{"id":1265,"date":"2026-10-12T11:54:28","date_gmt":"2026-10-12T15:54:28","guid":{"rendered":"https:\/\/thedoctorads.com\/blog\/google-ads-all-conversions-mix-drift"},"modified":"2026-10-12T11:54:28","modified_gmt":"2026-10-12T15:54:28","slug":"google-ads-all-conversions-mix-drift","status":"publish","type":"post","link":"https:\/\/thedoctorads.com\/blog\/google-ads-all-conversions-mix-drift","title":{"rendered":"Google Ads All Conversions Year Over Year: The Mix Moved 6.42 Points"},"content":{"rendered":"<div class=\"tldr\">\n<p>Your annual report shows more conversions than last year, and it feels like growth. But the report may be counting a different mix of things. In the median account we examined, the mix inside All conversions moved <strong><span data-claim-id=\"MX-01\">6.42 percentage points<\/span><\/strong> from 2024 to 2025. Roughly one tracked event in sixteen would have to switch category for the two years to match.<\/p>\n<p>All conversions is Google&#8217;s broad column, holding everything in your main Conversions column plus the other actions you track. Its name stays fixed while its contents move, and a year earlier they moved further: <span data-claim-id=\"MX-02\">20.95 points<\/span> from 2023 to 2024, about one event in five. That slowdown holds when the same accounts are compared in both years. We call this <strong>The Moving Conversion Denominator<\/strong>: the total you divide spend by keeps its label while its ingredients change.<\/p>\n<p>Before a higher total turns into a budget decision, put the categories inside it side by side. Our records show that the mix moves, not whether business, tracking or labels moved it.<\/p>\n<\/div>\n<aside class=\"key-stat\">\n<p><strong>The Moving Conversion Denominator: the median account&#8217;s All conversions mix moved <span data-claim-id=\"MX-01\">6.42 percentage points<\/span> from 2024 to 2025, after <span data-claim-id=\"MX-02\">20.95 points<\/span> from 2023 to 2024.<\/strong> For each pair of years to match, roughly one tracked event in sixteen would have to switch category, and a year earlier about one in five.<\/p>\n<p class=\"source-line\">Source: Doctor Ads Profit Forensics. Full calendar years 2023 to 2025; category shares compared within each account after reconciling to campaign-level All conversions; median account.<\/p>\n<\/aside>\n<h2>The total keeps its name while its contents change<\/h2>\n<p>Picture the annual review. The dashboard shows more conversions than last year, and the conversation moves on to next year&#8217;s budget. Nobody asks what a conversion was in each year, because the column heading never changed.<\/p>\n<p>That heading covers a lot. Google&#8217;s help page on <a href=\"https:\/\/support.google.com\/google-ads\/answer\/3419678?hl=en\" target=\"_blank\" rel=\"noopener\">All conversions<\/a> describes the column as a view of all the conversions your campaigns drive, including actions you chose to keep out of the main Conversions column. The column can also be segmented by conversion action. A purchase, a lead form and a page visit can all land in the same total, and they are not worth the same to your business.<\/p>\n<p>A rising total can mean that a valuable outcome grew, that an early step in the customer journey grew, or that a newly tracked action joined the count. The total looks the same in all three cases. Only the categories inside it tell them apart.<\/p>\n<p>A column heading is a label, not a unit of business. That is why we call the pattern <strong>The Moving Conversion Denominator<\/strong>. A conversion total sits at the bottom of every cost-per-conversion figure built on it, so when its ingredients change, the unit behind your efficiency figure changes with them.<\/p>\n<p>This is a different question from our earlier studies. We looked at <a href=\"https:\/\/thedoctorads.com\/blog\/google-ads-conversions-column-lead-counting\">what goes into the Conversions column<\/a> and at <a href=\"https:\/\/thedoctorads.com\/blog\/google-ads-conversion-goals-what-campaigns-bid-on\">what campaigns are marked to bid on<\/a>, both as settings at one moment. A <a href=\"https:\/\/getsecondstep.com\/attribution-gap-index\/\" target=\"_blank\" rel=\"noopener\">Second Step study of conversion categories<\/a> asked how much of a portfolio&#8217;s recorded activity could represent revenue, within one reporting window. Here we follow what each account actually recorded, from one full year to the next.<\/p>\n<h2>One event in sixteen, after one in five the year before<\/h2>\n<p>Between 2024 and 2025, the median account&#8217;s category mix moved <strong><span data-claim-id=\"MX-01\">6.42 percentage points<\/span><\/strong>. Put simply, roughly one tracked event in sixteen would have to switch category for the two years to look alike.<\/p>\n<p>The median hides a long tail. The middle half of accounts ran from <span data-claim-id=\"MX-01\">3.75 to 25.70 points<\/span>, so in the quarter of accounts that moved most, one event in four or more would have to switch. A calm median can sit beside accounts whose total now counts something quite different.<\/p>\n<p>The year before moved much further. Between 2023 and 2024, the median move was <strong><span data-claim-id=\"MX-02\">20.95 points<\/span><\/strong>, about one tracked event in five. The middle half ran from <span data-claim-id=\"MX-02\">8.00 to 35.81 points<\/span>, so in the quarter that moved most, more than one event in three would have to switch.<\/p>\n<figure class=\"chart\"><img decoding=\"async\" src=\"https:\/\/thedoctorads.com\/blog\/wp-content\/uploads\/2026\/10\/tda-w13-conversion-mix-years.png\" alt=\"The median account&#039;s All conversions category mix moved 20.95 percentage points from 2023 to 2024 and 6.42 from 2024 to 2025; the middle half of accounts ranged from 8.00 to 35.81 and from 3.75 to 25.70 points.\" title=\"\"><figcaption><strong>The mix moved less in the later year.<\/strong> About one event in five would have to switch category to match 2023 with 2024, and about one in sixteen to match 2024 with 2025. Dots show the median account; lines show the middle half. Source: Doctor Ads Profit Forensics.<\/figcaption><\/figure>\n<p>That rules out a simple correction. There is no fixed yearly drift to subtract from a report, because the size of the move changes from one pair of years to the next.<\/p>\n<p>The measure compares shares, not volumes. We turn each year&#8217;s total into a mix that adds up to <span data-claim-id=\"MX-08\">100%<\/span>, line up every category&#8217;s share against the year before, add up the gaps and halve the sum. Halving keeps one category&#8217;s loss and another&#8217;s matching gain from counting as two separate moves.<\/p>\n<p>So <span data-claim-id=\"MX-01\">6.42 points<\/span> is the share of the mix that would have to change category for the two years to agree. It is not a fall in conversions, and it does not mean that share of your sales was recorded wrongly.<\/p>\n<h2>The same accounts show the same slowdown<\/h2>\n<p>A fair first suspicion is the sample. An account enters a comparison only when it records enough events in both years, so the earlier and later comparisons do not hold exactly the same accounts. A slowdown could simply mean that different accounts were measured.<\/p>\n<p>So we kept only the accounts eligible for both comparisons. Their median move was <strong><span data-claim-id=\"MX-04\">22.74 points<\/span><\/strong> from 2023 to 2024 and <strong><span data-claim-id=\"MX-04\">6.28 points<\/span><\/strong> from 2024 to 2025. Same accounts, same pattern: the slowdown is not just a change in who was measured.<\/p>\n<figure class=\"chart\"><img decoding=\"async\" src=\"https:\/\/thedoctorads.com\/blog\/wp-content\/uploads\/2026\/10\/tda-w13-conversion-mix-common.png\" alt=\"With the same eligible accounts in both annual comparisons, the median All conversions category mix moved 22.74 percentage points from 2023 to 2024 and 6.28 from 2024 to 2025.\" title=\"\"><figcaption><strong>The slowdown holds when the account list stays fixed.<\/strong> The same eligible accounts in both annual comparisons. Source: Doctor Ads Profit Forensics.<\/figcaption><\/figure>\n<p>The first move was larger, and <span data-claim-id=\"MX-22\">the extra movement is not confined to the half-year that contains 1 July 2023<\/span>, the date <a href=\"https:\/\/support.google.com\/google-ads\/answer\/12495138\" target=\"_blank\" rel=\"noopener\">standard Universal Analytics properties stopped processing data<\/a>. Our category records cannot say what did change. A real shift in what customers did, a new tracking setup or a relabelled action could each move the mix. Google&#8217;s <a href=\"https:\/\/support.google.com\/google-ads\/answer\/14571185?hl=en\" target=\"_blank\" rel=\"noopener\">guidance on changing conversion goals and actions<\/a> covers changes of that kind, such as a new conversion source or a new category for an existing action.<\/p>\n<h2>Compare matching ingredients before comparing the total<\/h2>\n<p>A conversion total needs an ingredient list before it can carry a year-over-year comparison. For each category, put last year&#8217;s count next to this year&#8217;s, then add each category&#8217;s share of All conversions in both years. A category can grow in volume while shrinking in share, so you need both columns.<\/p>\n<table>\n<thead>\n<tr>\n<th>Report line<\/th>\n<th>Question it answers<\/th>\n<th>Decision it supports<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Total All conversions<\/td>\n<td>How much tracked activity was credited?<\/td>\n<td>Establish the reporting total before interpreting it.<\/td>\n<\/tr>\n<tr>\n<td>Counts by matching category<\/td>\n<td>Which kinds of activity rose or fell?<\/td>\n<td>Locate the source of the total&#8217;s movement.<\/td>\n<\/tr>\n<tr>\n<td>Category shares<\/td>\n<td>Did the ingredients change?<\/td>\n<td>Decide whether the two annual totals mean the same thing.<\/td>\n<\/tr>\n<tr>\n<td>Actions inside a changing category<\/td>\n<td>Are the underlying events still comparable?<\/td>\n<td>Separate business movement from measurement changes.<\/td>\n<\/tr>\n<tr>\n<td>Confirmed business outcomes<\/td>\n<td>Did customers, revenue or contribution improve?<\/td>\n<td>Make the budget decision on the outcome that matters.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Keep categories that appear or disappear. Dropping unmatched rows makes the rest of the report look steadier than the full count was. If the same business event moved to another category, reconcile the action before you read the move as a change in customer behaviour.<\/p>\n<p>Categories are only the first layer. Different actions can share one category, and their balance can shift while the category&#8217;s share holds still. Google&#8217;s <a href=\"https:\/\/support.google.com\/google-ads\/answer\/11461796?hl=en\" target=\"_blank\" rel=\"noopener\">primary and secondary action settings<\/a> govern reporting and optimization roles, which is a separate check from whether the event itself stayed the same.<\/p>\n<p>It is the same habit behind the way we <a href=\"https:\/\/thedoctorads.com\/blog\/google-ads-quality-scores-tested\">test platform scores against observed results<\/a>. A familiar metric can be useful without being the outcome you meant to measure. Keep the total as a summary, and rest the growth claim on matching events.<\/p>\n<h2>When this does not apply<\/h2>\n<p><strong>Your report counts one event that never changed.<\/strong> If a single event with the same definition, counting rule and coverage makes up the whole comparison, its annual count can stand on its own. A quick category check still confirms the assumption.<\/p>\n<p><strong>Your category labels changed during the period.<\/strong> Start with an action-level reconciliation. A relabel can create a large distance while the business event stayed the same, and offsetting changes can hide real movement elsewhere.<\/p>\n<p><strong>You are comparing part of a year.<\/strong> These results compare full calendar years. Match the months before borrowing them, because a seasonal change in what customers do is not the same as a yearly change in what gets counted.<\/p>\n<p><strong>Your category export does not add back to All conversions.<\/strong> Fix the missing or mismatched rows first. Turning an incomplete export into percentages makes a gap look like a precise mix.<\/p>\n<p><strong>Your account records fewer than <span data-claim-id=\"MX-08\">100 All conversions<\/span> a year.<\/strong> We only measured accounts with at least that many in each year. Below that line, a handful of events can swing the shares, so a large distance may say little about the business.<\/p>\n<h2>What to check in your account<\/h2>\n<ol>\n<li>Export All conversions for two matching full years, segmented by conversion action and by category.<\/li>\n<li>Check that the rows add back to the unsegmented total at the same campaign and month level. Fix any gap before you compare shares.<\/li>\n<li>Build one table with each category&#8217;s count and share in both years. Keep categories that exist in only one year.<\/li>\n<li>Open the actions behind the biggest share moves. Note whether the event, its label, its counting rule or its tracking coverage changed.<\/li>\n<li>Report confirmed business outcomes separately from the broader total, and say which like-for-like events support the growth claim.<\/li>\n<li>Make the comparison in your reports, not by relabelling actions in the account. If you do change a category or a goal for business reasons, read Google&#8217;s guidance on changing conversion goals and actions used for Smart Bidding first. The bidding needs time to learn after such a change.<\/li>\n<\/ol>\n<p><!-- cta-anchor --><\/p>\n<h2>Researcher&#8217;s take<\/h2>\n<p>I would rather see a modest growth figure built from matching events than a big one built from a moving mix. All conversions is useful precisely because it is broad. The trouble starts when that breadth is treated as a fixed unit of business.<\/p>\n<p>In an annual review, my first question is what grew inside the total. If the team cannot answer, I hold the verdict and ask for the category table. The count may still tell a good story, but it has to be the same kind of story in both years.<\/p>\n<p><em>Igor Ivitskiy, Doctor Ads Profit Forensics<\/em><\/p>\n<h2>FAQ<\/h2>\n<h3>How do I compare All conversions year over year?<\/h3>\n<p>Compare matching categories and actions alongside the total, after checking that the breakdown adds back to it. In the latest full-year comparison, the median account&#8217;s mix moved <span data-claim-id=\"MX-01\">6.42 points<\/span>, roughly one tracked event in sixteen, which is enough that the total alone no longer tells the whole story.<\/p>\n<h3>What is the difference between Conversions and All conversions?<\/h3>\n<p>All conversions includes everything in your main Conversions column plus the conversion actions you chose to leave out of it. Because it holds many kinds of actions, the mix inside it can change while the name stays the same. In the median account it moved <span data-claim-id=\"MX-01\">6.42 points<\/span> from 2024 to 2025.<\/p>\n<h3>Was the earlier year&#8217;s composition more stable?<\/h3>\n<p>No. From 2023 to 2024 the median distance was <span data-claim-id=\"MX-02\">20.95 points<\/span>, about one event in five. The smaller later move also holds for the same accounts across both years; the data do not establish why the mix moved less.<\/p>\n<h3>Does a 6.42-point distance mean 6.42% fewer sales?<\/h3>\n<p>No. <span data-claim-id=\"MX-01\">6.42 points<\/span> is a distance between category shares. It measures neither lost sales nor a change in the number of events.<\/p>\n<h3>Can a stable category mix prove tracking stayed unchanged?<\/h3>\n<p>No. Each year&#8217;s shares add up to <span data-claim-id=\"MX-08\">100%<\/span>, so changes among actions inside one category can stay hidden. Check the underlying event definitions before treating a stable mix as proof.<\/p>\n<h2>Method<\/h2>\n<p>Source: a <span data-claim-id=\"MX-16\">10 July 2026<\/span> snapshot of the managed portfolio, with one account refreshed in September 2026. That account appears only in the September data, so nothing is counted twice. Only months from <span data-claim-id=\"MX-16\">January 2023 through June 2026<\/span> are used, so the refresh adds no new months; the comparisons use the full calendar years 2023, 2024 and 2025. The portfolio is managed and audited by the agency, not a random sample of advertisers, and no client, niche, brand or country is identified.<\/p>\n<p>The unit is the account. An account enters a comparison with at least <span data-claim-id=\"MX-08\">100 All conversions in each year<\/span>: <span data-claim-id=\"MX-02\">n = 42<\/span> for 2023 to 2024, <span data-claim-id=\"MX-01\">n = 46<\/span> for 2024 to 2025 and <span data-claim-id=\"MX-04\">n = 37<\/span> eligible for both. Each category&#8217;s count is divided by the account&#8217;s annual total, absent categories count as zero, and the distance is half the sum of the share differences, each taken without its sign. Conversion counts are never compared across accounts.<\/p>\n<p>Category totals were reconciled to campaign-level All conversions on <span data-claim-id=\"MX-07\">16,756 campaign-month pairs<\/span>. <span data-claim-id=\"MX-07\">Six pairs in four accounts<\/span> missed a <span data-claim-id=\"MX-07\">0.5%<\/span> relative tolerance, with a minimum difference of <span data-claim-id=\"MX-07\">0.000001<\/span>, and those accounts were excluded entirely. No action keys were left unmatched, and identifiers are checked for uniqueness before every join.<\/p>\n<p>Medians and quartiles give each account equal weight and are computed <span data-claim-id=\"MX-16\">deterministically<\/span>, so a rerun returns identical figures. Leaving out one account at a time keeps the medians within <span data-claim-id=\"MX-05\">19.16 to 22.74<\/span> and <span data-claim-id=\"MX-05\">6.28 to 6.56 points<\/span>, so no single account carries the result. Dropping the largest account, ranked by portfolio spend from January 2023 to June 2026 or by audited campaign spend where an audit exists, gives <span data-claim-id=\"MX-05\">19.16 and 6.56<\/span>.<\/p>\n<p>Keeping only the highest-spend eligible account in each group of related accounts leaves <span data-claim-id=\"MX-06\">n = 42 and n = 44, one account per related group<\/span>, with medians of <span data-claim-id=\"MX-06\">20.95 and 6.94 points<\/span>. The equal-account means were <span data-claim-id=\"MX-03\">27.82<\/span> and <span data-claim-id=\"MX-03\">21.61 points<\/span>: a minority of large moves keeps the later mean well above its median. The means average account-level distances and do not pool events across advertisers.<\/p>\n<p>The half-year check uses accounts with at least 100 All conversions in both 2023 and 2024 and at least 50 in each half compared. Medians were <span data-claim-id=\"MX-17\">13.42 points<\/span> from early to late 2023, <span data-claim-id=\"MX-18\">14.11<\/span> from late 2023 to early 2024 and <span data-claim-id=\"MX-19\">3.82<\/span> from early to late 2024. Each pair has its own eligible accounts, so these figures describe timing and do not test a cause.<\/p>\n<p>Category records cannot separate a change in business from a change in setup or labels, and they miss shifts among actions inside one category. These are descriptive findings and do not guarantee a business result.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>All conversions year over year: the median account&#8217;s mix moved 6.42 points from 2024 to 2025, so one event in sixteen would need a new category to match.<\/p>\n","protected":false},"author":0,"featured_media":1264,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[16],"tags":[],"class_list":["post-1265","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-metrics-and-scores","generate-columns","tablet-grid-50","mobile-grid-100","grid-parent","grid-33"],"_links":{"self":[{"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/posts\/1265","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/comments?post=1265"}],"version-history":[{"count":0,"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/posts\/1265\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/media\/1264"}],"wp:attachment":[{"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/media?parent=1265"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/categories?post=1265"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/thedoctorads.com\/blog\/wp-json\/wp\/v2\/tags?post=1265"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}