From channel reports to managing against targets
Most paid programs are reported, not managed. Every channel has an owner, every owner has a report and every report shows the channel doing reasonably well by its own measure. Google Ads celebrates conversions, LinkedIn celebrates engagement, Meta celebrates cheap leads. Nobody is asked the only question that matters to the CEO: is this money producing the pipeline the revenue target needs, at a cost we planned for.
Managing against targets changes the conversation. Each channel gets a job, a budget and a band for its key numbers, all derived from the revenue goal. The weekly review then has a clear agenda: which channels are inside their band, which are outside, and what moves as a result. If you have not built the target and budget yet, start with marketing budget planning and come back here.
| Channel | CAC | ROAS | Payback |
|---|---|---|---|
| Branded search | $1,850 | 6.1 | 5 mo |
| LinkedIn Ads | $6,200 | 2.4 | 13 mo |
| Content and SEO | $2,900 | 4.0 | 8 mo |
| Meta prospecting | $9,400 | 1.1 | 27 mo |
Meta prospecting misses its payback band for the second week. Re-plan proposed.
How to set KPI bands for each channel
A KPI band is a range, not a single number. A point target invites panic at every small wobble; a band tells you when a change is real. Each paid channel needs a band for its primary metric and a minimum volume so that the metric means something.
For B2B, the primary metric is usually CAC or cost per sales-qualified opportunity, with payback as a check. For ecommerce it is ROAS, with new customer share as a check so the channel is not simply buying back existing buyers. Set the centre of the band from the plan and the width from your normal week to week noise, typically 20 to 30 percent either side.
- Primary metric. CAC, cost per opportunity or ROAS, depending on what the channel is for.
- Band. Upper and lower limits around the plan, for example a CAC of 5,000 to 7,000 USD.
- Minimum volume. Below a set number of conversions the metric is too noisy to act on, so the channel is judged on leading signals instead.
- Review window. One week for spend pacing, two consecutive weeks before any structural change.
When to cut, hold or scale a channel
Write the rules down before the numbers come in. The table below is the rule set VirtualCmo applies by default. You can tighten or loosen any line in your guardrails.
| Situation | Signal | Default action |
|---|---|---|
| Inside band, room to grow | CAC or ROAS inside band, impression share or audience not exhausted | Scale budget 10 to 20 percent and watch the next two weeks |
| Inside band, saturated | Metric holds but extra spend no longer adds volume | Hold budget, move growth money elsewhere |
| Outside band one week | Metric breaks the band once | Hold, check tracking and creative, no budget move |
| Outside band two weeks running | Metric breaks the band in two consecutive weeks | Cut 15 to 25 percent and re-plan the freed budget |
| Far outside band | CAC more than double the band or ROAS below 1 | Pause the campaign, keep retargeting audiences, flag for review |
| Too little data | Below the minimum volume for the metric | Judge on cost per lead and lead quality, keep a fixed test cap |
| New channel test | No history | Fixed budget and end date, success criteria set in advance |
See the pipeline your revenue target needs and three recommendations for your company.
Get my CMO planAuto re-plan when a KPI misses its band two weeks running
The two-week rule is the heart of the system. One bad week is noise: a holiday, a tracking glitch, an expensive auction day. Two in a row is a pattern, and patterns deserve a change to the plan rather than another week of watching.
Here is how that plays out with the default B2B SaaS assumptions. Meta prospecting carries 2,400 USD a month with a CAC band of 5,000 to 7,000 USD. In week one its CAC reads 8,900 USD, so VirtualCmo holds and checks tracking. In week two it reads 9,300 USD, still outside the band. The plan is revised: a recommendation to cut Meta prospecting 22 percent, about 528 USD a month, and move that money to branded search, which has been running at a CAC near 4,700 USD with room to grow.
The recommendation shows the reason, the expected impact on pipeline, the risk and the effort. Because a 22 percent shift is larger than the 10 percent auto-execute limit on the Growth plan, it waits for a person to press Approve. Smaller moves inside your limits can run on their own. Either way, the change, its reason and its result go into the decision log, and next week's CEO report lists it under decisions taken.
Guardrails that keep automation safe
Automation is only useful if you can trust it with money. VirtualCmo starts in Approve mode on every account and only acts alone inside rules you write. Connections to ad accounts are read-only by default.
- Maximum shift per channel. No single move can take more than a set share of a channel's budget, 10 percent on Growth and your choice on Scale.
- Spend caps. Total monthly spend and per-channel spend never exceed the limits you set, whatever the model recommends.
- Approval thresholds. Moves above a dollar amount or a percentage go to an Approver, even when auto-execute is on.
- Roles. On Scale, Owner, Admin, Approver and Viewer roles separate who can change rules from who can approve moves.
- Decision log and audit log. Every recommendation, approval, auto-execution and outcome is recorded with a timestamp and the person or rule behind it.
Running the weekly performance review
- 1 Open the dashboard and read pipeline against target first. If the company is on track, the review is about efficiency; if it is at risk, it is about volume.
- 2 Go channel by channel and mark each one inside band, outside band one week, or outside band two weeks.
- 3 Apply the rule table. Scale what is inside band with room, cut what has been outside two weeks, pause anything far outside.
- 4 Check the guardrails and approve, dismiss or ask why on each recommendation that needs a person.
- 5 Brief the specialists. Budget changes usually need new creative, landing pages or keywords to go with them.
- 6 Record what you expect to happen, so next week's review compares the outcome with the forecast instead of with memory.
What this approach does not replace
Rules handle allocation well. They do not write the ad that makes a buyer stop scrolling, they do not decide that your positioning is wrong, and they cannot judge a brand campaign whose payoff arrives in a year. Keep humans on creative direction, messaging and brand bets, and use the rule set to stop money leaking while they work. The numbers behind the bands come from a marketing analytics dashboard that treats every channel the same way, and the guide on when to cut a paid channel goes deeper on edge cases.
Frequently asked questions
Why wait two weeks before changing a channel?
Weekly numbers are noisy. A single bad week often comes from tracking problems, holidays or auction spikes. Two consecutive weeks outside the band is a much stronger signal and still fast enough to save real money.
Can VirtualCmo pause our campaigns without asking?
Only if your guardrails allow it. Approve mode is the default, and any move above your maximum shift or approval threshold waits for a person.
What if a channel has too few conversions to judge?
It is judged on leading signals such as cost per lead and lead quality, with a fixed test cap, until it reaches the minimum volume for its primary metric.
Does this work for brand campaigns?
Partly. Brand spend can have its own band on reach and branded search lift, but the payoff is slow. Most teams ring-fence a brand budget and manage it outside the cut or scale rules.
How often does the plan change on its own?
Growth re-plans monthly and Scale weekly, with daily pacing. Any KPI that misses its band two weeks running triggers a re-plan sooner. See pricing for plan details.