Comparison
Google Analytics tells you what happened. Wardly tells you what it was worth.
Both are honest tools doing different jobs, and most of our clients run both. Here is the specific line between them — including the case where you should stay exactly where you are.
First, the fair part
What Google Analytics does better.
Any comparison page that cannot answer this honestly is not worth reading.
- It is free. For an enormous number of businesses that is the correct trade, and free is a real feature rather than a compromise to argue away.
- The Google integration is genuinely good. Native connections to Google Ads and Search Console, and a free BigQuery export. If your ad buying depends on conversions flowing back into Google Ads, that is hard to give up and you should not.
- It measures far more than we do. Wide event coverage, app and web in one property, and years of production maturity across millions of properties.
- Everyone knows it. Skills are portable, the documentation is vast, and any question has already been answered somewhere.
- Its prediction is good — when you qualify. The models are sound. The issue is who is allowed to use them.
The threshold
Prediction is gated behind a number most sites never reach.
Google Analytics will produce a purchase probability only once your property has recorded at least 1,000 returning users who triggered the event and at least 1,000 who did not, within the last 28 days.
≈ 33,000–50,000
returning sessions every 28 days, at a typical 2–3% ecommerce conversion rate
And if the model quality for a property falls below Google's minimum threshold, the predictions stop updating and can disappear from reporting.
This is not a criticism of the models. A universal, free tool has to set a bar where its predictions will be reliable for everyone. But the effect is that the businesses most in need of help judging traffic — the ones without enough conversions to count — are exactly the ones excluded.
Even when you qualify, it is a different answer
Suppose you clear the threshold. What you get is a probability that a user active in the last 28 days will convert in the next seven. That is built to assemble a remarketing audience, and it is good at that.
It is not built to answer “was the traffic I bought last month any good?” — a retrospective question about sessions that have already happened, asked per channel. Those are different shapes of question and the second one is the one that changes a budget.
And it only predicts two things
The predictive metrics cover purchase and churn. If your commercial outcome is a booked demo, a qualified enquiry, a quote request or a subscription start, there is no predictive metric for it.
Wardly trains against whichever single event your business actually runs on, because that choice is made per deployment rather than baked into a universal product.
Side by side
Dimension by dimension.
| Dimension | Google Analytics | Wardly |
|---|---|---|
| What it measures | What happened: sessions, users, engagement, events, conversions — across a very wide surface. | What the traffic was worth: an intent score per session, and a quality grade per channel. |
| Prediction, and who gets it | Purchase and churn probability, but only once the property has 1,000 returning users who triggered the event and 1,000 who did not, within 28 days. | Intent scoring from about 50 conversions, not 1,000. Built for the volumes that make counting unreliable. |
| Unit of prediction | A user, forward-looking, over the next 7 days. Designed to build remarketing audiences. | A session, including sessions already past. Designed to grade traffic you have already paid for. |
| Which event | Purchase or churn specifically. | Whatever single event your business runs on — a demo, a qualified enquiry, a subscription, a purchase. |
| Collection | A tag served from a Google domain, so collection is a third-party request that can be blocked by domain. | Runs under your own domain. No third-party request in the measurement path. |
| Who configures it | You, or an agency. The defaults measure traffic until somebody changes them. | We do, as the engagement, including defining the conversion event with you. |
| Cost | Free at the standard tier. Analytics 360 is enterprise-priced. | A paid engagement, scoped per deployment and quoted after a call. No free tier. |
| Ads integration | Native, deep integration with Google Ads, Search Console and a free BigQuery export. A genuine strength. | None. If your buying depends on conversions feeding Google Ads, you keep Google Analytics for that. |
Stay on Google Analytics if
- Your Google Ads buying depends on conversions feeding optimisation, and you will not run a second measurement layer alongside it.
- You already clear the prediction threshold comfortably and only want remarketing audiences.
- You need app and web in one property — Wardly measures websites.
- Budget for measurement is genuinely zero. Free is a real feature.
- You need a tool every agency and contractor already knows.
Questions
Common questions.
Does Google Analytics really not predict conversions?
It does, and we want to be precise about it rather than score a cheap point. Google Analytics produces purchase probability and churn probability. The constraint is eligibility: the property needs at least 1,000 returning users who triggered the event and at least 1,000 who did not, measured within the last 28 days. At typical ecommerce conversion rates that is roughly 33,000 to 50,000 returning sessions a month. Below that the metrics simply are not available, and if model quality drops the predictions stop updating.
So this is only for small sites?
No — it is for any site where conversions are sparse relative to traffic, which includes plenty of large ones. A high-traffic business with a considered, low-frequency purchase can have millions of sessions and nowhere near 1,000 monthly converters. The binding constraint is conversion count, not traffic.
Is Wardly a replacement for Google Analytics?
Usually not, and we would rather you kept both at first. Google Analytics is free, universal, and holds the ads integration you may depend on. It answers what happened across a wide surface. Wardly answers whether the traffic was worth buying. Running both also gives you a reference point while you build confidence in a new measurement layer.
Will the numbers match?
No, and you should expect that. Different collection architectures, session definitions and consent handling produce different totals — Wardly will typically record sessions that a third-party tag loses to domain-level blocking. During deployment we reconcile against whatever source you already trust, so the difference has an explanation attached before anyone questions it in a meeting.
Couldn't we just configure Google Analytics properly instead?
For a lot of businesses, yes — and if that is what you need, hire someone to do it, because it is cheaper than this. A well-configured property with properly defined conversions and clean channel groupings is a large improvement over a default one. What it will not do is score individual sessions on intent, or grade a channel before enough conversions have accumulated to count.
Is Google Analytics' engagement rate not the same idea?
No. Engagement rate measures whether a session met an activity threshold — time, pageviews, or a conversion. It has no knowledge of what your business sells, so it cannot distinguish someone reading your careers page from someone comparing your pricing. Intent scoring is trained against your conversion event specifically.
Worth a call if you never qualified for prediction.
Thirty minutes. If configuring Google Analytics properly is the better answer for you, we will say so.