Audit Census AI Adoption Data

Build a Census AI adoption chart that marks a survey-question break.

Introduction

30 Second Summary

A viral chart can turn a pile of numbers into a story that feels obvious. If the question behind those numbers changes halfway through, the line can suggest growth that the data cannot prove.

In this project, you will rebuild a widely shared AI adoption chart from the U.S. Census Bureau's Business Trends and Outlook Survey data. You will calculate six-survey trailing averages in Google Sheets before plotting each survey wording as a separate series with a visible gap.

What You'll Build

When you open your finished PNG, you can follow AI use by firm size while seeing exactly where the survey's two wordings stop being comparable.

By the end of this project, you'll have:

Clean firm-size data that lets you compare seven business groups without unrelated survey rows getting in the way.
Two separate six-survey averages that smooth each survey wording without crossing the break.
A labelled line chart with the wording change held inside a shaded gap. You can export the finished chart as a shareable PNG.
Secret Mission: An optional challenge to push your chart-checking skills further.

Are there any prerequisites?

You only need a browser and a Google account for Google Sheets. Any free spreadsheet program that opens XLSX files also works.

Before We Start

This first checkpoint gives you a clear reason for rebuilding the chart before any spreadsheet work begins. You will carry that purpose into every choice about the survey wording.

Download the Census Survey Data

A viral chart can flatten a survey into one smooth story. The U.S. Census Bureau's Business Trends and Outlook Survey is the source you need to check that story.

In this step, you'll download the Employment Size Class workbook. You'll convert it into a Google Sheets spreadsheet for the analysis ahead.

In this step, get ready to:

Download the Employment Size Class workbook from the Census Bureau.
Convert the workbook into a Google Sheets spreadsheet.
Confirm that the survey estimates imported correctly.

Download the employment-size workbook

An XLSX workbook stores spreadsheet data in an Excel-compatible file. The Employment Size Class workbook contains the estimates needed to compare firms of different sizes.

The downloads page contains several data products, so the right workbook can take a moment to spot.

Visit the official BTOS Data Downloads page in your browser.
Find the Employment Size Class section on the page.
Click the XLSX download link in that section.
Wait for the workbook download to finish.

You'll see a spreadsheet file with an .xlsx extension in your browser's downloads list.

Download missing?

Check your browser's downloads panel for the workbook.
Allow the Census Bureau download if your browser blocked it.
Click the XLSX link again.

Troubleshoot the missing file with help for your browser.

Convert the workbook to Google Sheets

Google Drive stores the downloaded workbook online. Converting it creates a native Sheets copy for the filtering and charting steps.

Go to Google Drive in your browser.

You'll see the files stored in your Drive account.

Click New in the top-left corner.
Select File upload.

Your computer's file picker opens.

Choose the downloaded .xlsx workbook from your downloads folder.
Wait until Google Drive confirms that the upload is complete.

You'll see the uploaded workbook listed in Google Drive.

Double-click the uploaded workbook in Google Drive.
Click Open with Google Sheets at the top of the preview.

You'll see the imported Census worksheet in the Sheets editor.

Click File in the top menu.
Select Save as Google Sheets.

You'll see a new browser tab containing a native Google Sheets copy.

Click the spreadsheet title in the top-left corner.

The current title becomes editable.

Replace the title with AI Adoption by Firm Size.
Press Enter to save the new title.

You'll see AI Adoption by Firm Size at the top of the spreadsheet.

Spreadsheet still shows XLSX?

Click File in the spreadsheet's top menu.
Select Save as Google Sheets.
Use the new Sheets copy that opens in another browser tab.

Get help with converting the workbook.

Confirm the imported survey data

A quick import check confirms that you downloaded the correct source workbook. It also confirms that the original survey fields survived the conversion.

Before you check, which fields would prove that this workbook can support a chart of AI use by firm size?

Scan the header row in AI Adoption by Firm Size.
Locate the field or fields containing survey dates.
Locate the field containing question text or a question identifier.
Locate the field containing employment-size classes.
Locate the field containing survey estimates.

You'll see rows of Census survey estimates organised by employment-size class. The source fields remain available for the next step.

That source check is complete. Next, you'll isolate the AI use question from the rest of the Census survey data.

Filter the AI Use Estimates

Your imported Census worksheet contains estimates for many survey questions. A useful adoption chart needs the rows that measure current AI use across the seven employment-size groups.

You already have the source data in Google Sheets. In this step, you'll isolate the relevant estimates while preserving their dates and question wording.

In this step, get ready to:

Create a separate worksheet for the filtered observations.
Keep the estimates that represent reported AI use.
Retain the seven employment-size groups with the fields needed for charting.

Create a clean worksheet

The imported worksheet is your source record. A separate worksheet gives you a clean destination without removing any Census data.

Switch back to the imported worksheet from earlier.
Click the plus icon beside the worksheet tabs to create a blank worksheet.

You should see a new blank worksheet tab at the bottom of the spreadsheet.

Double-click the new worksheet tab.
Type AI use filtered into the worksheet name field.
Press Enter.

The new tab should now read AI use filtered.

Switch back to the imported worksheet.

Filter for reported AI use

The workbook can contain questions about reported use and future plans. This chart needs the affirmative estimate from each question that measures use during the survey period.

Click any cell inside the main estimate table.
Click Data in the top menu.
Click Create a filter.

You'll see a filter control in each header cell across the table.

Click the Filter control in the column that contains the survey question wording.
Expand Filter by values.
Clear the current selections.
Type Intelligence into the search box.
Select every wording that asks about actual use during the survey period.
Click OK.

The visible rows should now describe recent AI use. Rows about future plans should be hidden.

Why keep every current-use wording?

The current-use question appears with different wording across the survey dates. Keeping each wording preserves the evidence you need to identify comparability breaks later.

The question text belongs in the dataset beside every estimate. That wording tells readers what each number measures.

Which estimate belongs in the chart?

A current-use question can produce estimates for several response categories. The adoption chart uses the estimate for businesses reporting that they used AI during the measured period.

Click the Filter control in the response or answer column.
Clear the current selections under Filter by values.
Select the response value that indicates the business reported using AI.
Click OK.

Each visible row should now represent the reported-use estimate. Other response categories should be hidden.

Click the Filter control in the employment-size column.
Clear the current selections under Filter by values.
Select only 1–4, 5–9, 10–19, 20–49, 50–99, 100–249, and 250+.
Click OK.

The visible data now covers reported AI use for the seven target employment-size groups.

Cannot find the right question or response?

Check that you opened the filter for the column containing question wording instead of the estimate column.
Use the broad search term Intelligence so the results include each version of the current-use question.
Read the response values carefully so you keep reported use instead of future plans or another answer category.

Ask for help identifying the correct rows in your workbook.

Build the four-column table

A filter hides unrelated observations in the source worksheet. Copying the visible result into your new worksheet creates the focused table used by the rest of the project.

Click any visible cell inside the filtered table.
Press Cmd+A on macOS or Ctrl+A on Windows once to select the table range.
Press Cmd+C on macOS or Ctrl+C on Windows to copy the visible data.
Switch to the AI use filtered worksheet.
Click cell A1.
Press Cmd+V on macOS or Ctrl+V on Windows to paste the filtered data.

You should see the filtered headers and observations fill the new worksheet.

Return to the imported worksheet.
Click Data in the top menu.
Click Remove filter to restore the complete source table.
Return to the AI use filtered worksheet.

Your imported worksheet is back to its complete view. The copied observations remain in AI use filtered.

Select every column unrelated to the survey date, question wording, employment-size group, or estimate.
Use the column-header menu to delete the selected columns.
Rename the four remaining headers to Survey date, Question wording, Employment-size group, and Estimate.

You should now see one compact table with four columns. Every row should have a date, wording, size group, and estimate.

Before you check one survey date, how many employment-size rows do you expect to see?

Click any cell inside the compact table.
Click Data in the top menu.
Click Create a filter.
Click the Filter control in the Survey date column.
Clear the current selections under Filter by values.
Select one survey date.
Click OK.

You should see exactly seven data rows for that date. The employment-size values should run from 1–4 through 250+ with one row for each group.

Do you see fewer or more than seven rows?

Check the employment-size values for a missing target group.
Check the response column in the imported worksheet if several answer categories were copied.
Repeat the source filters if unrelated question rows appear in the compact table.

Ask for help checking why the row count is different.

Reopen the Filter control in the Survey date column.
Show all survey date values again.
Click OK.

Great work. Your full filtered table is back on screen with every survey date and all seven employment-size groups.

Your AI-use observations are ready. Next, you'll separate the rows where the survey wording changes so the chart does not connect incomparable estimates.

Separate the Survey Wordings

Your Google Sheets spreadsheet now gathers every firm-size estimate for the AI-use question in one place. The next job is to protect your chart from a hidden break in that question.

On November 17, 2025, the Census wording changed from AI used to produce goods or services to AI used for any business function. The later observations measure a broader activity.

A single uninterrupted line would turn that wording change into a false trend. You will separate the observations into two tables before any smoothing or charting begins.

In this step, get ready to:

Locate the November 17, 2025 question-wording break.
Create one table for each survey wording.
Record the wording break as a chart gap.

Identify the wording break

Chronological sorting places the last observation under the old wording directly before the first observation under the new wording. That boundary becomes your audit trail.

Return to the AI use filtered worksheet in the spreadsheet from earlier.
Select the entire populated table.
Click Data in the menu bar.
Choose Sort range.
Choose Advanced range sorting options.
Check Data has header row.
Set the primary sort to the survey date column in A to Z order.
Click Sort.

You will see the survey dates run from earliest to latest. Each date stays attached to its complete observation row.

Find the first row dated November 17, 2025.
Compare its question wording with the wording in the previous row.
Select the header cell at the top of the survey date column.
Press Shift+F2 to add a note.
Type Question-wording break: November 17, 2025. Keep a chart gap between the two series. into the note.
Click another cell to save the note.
Hover over the survey date header to confirm the note appears.

The row before the boundary should use the wording about AI used to produce goods or services. The first November 17, 2025 row should use the wording about AI used for any business function.

Why the gap matters

The newer question expands what counts as AI use. The chart gap tells readers that the later points answer a different question.

That note now carries the rule that prevents two different measures from becoming one misleading trend.

Create the producing-goods-or-services table

The first table represents the original measure. Every row in it must come from before November 17, 2025.

Return to the AI use filtered worksheet.
Click the down arrow on its sheet tab.
Choose Duplicate.
Double-click the copied sheet tab text.
Type Producing goods or services.
Press Enter.

Good. The original filtered worksheet remains untouched while the new tab becomes the home for the older wording.

Find the first row dated November 17, 2025.
Click that row number.
Scroll to the last populated row.
Shift-click the last populated row number to select the full block.
Right-click the selected row numbers.
Choose Delete rows.
Inspect the final remaining date block.

The latest remaining survey date will be before November 17, 2025. You will see one row for each of the seven employment-size groups under that date.

Rows missing from the older table?

Click Undo if the header row disappeared.
Restart the selection at the first November 17, 2025 row if later observations remain.
Confirm that you edited Producing goods or services instead of the original filtered worksheet.

Help me check which rows belong in the producing-goods-or-services table.

Create the any-business-function table

The second table represents the broader measure. Its first observations belong to November 17, 2025.

Return to the original AI use filtered worksheet.
Click the down arrow on its sheet tab.
Choose Duplicate.
Double-click the copied sheet tab text.
Type Any business function.
Press Enter.

Your spreadsheet now has a separate destination for the broader survey question.

Locate the last row dated before November 17, 2025.
Click the first data row number below the header.
Shift-click the last pre-break row number to select the full block.
Right-click the selected row numbers.
Choose Delete rows.
Inspect the first remaining date block.

The first remaining survey date will be November 17, 2025. You will see all seven employment-size groups under that date.

Before you compare the finished tables, which worksheet do you expect to contain the cutoff date?

Switch to Producing goods or services.
Read the latest survey date.
Inspect the question wording in that final date block.
Count the employment-size groups in that final date block.

Keep that boundary in mind while you check the broader table.

Switch to Any business function.
Read the earliest survey date.
Inspect the question wording in that first date block.
Count the employment-size groups in that first date block.
Hover over the survey date header to reveal the wording-break note.

The older table ends before November 17, 2025 with the producing-goods-or-services wording. The broader table begins on November 17, 2025 with the any-business-function wording.

The header note records the boundary as a chart gap. Strong work. Your spreadsheet now keeps the wording change visible instead of disguising it as growth.

Does the cutoff date appear in both tables?

Remove the November 17, 2025 rows from Producing goods or services.
Keep the November 17, 2025 rows in Any business function.
Confirm that each boundary date block contains all seven employment-size groups.

Help me diagnose why the cutoff date appears in the wrong survey table.

Your two survey wordings now have separate tables with a visible boundary between them. Next, you will calculate a trailing average that restarts at this break.

Calculate Separate Trailing Averages

Your two wording tables now protect the November 17, 2025 break. Raw estimates can still rise or fall sharply from one survey to the next.

A six-survey trailing average smooths each point with the current survey plus as many as five earlier surveys. You will calculate that average separately in Google Sheets.

Every averaging window must stay on one side of the wording break. Mixing the two questions would recreate the misleading continuous trend you have worked to prevent.

In this step, get ready to:

Add a six-survey trailing-average column to each wording table.
Calculate smoothed values for all seven employment-size groups under the original wording.
Restart the calculation inside the any-business-function table.

Set up the average columns

Each wording table needs its own destination for the smoothed values. Matching headers make the finished tables easier to audit.

Column map for the formulas

Column A holds the survey date.
Column B holds the question wording.
Column C holds the employment-size group.
Column D holds the raw estimate.

Switch back to the AI use filtered worksheet from earlier.
Select the first empty header cell beside column D in the producing-goods-or-services table.
Type Six-survey trailing average into the selected header cell.
Select the first empty header cell beside column D in the any-business-function table.
Type Six-survey trailing average into the selected header cell.

You should now see one empty average column beside each table's raw estimates.

Calculate the original-wording averages

A trailing window follows one employment-size group through time. The formula chooses the six most recent matching observations up to the current survey date.

Find the first data row in the producing-goods-or-services table.
Record that row number here: first old-wording data row.
Select the first empty cell beneath the new trailing-average header.
Calculate the first old-wording average by pasting this formula:

=AVERAGE(SORTN(FILTER($D$[[OLD_FIRST_ROW="first old-wording data row"]]:$D,$B$[[OLD_FIRST_ROW="first old-wording data row"]]:$B=B[[OLD_FIRST_ROW="first old-wording data row"]],$C$[[OLD_FIRST_ROW="first old-wording data row"]]:$C=C[[OLD_FIRST_ROW="first old-wording data row"]],$A$[[OLD_FIRST_ROW="first old-wording data row"]]:$A<=A[[OLD_FIRST_ROW="first old-wording data row"]]),6,0,FILTER($A$[[OLD_FIRST_ROW="first old-wording data row"]]:$A,$B$[[OLD_FIRST_ROW="first old-wording data row"]]:$B=B[[OLD_FIRST_ROW="first old-wording data row"]],$C$[[OLD_FIRST_ROW="first old-wording data row"]]:$C=C[[OLD_FIRST_ROW="first old-wording data row"]],$A$[[OLD_FIRST_ROW="first old-wording data row"]]:$A<=A[[OLD_FIRST_ROW="first old-wording data row"]]),FALSE))

How does the trailing window work?

The first FILTER keeps rows with the current question wording.
A second condition keeps the current employment-size group.
The date condition keeps observations through the current survey date.
SORTN orders the matching surveys from newest to oldest.
The 6 limit keeps at most six estimates.
AVERAGE returns the mean of the available estimates.

Commit the formula by pressing Enter.
Select the formula cell through the final data row in the producing-goods-or-services table.
Fill the selected range downward by pressing Cmd+D (macOS) or Ctrl+D (Windows).
Pause while Google Sheets saves the filled formulas automatically.
Find the earliest survey for 1–4 employee firms in this table.
Compare its raw estimate with its trailing average.

The two values should match because only one old-wording observation is available at that point.

Good progress: the original wording now has a smoothed value for every firm-size observation.

Seeing blank cells or errors?

Place the first formula in the first data row beneath the trailing-average header.
Check the column letters against the four-column map above.
Confirm your recorded row number matches the first old-wording data row.

Ask for help with the formula references:

Restart the average after the break

The first broader-wording observation needs a fresh averaging window. This fresh start protects the new series from earlier estimates produced by a different question.

Find the first data row in the any-business-function table.
Record that row number here: first new-wording data row.
Select the first empty cell beneath this table's trailing-average header.
Calculate the first broader-wording average by pasting this formula:

=AVERAGE(SORTN(FILTER($D$[[NEW_FIRST_ROW="first new-wording data row"]]:$D,$B$[[NEW_FIRST_ROW="first new-wording data row"]]:$B=B[[NEW_FIRST_ROW="first new-wording data row"]],$C$[[NEW_FIRST_ROW="first new-wording data row"]]:$C=C[[NEW_FIRST_ROW="first new-wording data row"]],$A$[[NEW_FIRST_ROW="first new-wording data row"]]:$A<=A[[NEW_FIRST_ROW="first new-wording data row"]]),6,0,FILTER($A$[[NEW_FIRST_ROW="first new-wording data row"]]:$A,$B$[[NEW_FIRST_ROW="first new-wording data row"]]:$B=B[[NEW_FIRST_ROW="first new-wording data row"]],$C$[[NEW_FIRST_ROW="first new-wording data row"]]:$C=C[[NEW_FIRST_ROW="first new-wording data row"]],$A$[[NEW_FIRST_ROW="first new-wording data row"]]:$A<=A[[NEW_FIRST_ROW="first new-wording data row"]]),FALSE))

Why does this formula restart?

The fixed ranges begin at the first row of the any-business-function table. Earlier producing-goods-or-services observations sit outside that averaging range.

The current-row references move as you fill the formula downward. Each result still follows one wording plus one employment-size group.

Commit the formula by pressing Enter.
Select the formula cell through the final data row in the any-business-function table.
Fill the selected range downward by pressing Cmd+D (macOS) or Ctrl+D (Windows).
Pause while Google Sheets saves the second set of formulas automatically.

Before you inspect the break, ask yourself whether the first broader-wording average will match its raw estimate.

Find the earliest broader-wording observation for 250+ employee firms.
Compare its raw estimate with its trailing average.

The two values should match. This first average uses only the first observation collected under the broader wording.

Repeat the comparison for the remaining six employment-size groups.

Each group's first broader-wording average should match its raw estimate.

Scan both trailing-average columns for blank cells or errors.

You should see numeric values for every observation across all seven groups. That is the critical check complete: every smoothed series now restarts at the survey break.

Does the first new average look too low?

Confirm the fixed ranges begin at the first row of the any-business-function table.
Check that the first broader-wording formula references its own row.
Verify the wording condition points to column B.

Ask for help checking the restart:

Your smoothed series now stop cleanly at the November 17, 2025 boundary. Next, you will plot both wordings with a visible gap between them.

Chart the Break in the Series

Your two wording tables now have six-survey trailing averages that restart at the survey break. A standard line chart could still bridge those periods into a false trend.

In this step, you will use Google Sheets to keep each wording in its own series set. The visible gap will protect the AI adoption story from a misleading connection.

In this step, get ready to:

Plot the original producing-goods-or-services series.
Add the broader any-business-function series as a separate set.
Make the November 17, 2025 wording break visible.

Plot the original wording

The producing-goods-or-services table supplies the first half of the chart. Its survey dates provide the horizontal axis. Its seven trailing-average columns provide the firm-size lines.

Switch back to the AI use filtered worksheet.
Locate the producing-goods-or-services table from the previous step.
Select its survey-date column plus all seven six-survey trailing-average columns.
Click Insert in the top menu.
Click Chart.

You will see a chart appear over the worksheet. Its first version may use a different chart type.

Double-click the chart to open the chart editor.
Click Setup in the chart editor.
Open Chart type.
Select Line chart.
Confirm the chart uses the selected header row for the series names.

You should now see seven labelled lines that stop before November 17, 2025. Each line represents one employment-size group under the original wording.

Why Start With One Wording?

Starting with the original wording gives you a clean first series set. Every line shares the same question definition during this period.

The six-survey average remains inside that definition. This keeps the smoothing faithful to the source question.

Missing Lines From the First Table?

Check that your selected range includes all seven trailing-average columns.
Remove any raw-estimate columns that appeared as extra lines.
Confirm the survey-date column is assigned to the horizontal axis.

Help me fix the missing or extra lines in my Google Sheets chart.

Add the broader wording as a second series set

A series set lets the same chart use another date range without joining the values to the first set. The second set begins with the any-business-function wording on November 17, 2025.

Double-click the chart to return to the chart editor.
Click Setup.
Click Add axis & series set at the bottom of the panel.
Choose the survey-date column from the any-business-function table as the new horizontal axis.
Choose all seven six-survey trailing-average columns from that table as the new series.

You will see a second family of lines begin on November 17, 2025. The original lines should still stop before that date.

Click Customize in the chart editor.
Click Series.
Use Apply to to select the first pre-break series.
Choose a dashed option under Line dash styles.

The selected pre-break line now appears dashed. That visible change identifies it as part of the earlier question period.

Apply the same dashed pattern to the other six pre-break series.
Keep all seven post-break series solid.

Your chart should now show dashed lines ending before the break. Solid lines should begin with the broader wording.

Why Use Separate Line Styles?

Dashed lines identify estimates from the producing-goods-or-services question. Solid lines identify estimates from the any-business-function question.

The style change gives every employment-size group the same visual warning. Readers can compare groups within each period without treating the break as ordinary movement.

Did the Lines Connect Across the Break?

Confirm you used Add axis & series set for the second table.
Remove any series that combines values from both wording tables.
Check that the pre-break trailing averages end before November 17, 2025.

Help me separate the two wording periods without connecting their lines.

Make the wording break visible

An annotation names the methodological break at the exact date it occurs. This marker explains why the dashed lines stop before the solid lines begin.

Return to the any-business-function table.
Right-click its survey-date column header.
Choose the option that inserts one column to the right.
Enter Question wording break in the new column header.
Locate the row dated November 17, 2025.
Enter Question changed to any business function in the new column on that row.

The table now carries a note on the first date covered by the broader question. You can attach that note to the chart's horizontal axis.

Double-click the chart to reopen the chart editor.
Click Setup.
Click More beside the post-break horizontal axis.
Click Add labels.
Select the new question-wording-break column.
Click OK.

You should now see a note attached to November 17, 2025. The note marks the point where direct comparison stops.

Click Customize in the chart editor.
Click Chart & axis titles.
Choose Chart title under Type.
Enter AI use by firm size across two Census question wordings under Title text.

The chart title now tells readers that the figure contains two survey questions. The subtitle will state how the values were smoothed.

Choose Chart subtitle under Type.
Enter Six-survey trailing averages. Wording break: November 17, 2025. under Title text.
Resize the chart until the annotation plus all seven employment-size labels are readable.

Why the Marker Matters

The marker identifies a change in what the survey measured. It prevents the apparent increase from reading like growth under one stable question.

The gap becomes part of the evidence. Readers can see where the comparison loses continuity.

Before you inspect the finished chart, predict where the apparent rise for 250+ employee firms should sit.

Inspect the 250+ pre-break series plus its post-break counterpart.

What Should You See?

The dashed pre-break 250+ series sits near 12% in September 2025.
The annotation marks the November 17, 2025 wording break.
The solid post-break 250+ series sits near 30% by February 2026.
The same post-break series approaches 40% by October 2026.
No line connects the pre-break value to the post-break value.

Is the Break Hard to Read?

Widen the chart until November 17, 2025 has enough horizontal space.
Confirm the question-change note is attached to the post-break horizontal axis.
Check that pre-break lines use a dashed style.

Help me make the wording break clearer in my Google Sheets line chart.

That is the integrity check complete. Your chart now shows exactly where comparability stops.

Your two survey periods now share one chart without sharing one trend line. Next, you will add the final wording note and export the labelled chart.

Annotate and Export the Chart

Your Google Sheets chart already keeps the November 17, 2025 question-wording break visible. The 250+ employee series no longer turns that break into a continuous trend.

An exported chart needs to carry its context wherever it goes. In this step, you will add the later wording note before saving the finished chart as a PNG image.

In this step, get ready to:

Add the October 19, 2026 Super Intelligence wording note.
Verify the final chart annotations.
Save the PNG beside your spreadsheet in Google Drive.

Add the Super Intelligence wording note

The marked November gap explains the first wording break. A second note records the October 19, 2026 change to Super Intelligence wording.

Switch back to the AI Adoption by Firm Size spreadsheet from earlier.
Double-click the line chart to open its editor.
Click Customize in the chart editor.
Expand Chart & axis title.
Select the subtitle entry from the Type menu.
Enter October 19, 2026 change to Super Intelligence wording in the Title text field.
Click an empty cell to finish editing the chart.

You should see the Super Intelligence wording note beneath the main chart title.

Why does this note matter?

The November 17, 2025 gap prevents the first two survey questions from appearing as one continuous measure.

The October 19, 2026 note flags another wording change. Readers can now see where comparisons need extra care.

That is the interpretation safeguard in place: your exported chart now carries the warning with it.

Check the chart's final story

Before you inspect the final chart, where do you expect the apparent jump for 250+ employee firms to appear?

Close the chart editor side panel.
Inspect the marked gap at November 17, 2025.
Follow the 250+ employee line to the edge of the gap.
Confirm the producing-goods-or-services series stops before the gap.
Confirm the any-business-function series starts after the gap.
Confirm the October 19, 2026 note remains visible.

You should see two disconnected wording periods. The apparent 250+ employee jump should sit inside the marked gap.

The Super Intelligence wording note should remain visible beneath the chart title.

Missing a label or gap?

Reopen the chart editor to check the subtitle under Chart & axis title.
Compare the subtitle with the required October 19, 2026 wording.
Check that the two wording series remain disconnected at November 17, 2025.

Ask for help with the chart settings if the annotation still does not appear: Help me check why my Google Sheets chart subtitle or wording gap is missing.

Export and save the PNG

A PNG is a static snapshot. Later spreadsheet changes do not update the exported image.

Click the chart once to select it.
Click More at the top right of the chart.
Click Download as.
Choose PNG as the file type.

Your chart is now a portable image. Rename the downloaded file so it is easy to identify.

macOS

Press Cmd+Space to open the search bar.
Type Finder.
Press Enter to open Finder.
Select Downloads in the Finder sidebar.
Select the downloaded chart image.
Press Return to edit its name.
Enter ai-adoption-by-firm-size.png.
Press Return to save the new name.

Windows

Press the Windows key to open search.
Type File Explorer.
Press Enter to open File Explorer.
Select Downloads in the left sidebar.
Select the downloaded chart image.
Press F2 to edit its name.
Enter ai-adoption-by-firm-size.png.
Press Enter to save the new name.

The image now has the required filename. The last task places it beside the spreadsheet in Drive.

Return to the Google Drive folder containing AI Adoption by Firm Size.
Click New at the top left.
Click File upload.
Select ai-adoption-by-firm-size.png from your Downloads folder.
Wait until Upload complete appears.

Before the final check, which two files do you expect to see together in the Drive folder?

Look at the open Drive folder to confirm the uploaded image is listed beside the spreadsheet.

You should see AI Adoption by Firm Size beside ai-adoption-by-firm-size.png in the same Drive folder.

The PNG should show separate wording series across the marked November 17, 2025 gap. The October 19, 2026 note should appear beneath the main chart title.

Can't find the exported PNG?

Check your computer's Downloads folder for the exported chart.
Confirm the filename ends with .png.
Repeat the Drive upload from the folder containing the spreadsheet.

Get help locating or uploading the file: Help me find my exported Google Sheets chart and upload it to the correct Google Drive folder.

You have finished the hard part: the chart now carries its comparability warnings wherever it is shared.

Secret mission

Audit the Chart's Integrity

Put your finished chart through a three-point integrity audit. Trace the 250+ employee series to its source values. Prove that the averaging boundary protects readers from a false trend.

Clean Up Your Resources

Your public Census data and the files in your existing Google account create no ongoing project charges. Choose whether to keep your work ready, pause it for later, or delete it.

Resources you used:

The AI Adoption by Firm Size Google Sheets spreadsheet in Google Drive.
The ai-adoption-by-firm-size.png chart image in Google Drive.
The downloaded Employment Size Class XLSX source file on your computer.

Keep everything running

No action is needed. Choose this option if you want to add future Census releases or revisit the question-wording breaks.

Keep AI Adoption by Firm Size in Google Drive as your editable analysis.
Keep ai-adoption-by-firm-size.png beside the spreadsheet as your finished export.
Keep the downloaded Employment Size Class XLSX as an untouched copy of the source data.

Pause - I'll come back to this later

There is no running process to stop. Your spreadsheet and image can remain idle without creating ongoing project charges.

Leave AI Adoption by Firm Size in Google Drive.
Leave ai-adoption-by-firm-size.png beside the spreadsheet.
Close the Google Sheets browser tab to pause your work.
Return to the spreadsheet when you are ready to add another Census release.

Delete - I don't want to use this again

Deleting these files removes your working analysis plus its exported image. Save another copy first if you may want to revisit the chart.

Find AI Adoption by Firm Size in Google Drive.
Use the spreadsheet's context menu to move it to the trash.
Find ai-adoption-by-firm-size.png in the same Google Drive location.
Use the image's context menu to move it to the trash.
Open Google Drive's trash folder from the left sidebar.
Permanently delete the spreadsheet and PNG.

That is the cloud cleanup done. The downloaded Census workbook is the only local project file left.

Search for the downloaded Employment Size Class XLSX in Finder on macOS or File Explorer on Windows.
Move the downloaded XLSX to your computer's trash or recycle bin.
Empty your computer's trash or recycle bin to remove the source file permanently.

Nice Work!

You did it! Your Google Sheets analysis now exposes the question-wording break that a continuous trend line would hide.

You've learned how to:

Filter Census survey estimates into seven employment-size groups while preserving each survey date.
Separate non-comparable question wordings at November 17, 2025. You also restarted each six-survey trailing average after the break.
Plot the wordings as separate chart series with a visible gap. You exported the result as the PNG file ai-adoption-by-firm-size.png.
Secret Mission: Applied the same careful data-checking method in an optional challenge that pushes your survey-analysis skills further.

Ready to quiz yourself?