, and how newly reached audiences develop over time, alongside total reach. 

For Brand Lift studies, we can compare results across campaign flights and available segments.

The next step matters: what should the client do based on the analysis? Reassess a channel’s role, adjust frequency, prepare a test, or maintain the current setup while adding measurement?

AI helps us compare more options and investigate more relationships in the same amount of time. We can then explain more clearly why we recommend a particular change and how we will assess whether it helped.

We bring together the size and cost of exclusive reach, overlap between channels and the development of newly reached audiences over time. Together, these provide inputs for assessing each channel’s role and future budget allocation.

Clients need to understand what their budget delivers and how to use it next. Connected data, a knowledge base, skills and custom applications help us investigate these questions more thoroughly.

The aim is to deliver more thorough campaign management in the same amount of time: more relationships examined, clearly explained results and recommendations that translate into action.

Do you know which questions your media evaluation should answer? 

The e-book What your media report isn’t telling you explores viewability, reach and overlap, frequency, cost transparency and measuring media impact.

Download the e-book