Traffic can hold steady while the advertising income disappears. Waiting for the next seasonal upswing feels reasonable; replacing the ad network feels decisive. Both can miss the more useful question: did each visit produce fewer served ads, or did those ads earn less?
Start there. A falling RPM is the alarm, not the diagnosis. Before changing partners or trying to buy more traffic, separate the number of visits, the number of ads served and the money those ads generated.
Count what the visit actually earned
A utility-site publisher using Journey by Mediavine reported unchanged traffic while impressions fell from roughly 2–2.5 per session to 0.3. Some pages showed zero RPM. That is a much sharper diagnostic clue than the revenue chart alone.
The account is not an audited explanation of Journey's October performance. Its value is the distinction it exposes: a visit can still arrive while far fewer ads are served during it.
Journey's earnings documentation separates ad-server earnings and impressions from the analytics used for traffic and RPM. RPM is revenue divided by traffic, multiplied by 1,000. An analytics change can move that calculation without changing the money earned.
So put revenue, sessions, served impressions and CPM beside each other for the same dates. If earnings are stable and only RPM falls, investigate the traffic denominator first. If revenue and impressions per session both fall, inspect delivery and reader behavior before treating cheaper advertiser bids as the whole explanation.
A placeholder is not an impression
A page can contain every expected ad slot without serving the same number of ads. Journey explicitly distinguishes slots from impressions . Its placeholder test shows where slots exist; impressions depend on how readers move through the page and include refreshes.
Choose high-traffic pages with the largest before-and-after change. Open them privately on mobile and desktop, with ad blocking disabled, and actually move through the content. Compare affected templates with pages that still earn. Check whether a new overlay, navigation shortcut or layout sends readers past the places that used to monetize.
Then review recent theme, plugin and ad-script changes. Check the current ads.txt configuration against the provider's requirements. These are checks to perform, not repairs already proved to work.
Keep evidence at page level. The exact URL, date, device and missing placement are more useful than a screenshot saying revenue is down.
Give the demand explanation a fair test
The strongest argument for waiting is straightforward: visits do not guarantee buyers. A temporary campaign-budget reset or weaker advertiser demand could lower what available impressions earn. Moving networks immediately may simply move the same inventory into another disappointing auction.
Journey's ramp-up guidance describes variable results shaped by site history, traffic volume, geography and seasonality, and stresses that advertisers need the correct ads.txt file. It offers context, not a promised recovery date.
Waiting deserves more weight when served impressions and the traffic mix remain comparable while the price paid for those impressions falls. It deserves less when a heavily visited page suddenly serves very little. A weak market and a delivery change can coexist; choosing one explanation prematurely hides the other.
Another site's result cannot settle this. Stable, rising and falling yields across different audiences are not a controlled comparison of networks. Nor does a sudden homepage traffic spike prove bots. Check where the extra visits came from and whether they monetized before declaring either a market slump or an invalid-traffic incident.
Make support part of the switching decision
Prepare a short escalation record: affected URLs, comparable dates, revenue, served impressions, traffic mix and recent changes. Ask the provider to distinguish ad requests, filled opportunities and the behavior it can actually observe. Do not ask only when RPM will recover.
If useful help remains unavailable, support quality becomes a legitimate commercial reason to leave. Even when a partner is not responsible for the decline, an inability to help you investigate can become expensive.
But secure the alternative before counting its projected revenue. Confirm that the site, audience geography and formats are acceptable, and understand the operational cost of changing. A higher-tier label is not a repair for a fault you have not located.
The next decision should follow the missing evidence: correct a measurement problem, investigate lost delivery, monitor a documented demand decline, or choose a partner that can support the business. Do not spend another month waiting for a better RPM when the more revealing number is how many ads each visit actually received.