Every fall, hearth dealers get reports that show their advertising working. Leads are up from August. Clicks are up from September. Some of those campaigns are working and some are not. A month over month comparison cannot tell them apart, because fireplace shopping climbs every fall whether anyone advertises or not. The same weeks last year can. Here is how to run that check in Google Analytics and Search Console. We also cover what to do when there is no last year to compare against.
Fall makes every campaign look like it works
Fireplace shopping follows the weather. Google Trends shows the shape for the whole country. US search interest in "gas fireplace" is lowest from April through August. It climbs through the fall and peaks between November and January.
The climb is steep. In each year from 2022 through 2025, the October average ran 1.6 to 2 times September's. November ran about 3 times August's.
So pick any two back to back months between August and December and the second one wins. A campaign that started in August will show growth. So will one that has run unchanged for years. So will one that is quietly getting worse.
A campaign in its fifth year still nearly doubled in October
Bromwell's is a fireplace retailer in the Washington, D.C. area and a client of ours since 2021. Its 2025 leads by month are on our results page. August brought in 69 leads. September brought in 119. October brought in 227.
That is 72 percent more in September and 91 percent more again in October. This was a campaign in its fifth year, not a new launch.
Now picture a new agency taking over that account on September 1. Its first report would show leads up 72 percent. Its second would show another 91 percent. None of that would prove the new agency did anything. The same holds when the agency is us.
Compare the same weeks last year
The fix is to hold the calendar still. Compare this October to last October, not to this September. The season sits in both numbers, so most of it cancels out.
Then go one step further. Compare the gap with last year before the change to the gap after it. A worked example shows why. These numbers are made up for illustration. They are not a client result.
- This year: a dealer's site got 400 organic clicks in the 4 weeks before a new campaign started in September. It got 560 in the 4 weeks after. That is up 40 percent. It looks like a win.
- Last year: the same 4 weeks before got 420 clicks. The same 4 weeks after got 650. That is up 55 percent with no campaign at all.
- The gap: before the change, the dealer ran 5 percent behind last year (400 against 420). After it, the dealer ran 14 percent behind (560 against 650).
The raw before and after says plus 40 percent. The honest reading is that the site fell further behind last year's pace once the change went in. That is the opposite story. It is also the one a dealer needs before deciding anything.
Sometimes the causes cannot be separated. If the site was rebuilt, ads started and the season turned in the same month, no report can pull those three apart. When that happens, say so and make no claim. A flattering number that only holds because of the calendar does a dealer no good.
Use whole weeks so the weekdays line up
A showroom's Saturday is not its Tuesday. October 1 fell on a Wednesday in 2025. It falls on a Thursday in 2026. Compare one day to the other and you are comparing different kinds of days.
Pick ranges that run a whole number of weeks: 28 days, 56 days or 91 days. Each one holds every day of the week the same number of times. The weekday mix matches even when the dates do not.
Holidays still move. Thanksgiving was November 27 in 2025 and falls on November 26 in 2026. Give Thanksgiving week its own look rather than letting it land on different sides of a cutoff.
Set up the comparison in GA4
- Open a report such as Traffic acquisition. Click the date picker in the top right.
- Choose a custom range in whole weeks using the calendar.
- Turn on Compare and choose Previous year. Google describes it as comparing your range to the same range last year.
- Click Apply. Then switch the line chart to week view so one odd day does not drive the picture.
Google has removed the old option that matched the day of the week for the previous year. Previous year now lines up calendar dates. That is why the whole-week rule above matters.
One trap: explorations follow your data retention setting. GA4 lets a property keep event data for as little as 2 months. Google says that setting does not affect standard reports, comparisons included, but it does limit explorations and funnel reports. If last year comes up empty in an exploration, run the comparison in a standard report.
Set up the same comparison for organic search
- Open Performance, then Search results.
- Open the date filter and choose Compare. Set this year's weeks and the same weeks last year.
- Click Apply. The table adds a Difference column for every query and page.
- Set the chart to Weekly. Google recommends weekly or monthly views when comparing date ranges, because they remove the day of week effect.
Search Console keeps 16 months of data. That covers last fall, but older data drops off. Export last year's fall numbers before they are gone.
Check the signals the campaign does not touch
Year over year still mixes the campaign with everything else that changed. The weather does not repeat on schedule. A competitor may have closed or opened. Three numbers show how the season is running for the store as a whole.
- Searches for your store's name. In Search Console, filter queries containing your business name. These are people who already know you.
- Direct traffic. In GA4, the Direct channel counts visitors who arrive from a saved link or by typing your address.
- Total phone calls. Your phone system or carrier counts every call to the store, from every source.
Read them for the same weeks, against the same weeks last year. If they are up 20 percent and campaign leads are up 20 percent, the campaign held its share of a busier season. If campaign leads are up 50 percent while the baseline is flat, the difference belongs to the campaign. Good advertising can lift brand searches over months. A jump that starts the week the weather turned is the season.
If search ads started inside the window, read paid and organic together for your own name. Shoppers who used to click your organic listing can click the ad above it instead. An organic drop on brand searches is then not a ranking loss.
What to do when there is no last year
A new domain, rebuilt tracking or a changed lead definition can each break the comparison. These are the options we use, strongest first.
- Compare only what was measured the same way. A new GA4 setup may count leads differently. Visits by channel usually survive a change like that. A lead count that changed its definition does not.
- Lean on Search Console. It measures searches on Google's side, not with a tag on your site. A new analytics setup does not reset it.
- Use your own records. Phone bills and monthly sales go back years and nobody retagged them.
- Use Google Trends as the yardstick. Pull your state for the same weeks. If search interest doubled and your leads doubled, the campaign kept pace. That is not a gain.
- Label the gap. Mark the report "no prior-year comparison available" rather than presenting a month over month gain as a result.
Ask for the year over year line in every report
Whoever runs your advertising, us included, should show each month next to the same weeks last year with the baseline signals beside it. We made the same point about lead costs in what a fireplace lead should cost. Our post on marketing analytics for fireplace dealers covers what to track in the first place.
If you want us to run this check on your own numbers, send us read access to Search Console and GA4. We will send back what we find whether or not it turns into a conversation about working together.
Sources
- Bromwell's 2025 leads by month, its location and the year it became a client are from our results page. The percent changes are our arithmetic on that table.
- Google Trends, "gas fireplace", United States, past 5 years, weekly values read October 21, 2026. The low months, the peak months and the October and November ratios are our arithmetic on monthly averages of those weekly values. Trends numbers are relative and sampled, so a fresh pull can differ slightly.
- Google Analytics Help, Change and compare date ranges in reports, for the date picker, Compare, the Previous year option, the removal of Previous year (match day of week) and the day, week and month chart views.
- Google Analytics Help, Data retention, for the 2 month minimum and for retention not affecting standard reports while limiting explorations and funnel reports.
- Google Analytics Help, [GA4] Default channel group, for Direct as arrival by a saved link or by entering the URL.
- Search Console Help, Performance report (Search results): Advanced filtering and comparison, for comparing time periods, the Difference column, filtering queries that contain a word and the weekly or monthly recommendation.
- Google Search Central Blog, Introducing the new Search Console (January 8, 2018), for 16 months of Performance data and year over year comparisons.
- U.S. Office of Personnel Management, Federal Holidays, for Thanksgiving on November 27, 2025 and November 26, 2026.
- The clicks in the worked example are invented to show the method. They are not from any account.
Every market is different. A year over year check will not tell you why a number moved. It does keep the calendar from telling the story for you.