Click-Through vs View-Through Conversions: What to Count

It is written for the people who have to reconcile a platform dashboard with a pipeline report: performance marketers, marketing leaders and the agencies that run paid channels for them. If your question is only why four dashboards disagree, why GA4, Meta and Google conversions don't match your CRM is the shorter read. If you have to decide what to count, start here.

Both are conversions an ad platform credits to its own ad. What separates them is what the person did with the ad, and what that leaves behind for anyone else to check.

• Click-through conversion

  The person clicked the ad and converted inside the click window: 7 days by default on Meta, 28 on Reddit, 30 on Google Ads and Microsoft, 90 on LinkedIn. The click leaves a visit with a click ID and UTMs that GA4 and your CRM can see.

• View-through conversion

  The person was shown the ad, did not click, and converted inside the view-through window: 1 day by default on most platforms, 90 days on LinkedIn. Only the platform that served the impression knows it happened.

• Engaged-view conversion

  The video case in between. Google counts 10 seconds of a skippable YouTube ad, Meta's engage-through 5 seconds of video or a non-link interaction. Google puts these in its Conversions column.

The difference that matters in practice is not the window but the witness. A click-through has a visit behind it: a landing page, a click ID, UTM parameters, a session that GA4 recorded and a lead that your CRM can trace back to it. A view-through has only the platform's word, matched through its own pixel or conversion API against a person it believes it showed the ad to. The glossary entry on view-through conversion has the long definition, the full table of windows and the view-through rate, and attribution window covers how windows work in general.

Grouped by channel type, because that is how the decision reaches you: what a search, social, display or video line in the plan is really reporting. Defaults as read on each platform's own help pages on 28 September 2026.

Google documents view-through conversions for the Display Network, video, Demand Gen and App campaigns; search ads are not among them. Microsoft counts them on its Audience ads and on search ads extended to its audience network. Google adds a limit most reports never mention: view-throughs from browsers that do not allow cross-site cookies cannot be reported, so the column is not only generous in some places but blind in others.

Meta moved three of its defaults this year. The 7-day and 28-day view windows stopped being returned by its reporting API on 12 January 2026, so the view window is now 1 day or nothing. In March 2026 click-through was narrowed to link clicks, and the old engaged-view class was renamed engage-through, taking in shares, saves and other non-link interactions, with the video threshold cut from 10 seconds to 5. The default today is 7-day click, 1-day view and 1-day engagement. A year-on-year comparison of Meta conversions across those dates compares two different definitions.

Inside one platform the rule is the same everywhere: a click beats an impression. Google and Microsoft exclude anyone who clicked any of your ads on their network from the view-through count, TikTok attributes each conversion to one touchpoint, and a DSP like Basis counts a conversion as click-through when a click came first. Across platforms there is no rule at all, because no platform knows what the others served.

Impressions do work. Most people who see an ad never click it, and some of them act on it later, by searching your name, typing your URL or clicking a different ad. A report that credits only clicks gives those impressions nothing, and the channels that live on impressions end up looking worthless whatever they did.

• Video and CTV have almost nothing else to countA video ad is watched rather than clicked, and a television screen has no cursor. On CTV a view-through or a test is nearly all there is to measure with.
• Brand search collects credit it did not createThe person who saw a LinkedIn ad on Monday and searched your brand on Thursday is a Google Ads click-through. On a click report, brand search looks like the best campaign in the account.
• Upper-funnel spend gets cut firstChannels that start journeys are judged against channels that finish them. Ignore every view and the budget drifts, quarter by quarter, towards retargeting and brand terms, until nothing is filling the funnel they harvest.
• The bidding needs a signalWhere clicks are rare, conversions after an impression are the only feedback an algorithm gets, which is why Google lets YouTube-only Demand Gen campaigns opt into bidding on view-throughs.

That is a real argument and it deserves a real answer, not a dismissal. Ignoring view-throughs entirely is its own bias: it systematically under-credits the upper funnel and hands the difference to whatever channel sits closest to the form. The question is not whether impressions matter but whether a platform's view-through column is a good measure of how much.

Now the other side, with the month from the introduction. The windows are the defaults, except the DSP, set to the 30 days a programmatic platform commonly starts from. The figures are illustrative, and every column adds up.

The CRM has 100 leads, from every source including organic search and direct. The platforms claim 190, and 101 of the claims are view-throughs. Three mechanisms produce the 90 extra conversions, and none of them requires anybody to be dishonest.

• The same conversion, claimed by every platform that served an impression. A lead who scrolled past a Meta ad, saw a LinkedIn post and was served a banner in the same week is a view-through in three dashboards, and a click-through in a fourth if she then clicked a search ad. Each platform is right about its own impression. The sum is wrong.
• Retargeting counts people already converting. A retargeting audience is people who already visited. The ones who come back on their own inside the window after an impression are view-throughs whether the ad changed anything or not, and the DSP's 43 view-throughs against 3 clicks is what that looks like at scale.
• Windows that outlast the decision. LinkedIn's default credits a lead to any impression served in the previous 90 days. On a B2B account that is often before the buyer had a project at all, so the 16 view-throughs include people who decided for reasons LinkedIn never saw.

Two details in the table are worth a second look. Google's 18 view-throughs sit in their own column, outside Conversions, so the Google Ads headline reads 41, while Meta's and LinkedIn's headlines include theirs. And the click-through column is not clean either: 89 click-throughs against 100 leads looks close, but a person who clicked a Meta ad and then a Google ad is a click-through on both, and 30 of the 100 leads never clicked a paid ad at all.

Your CRM only knows the visits that arrived. A lead is recorded with the source of the session that produced it: UTM parameters, a click ID, a referrer, or a field somebody typed. An impression that was never clicked produces no visit, so it cannot be the source of anything in the CRM. The lead who saw the DSP's banner and then searched your brand arrives as search, correctly, because that is the visit that happened.

GA4 is almost the same, and the exact answer matters because it is asked so often. Its attribution settings are a reporting model, the channels that can receive credit and a key event lookback window; there is no view-through window among them. Google describes its last click model as crediting "the last channel that the customer clicked through (or engaged view through for YouTube)", and its data-driven model as measuring the contribution of each click interaction.

The exception is YouTube. With Google Ads linked, GA4 records engaged-view key events: somebody watched a YouTube ad for at least 10 seconds and triggered a key event on your site within 3 days. GA4 files those under google / cpc with the YouTube campaign's name, so a YouTube engaged view can take credit in GA4 much as a click would. A display banner, a Meta or LinkedIn ad, or a programmatic impression that was only seen never reaches GA4 at all.

The consequence is predictable. A channel whose results are mostly view-through, a DSP running display and CTV above all, reports dozens of conversions in its own dashboard and close to nothing in GA4 and the CRM. Neither side is broken. Programmatic attribution covers how to get a DSP's clicks into the CRM at all.

The way out of the argument is to stop looking for one true number and give each decision the number built for it. Five decisions cover almost everything a marketing team does with conversion data.

Bidding deserves a closer look, because the attribution setting is not only a reporting choice. Meta says in its own help centre that standard attribution optimises delivery for the time windows you select. A 1-day view setting therefore teaches delivery to find people who convert within a day of seeing an ad, and on an account heavy with retargeting those are often people who were converting anyway. Meta also offers an incremental attribution model, which it describes as optimising for conversions its models predict were caused by the ad. Google, by contrast, keeps view-throughs out of bidding for Display and video unless a campaign type opts into it.

The strongest fix for bidding is upstream of the window: send the platforms the conversions your business actually wants. When a lead qualifies or a deal closes in the CRM, that event goes back to Meta, Google, LinkedIn and Microsoft with the click ID and the deal value, and the algorithm learns from deals rather than from form fills and impressions. Sending CRM data back to Meta and Google and offline conversion tracking cover the mechanics.

A view-through count is a hypothesis: these people converted because they saw the ad. The only way to test it is to compare them with similar people who did not see it. There are four ways to do that, from rigorous to rough.

• A holdout or PSA testSplit the target audience at random, show one half your ad and the other half nothing, or a public service ad, and compare conversion rates in your CRM rather than in the platform. The PSA version gives the control group a real impression too, so both halves carry the same counting bias and only the message differs.
• A geo splitSwitch the channel on in some regions and off in comparable ones for several weeks, then compare leads per region in the CRM. Cruder than a holdout, but it needs no platform cooperation and it works for CTV.
• A platform lift studyGoogle runs Conversion Lift studies based on users or on geography, though not for every account; Meta describes an incremental attribution model; Reddit lists conversion lift among its measurement products. They are randomised, which is the point, but measured on the platform's own conversions.
• The cheap proxy: branded search and directWatch branded search clicks and direct visits in the weeks a channel goes on and off. Display and CTV that work send people to Google with your name; a channel whose view-throughs climb while branded search stays flat is claiming conversions it did not start.

Academic work points the same way. Gordon and colleagues, in Marketing Science, compared the observational methods advertisers use with the results of large randomised experiments run on Facebook, and found that the observational estimates often failed to recover what the experiments measured. A view-through column is an observational method of the simplest kind.

Read the result in both directions. If the holdout converts at nearly the same rate as the exposed group, most of the view-throughs were people converting anyway. If it converts far less, the impressions worked and a click-only report was under-crediting them. Either answer is worth more than the view-through column, and it is worth repeating when the creative, the audience or the budget changes a lot.

None of these makes view-throughs true or false. They make the column smaller where it is least likely to mean anything, and keep it apart from the numbers you add up.

• Shorten the long view windows. LinkedIn's default is 90 days of view, which on a B2B account credits impressions served a quarter before the lead; 7 days, or 1, is nearer to what an impression can plausibly do, and LinkedIn offers both. A DSP's 30-day view window deserves the same look. Google Ads, Meta, Microsoft and Reddit already default to 1 day.
• Decide Meta's view-through setting on purpose. Meta offers None or 1-day, and because standard attribution optimises delivery for the windows you select, the setting changes who sees the ads, not only what is reported. Compare attribution settings in Ads Manager shows 1-day view, 1-day click, 7-day click and 28-day click side by side without touching delivery.
• Pick LinkedIn's last ad set model. Under "Last touch, each ad set" every ad set that reached the member inside the window is credited with the conversion; "Last touch, last ad set" credits only the most recent one.
• Exclude existing customers and recent converters from prospecting. Upload the CRM's customer list and open pipeline as exclusions and exclude recent converters from retargeting, so impressions are not spent on people who are already buying and view-throughs are not collected from them either.
• Keep view-throughs out of blended totals. Microsoft adds Audience view-throughs to All conv. by default and has an account-level option to stop it; Google already keeps them out of Conversions for Display and video. On a DSP, Basis lets you value view-through conversions at a percentage of a click-through's revenue, 100% by default.
• Deduplicate across platforms in your own reporting. No platform will do it for you. One lead gets one credit, in the CRM, and the platforms' claims sit beside it as context, never summed with each other.

On a DSP the view-through column is not a side effect, it is usually most of the result. Simpli.fi, a programmatic platform, writes that the majority of digital campaign conversions are classified as view-through. Part of the reason is its own definition, under which a click-through only counts when the conversion happens in the same session as the click, but the direction is not in doubt: display is clicked rarely and CTV hardly at all.

The windows are long as well. StackAdapt's own blog says a DSP's view-through and click-through windows default to 30 days, and Campaign Manager 360's Floodlight uses one conversion window for clicks and impressions, 30 days by default. A 30-day view window on retargeting impressions, matched by the DSP's own identity data, is the setup most likely to credit conversions that would have happened anyway, and also the one where a real upper-funnel effect has no other way to show up.

So programmatic needs both halves. The clicks a DSP does produce can be tracked like any other click, with the DSP's macros filling UTM parameters so each lead reaches the CRM with the campaign and the site it came from; programmatic attribution shows how. The view-through part needs a test, and programmatic measurement covers which DSP metrics are worth reporting next to the CRM's.

LeadJourney is an all-in-one attribution platform for lead generation, B2B SaaS, e-commerce and agencies, and it sits on the click side of this argument. Tracking runs server-side on your own domain at 95%+ accuracy. On the first visit it stores the UTM parameters, the referrer, the landing page and the ad click IDs it knows (gclid, gbraid, wbraid, fbclid, msclkid, li_fat_id, ttclid) first-party on the LeadJourney Click ID, kept for weeks or months, and it writes the source onto the lead in the CRM. The visitor is joined to the lead at the form, the call or the booking, and the record follows the CRM stages to the closed deal, with first click, last click, linear, position-based and time decay switchable without re-tracking.

The loop runs back to the platforms as well: CRM stages and closed deals go to Meta, Google, LinkedIn and Microsoft as conversions with the deal value, which is the bidding fix from the section above. DSPs are tracked as custom traffic channels: LeadJourney does not connect to The Trade Desk, StackAdapt or DV360 today, so their clicks arrive through UTMs and the spend is entered as a channel cost.

What it cannot say, stated plainly: LeadJourney does not see impressions and counts no view-through conversions. It does not run lift studies or incrementality tests. That is exactly why its numbers are lower than the platforms', and why the gap is the thing worth looking at. In the month above, 70 of the 100 leads have a paid click in their journey. Credited to the first one, 36 go to Google Ads, 22 to Meta, 9 to LinkedIn and 3 to the DSP; the other 30 came in through organic search, direct and referral without a paid click. Meta claims 58 against the 22 journeys its clicks opened; the DSP claims 46 against 3. That difference, per channel, is the view-through and cross-platform share of each claim, and pixel and CRM mismatch shows how to read it.

Setup takes about 21 minutes, from €129 a month, with a 14-day free trial and no credit card, and the reports can be clicked through on demo data in the live demo first.

1. Know which column holds the view-throughs on each platform: a column of their own on Google Ads, Results on Meta, Conversions on LinkedIn, All conv. on Microsoft.
2. Write down every view window you run, and shorten the long ones: LinkedIn's 90 days first, then any DSP at 30.
3. Never add platform conversions together. Each platform deduplicates only against itself.
4. Count leads and revenue once, in the CRM, from click-based journeys joined to the deal.
5. Send qualified CRM conversions back to Meta, Google, LinkedIn and Microsoft, so bidding learns from deals.
6. Exclude customers and recent converters from prospecting and retargeting.
7. Test any channel whose case rests on view-throughs with a holdout, a geo split or a lift study before scaling it.
8. Watch branded search and direct in the weeks a channel goes on or off.
9. Re-read the platform defaults twice a year. Meta changed three of them in 2026 alone.

The definitions: view-through conversion, attribution window, click ID and multi-touch attribution. Programmatic in depth: programmatic attribution and programmatic measurement. Why the dashboards disagree: why GA4, Meta and Google conversions don't match your CRM, cross-channel attribution and, for one platform's windows in detail, Reddit Ads attribution. On the product side: pixel and CRM mismatch, optimising on revenue and multi-touch attribution.