Read Ad Intelligence Channel Mix Without Overstating the Data

A channel-mix chart is easy to read too quickly. If Social takes up three quarters of a bar chart, it is tempting to say that a competitor puts three quarters of its advertising budget into social media. That conclusion needs much more evidence than a few collected campaign records.

In Shane’s Ad Intelligence, Channel mix shows the share of supplied spend within the records that match your current filters. That makes it useful for reviewing a research dataset. To explain it accurately, keep the included records, missing figures and reporting dates beside the percentages.

This workflow reflects the tool’s behavior checked on November 4, 2026. The workspace uses observations you enter or import and saves them in your browser. It does not connect to advertising libraries, fetch competitor creatives or independently verify spend.

1. Decide which records belong in the comparison

Before entering amounts, write one sentence describing the dataset. For example: “Published USD spend figures for three fictional campaigns in the United States, covering the same October reporting period.” The example is illustrative, not market research or a real advertiser’s budget.

For your own work, check each row against that scope:

• Currency: the spend input is labeled USD. There is no currency selector or conversion step. Do not combine dollar, pound and euro figures as though they were the same unit.
• Reporting window: spend and impressions should cover the same dates and campaign scope. First seen and Last seen describe observation dates; they do not prove the period covered by a financial figure.
• Source quality: record whether the amount is published, estimated or unavailable, plus the source and date checked.
• Overlap: avoid adding a campaign total alongside component creative totals that are already included in it. Reimporting a CSV also appends records rather than removing duplicates.

Use Notes for source context and a consistent research-batch label such as “October review.” Search includes notes as well as advertiser, campaign and publisher text, so a distinctive label can help retrieve a batch. It is a text match, not a dedicated batch field: check the returned rows for accidental matches.

2. Work out what the percentage actually divides

The chart groups matching records by channel, adds their supplied spend, and divides each channel subtotal by the total supplied spend in that view. It does not calculate a channel’s share of impressions, campaign count or the advertiser’s complete budget.

Consider these fictional observations:

• Social: $900 supplied spend and 60,000 impressions.
• Video: $300 supplied spend and 20,000 impressions.
• Display: spend unavailable and 10,000 impressions.

If Display spend is left blank when the observation is saved, the tool stores that numeric field as zero. With all three rows included, the cards show three campaign records, $1,200 in observed spend and 90.0K observed impressions. Channel mix shows Social at 75.0%, Video at 25.0% and Display at 0.0%.

The useful statement is: “Social accounts for 75% of the supplied spend in these records. Spend was available for two of the three observations.” The workspace does not calculate that availability count for you; maintain it in your review notes or source sheet.

A zero-width Display bar does not establish that the advertiser spent nothing on Display. It reflects the values stored in this dataset.

3. Preserve “unknown” outside the numeric total

Blank spend and impression entries become zero in the current form and CSV import. The numeric fields therefore cannot distinguish “not supplied” from a confirmed zero. Keep that distinction explicitly in Notes and retain the original source sheet.

A useful note might read: “Spend unavailable; impressions supplied by source report; October review; checked 2026-11-04.” For a confirmed zero, say that instead and retain the evidence. Do not invent a spend estimate merely to fill the field.

Before reporting channel share, decide whether to compare only records with usable spend or to show the broader dataset with a clear availability warning. There is no dedicated missing-spend filter. If you use consistent note labels to select a reviewed subset, inspect every included row before relying on the totals.

If all included spend values are zero, every channel percentage displays zero. That is an absence of a positive supplied-spend denominator, not an even allocation and not proof of no advertising activity.

4. Treat period filters as a record-selection rule

The period selector checks Last seen against the current time. A record without Last seen remains included when you choose Last 30 days, Last 90 days or Last 12 months. First seen does not determine that inclusion.

The filter also does not split campaign amounts across days. If an included row contains a $900 campaign total, the full $900 contributes to the view, even when the campaign started before the selected window. Calling that number “spend during the last 30 days” would require separate source data supporting that period.

Review dates before applying the filter. A future Last seen date currently passes the recency check, so a date-entry error can keep a record in view. Check undated and future-dated rows rather than assuming the selector has validated them.

For a monthly report, prepare records whose amounts genuinely cover that month. Then document how the tool filters were used to select them. Do not rely on the period selector to calculate a monthly allocation.

5. Recheck the denominator whenever you filter

Search, channel, market and period filters affect the summary cards and Channel mix chart. Selecting Social alone makes Social 100% of the supplied spend in that view when its total is positive. That is an expected consequence of narrowing the dataset, not new evidence of a social-only strategy.

Before sharing a chart, record the search text, channel, market and period settings together with the included record count. The advertiser comparison selector controls its own comparison chart; choosing brands there does not narrow the campaign table or Channel mix.

Keep the underlying amounts when precision matters. Spend is displayed in whole dollars and channel shares are displayed to one decimal place. Rounded percentages may not sum to exactly 100%, and a very small nonzero amount may display as $0.

6. Check the downloaded file before handing it over

Export CSV downloads all stored observations, including rows hidden by your current filters. It is not a saved copy of the chart’s selected dataset. Keep an untouched copy, then make a separate reporting sheet with the intended rows and clearly stated selection rules.

Inspect the exported columns as well as the row count. The exporter preserves commas, quotes and line breaks by quoting each field and escaping quotation marks. If an earlier download had shifted columns, create a fresh export from the workspace and compare it with the original source sheet. Reimporting still appends observations, so importing an export into the same populated workspace creates duplicate records.

For a small review, it can be safer to assemble a concise summary directly from the verified source records and use the workspace chart as a cross-check. Browser storage and a download are useful working aids, but neither replaces a reviewed source record.

A handoff another person can verify

Finish with a short caption and a source checklist:

• The question, market, currency and reporting period.
• The exact filters and number of included records.
• The supplied spend total and the number of observations with unavailable spend.
• Source references, review dates and any estimates.
• A check for duplicates, overlapping totals and date errors.
• A checked reporting sheet that matches the chart’s intended scope.

For the fictional example, a clear caption is: “Within this three-record research sample, Social represents 75% of the $1,200 in supplied spend. Display spend is unavailable. These records do not establish the advertiser’s total budget or campaign performance.”

That gives the reader a useful finding, its denominator and its limits in one place. For the broader research process, see the guide to organizing advertising observations.