Repeat Purchase Rate Cohorts: Reading Meta vs Google Buyer Quality

First-order ROAS tells you which ad platform wins the first sale, and it is the wrong number to divvy a budget by. A customer who buys once from a Meta ad and never returns, and a customer who comes from Google search and orders three more times that year, look identical in a first-order report while being worth wildly different amounts to the store. Repeat purchase rate, measured as a cohort by acquisition month and split by platform, is the metric that separates the two, and the Shopify Meta ads optimization guide treats cohort quality as the final check on any optimization decision. This article explains what a repeat purchase cohort is, why acquisition source changes it, and how to read one from your Pulse data before you move another dollar between Meta and Google.

The Quick Answer

A repeat purchase rate cohort groups customers by the month of their first order and counts how many of them order again in the months that follow, separated by the platform that acquired them. The read is simple: an acquisition month where Google buyers repurchase at double the rate of Meta buyers means Google is recruiting real customers while Meta is recruiting discount-driven one-time orders, even if first-order ROAS says Meta won. Check it monthly, compare the two platform lines, and let the gap between them, not the first-order report, decide where incremental budget goes.

What a Repeat Purchase Cohort Is

A cohort is a group of customers defined by a shared start point. For this metric the start point is the month of the first order, and every customer whose first order happened in, say, March 2026 belongs to the March cohort no matter what they do afterward. The cohort’s repeat purchase rate at four months is the share of those customers who placed a second or later order within four months of their first.

The two numbers on each axis are what make the view honest. Grouping by acquisition month stops the metric from blending people who arrived during a sale and people who arrived at full price, because a monthly group shares the same promotion calendar, the same seasonality, and roughly the same ad creative. Splitting by platform is the second axis: it separates customers whose first order came through a Meta attribution click from those whose first order came through a Google click, so each platform gets its own repurchase line instead of a store average that hides the difference.

Marketing research has put hard numbers on why this matters: the probability of selling to an existing customer sits in the sixty to seventy percent range, while the probability of selling to a brand new prospect is somewhere between five and twenty percent. A customer who repurchases is therefore worth several times a customer who does not, and the platform that recruits repurchasers is worth several times what its first-order report suggests.

Why Acquisition Source Changes Buyer Quality

Meta and Google do not just deliver impressions from different placements. They deliver intent from different moments, and the moment shapes what the customer expects.

A Google search ad or shopping result usually catches a shopper with an active, nameable need: a specific product, a comparison, a price check. The store wins the shopper by being available at the exact moment the need surfaces. These buyers often repurchase, though not always, because the loyalty formed is to the need being met, and only partly to the brand.

A Meta ad usually does the opposite. It manufactures desire for a product the customer had not thought to look for. The discovery happens on impulse, often against lifestyle creative, and the customer completes the order while the impulse is alive. Whether that customer returns depends on brand fit, not on the search intent that brought the Google buyer. Both platforms can recruit loyalists and both can recruit one-time bargain hunters, but they do so in different proportions, and those proportions are exactly what a cohort view measures instead of guesses.

The practical consequence is that first-order ROAS is a biased comparison. Meta’s best campaigns are engineered for a fast, cheap first conversion, which flatters the first-order report. Google’s wins often come from customers with slower, larger intent, which the first-order report undersells. The cohort curve is the correction for that bias.

Reading the Cohort Curve by Acquisition Month

A cohort view places acquisition months down one axis and months-since-first-order across the other. Reading it well takes three habits.

First, compare months-since-first-order at fixed offsets, not cumulative totals. Compare every cohort at month two, month four, and month six. A cohort that hits a twelve percent repeat rate at month four while the store’s trend line sits at seven is an acquisition month worth repeating, and you can look up which campaigns carried it.

Second, read the platform split inside each cohort. When one acquisition month shows a strong overall repeat rate, check whether Meta and Google contributed evenly. A month that looks healthy on average may be a Meta-led month with a flat Google line, or the reverse, and the average conceals which. The gap between the two lines in the same month is the cleanest evidence of a buyer quality difference, because both groups bought during identical promotions at identical prices.

Third, isolate what you changed. Every acquisition month carries its own mix of offers, creatives, and seasonal traffic. When a cohort’s repeat rate moves, attribute the movement to whatever changed that month: a sitewide discount that pulled in deal seekers depresses the curve for months afterward, while a full-price prospecting push can lift it. This is why the monthly grouping matters; it turns the repurchase curve into a record of what your acquisition levers actually produced.

The Data Problems That Distort the Metric

Three data problems bend the cohort curve, and the first two are the familiar conversion losses already covered throughout this skill’s guides.

The first is missing first orders. Julian Juenemann of MeasureSchool has documented that the average Shopify store loses forty to sixty percent of purchase events to browser-side restrictions, with the biggest holes in iOS and Safari traffic. When a first order never reaches the ad platform, the cohort denominator is wrong twice: the customer exists in Shopify but not in the platform’s account of who was acquired, and the platform credits the second order as the first. Both defects shrink measured repeat rate and make the platform look worse than it is.

The second is identity failure on the second order. A repeat purchase only joins the right cohort when both orders link to the same person. If the first order arrived without enough identity for the platform to match, the second order may be credited as a new acquisition, or credited to no acquisition at all, and neither outcome lands in the cohort where it belongs. Failed identity is the same mechanism that drags down Match strength, and the Meta Business Engineering Team’s guidance applies here as elsewhere: send every identifier you have, email, phone, and browser ID, so each event can be matched to the person behind it.

The third is silent gaps in the customer record. If a checkout change stops capturing phone numbers for two weeks, or a theme update drops the browser ID, the orders from that period match worse than the orders around them and the cohort lines bend for reasons that have nothing to do with buyer quality. Meta’s published position is that event match quality directly influences delivery and optimization, so an identity gap does not just distort the reporting; it changes how the algorithm spends in the days that follow. A balanced read across both platforms, the Meta CAPI and Google Ads second channel setup covered separately, only stays honest when both streams carry complete data.

How Cohort Reporting Works on Top of Recovered Pulses

A cohort view is only trustworthy when it is computed from a purchase record where every order arrived once, carried identity, and was attributed to the platform that drove it. That record is what recovered, server-side Pulses build. Each order becomes a Pulse in the Stream, delivered to each connected Channel through Meta CAPI and the Google Ads integration, with a stable event ID so the browser and server copies of the same purchase count once.

On top of that record, the reporting layer computes repeat rate and churn per acquisition month and per platform, alongside return on ad spend by cohort rather than by first order alone. Rudy Sholtes, who manages growth for a DTC brand, frames the shift the way many operators experience it: the platform that looked cheaper on first orders was not the one that produced the customers who stayed, and the cohort view is what exposed the difference. The result is a budget conversation anchored in recruited-buyer lifetime rather than a single order.

Delivery health still gates the whole thing. A Clarity Score that holds high across both Channels, checked on the Overview, is what lets you trust that the repurchase lines reflect customers rather than data holes, and recurring research has found that keeping an existing customer costs close to five times less than acquiring a new one, which is why the quality of the first acquisition decides so much of the store’s economics later.

Step-by-Step: Reading Your First Platform Cohort Split

1. Confirm the record is complete. Check the Overview. Both Channels should show a Clarity Score above ninety percent, because a cohort built on a leaking Stream inherits the leak.
2. Confirm identity is complete. Check Match strength on recent purchase Pulses. If it sits low, fix collection gaps at checkout before reading the curve, since unmatched second orders would otherwise deflate the rate.
3. Pull cohorts by acquisition month. Take the last six to nine months, group first orders by month, and compute the share of each group that ordered again at two, four, and six months.
4. Split each cohort by platform. Separate customers by the Channel that drove their first order so Meta and Google each get a line.
5. Read the gaps, not the averages. A persistent platform gap at fixed offsets is buyer quality, and it is evidence, not noise, when both groups bought under the same promotions and prices.
6. Match the movement to what you changed. Tie curve shifts to the offers and campaigns of each acquisition month. Discount-heavy months should show up as flat lines weeks later; healthy prospecting months should hold their curve.
7. Re-check after storefront changes. Rerun the cohort view after any checkout or theme change, because collection regressions reshape the curve faster than campaign work does.

FAQ

What is a repeat purchase rate cohort? A group of customers defined by the month of their first order, tracked over the following months to see how many ordered again. Splitting each monthly group by acquisition platform shows which one recruits buyers who return.

Should I compare Meta and Google on first-order ROAS or repeat rate? Use both, weighted differently. First-order ROAS describes the first sale; repeat rate by acquisition month describes the customer the platform produced. When they disagree, the repeat rate is usually the better predictor of true return because existing customers convert at far higher rates than new prospects.

Do discounts ruin the cohort curve? They depress it. Deal-driven customers repurchase at lower rates, so acquisition months dominated by discounts produce flat cohort lines weeks later. That is a real read, not a data problem, and it belongs in the comparison rather than behind a store average.

Does this metric require server-side tracking? Not literally, but the record behind it degrades fast without it. Browser-side losses and identity gaps distort the denominator, misattribute second orders, and make acquisition platforms look worse than they are. Recovered server-side delivery is what makes the cohort trustworthy.

Related Resources

• The Shopify Meta ads optimization guide covers how cohort quality and Match strength feed campaign decisions.
• Bringing Google Ads in as a second Channel for Meta CAPI explains why balanced coverage across both platforms is what makes the cohort split trustworthy.
• How clean signal lowers cost per purchase covers the delivery and identity mechanics behind every Pulse that feeds the cohort record.