About VASTA
Who is actually going to look at your store? Most agencies will not tell you, because the answer changes with whoever is available that month. Ours does not change.
Who does the work
VASTA is 29 people in two squads. One is build-heavy, for architecture and technical work. The other is full-spectrum, covering conversion, development, user experience and organic. A project manager and a QA lead sit inside each.
They are permanent and they work together, which matters for a reason that is easy to miss: the value of a testing programme is not the tests, it is what the team remembers from the ones that lost. A rotating contractor pool cannot hold that. Neither can a tool.
Igor Silva, founder
Igor left Brazil for the United States in 2008. A winter in Avon, Colorado, working as a ski valet and bellman. Four years in Los Angeles driving deliveries. Then a medical marketing group hired him as a developer, which he was not: he is a marketer by training who had taught himself to build.
He built that agency's development team in Vitória da Conquista, his home town, and grew it to about ten engineers. Three years in he was VP of Marketing.
Vasta started in 2015, alongside that job. Selling on Amazon, running dropshipping, a few clients on the side. By his sixth year at the agency there was nowhere left to grow, so he left and took the Conquista team with him, into Vasta. His old employer became its first outsourced development client. An ecommerce community invited him to run their technical calls around the same time, and the Shopify work started arriving from there.
Conversion optimization became the centre of it, and has been ever since.
He still runs the first call on every engagement personally.
Why one team, in one place
The team is in Vitória da Conquista, Bahia. Igor was born and raised here, spent fifteen years in California, and came back. The company is registered in Long Beach, California, but the people who do the work sit together in one building.
That is a deliberate structural choice, not a cost decision. The three complaints operators make about agencies all have the same root: departments that do not talk to each other, delivery that is slow and expensive, and no institutional memory of what was already tried. A single permanent team in one location is the structural answer to all three.
The agreement
Those four lines cost us work. The second one is the expensive one: it means we cannot use a number until it has a proof pack behind it, and it is why some of our results are attributed by name and others are described only by what changed. If you want to see how far that goes, the measurement methodology is published in full.
The evidence
Every result below names the change we made, the sample it was measured across, and the window it was measured over. Where we do not have a published result yet, we say that instead of filling the gap.
| Measure | Value |
|---|---|
| Conversion rate | +20% |
| Revenue per visitor | +3% |
| Window | 8 days |
| Sessions | 865,909 |
| Confidence | 90% |
Largest session sample in our test history. Conversion rate was the primary metric. Revenue per visitor rose 3%, which did not clear the decision threshold on its own.
| Measure | Value |
|---|---|
| Projected monthly contribution | $175K |
| Window | 5 days |
| Sessions | 73,873 |
| Confidence | 90% |
Projected from a 5-day window. Short-window directional projection, not yet validated across a complete trading cycle.
| Measure | Value |
|---|---|
| Projected monthly contribution | $71K |
| Window | 30 days |
| Sessions | 71,833 |
| Confidence | 90% |
Our longest test window. When the window matters more than the number, this is the one to read.
Projected monthly contribution is extrapolated from the test window. It is not revenue banked, and we will not present it as though it were.
Results are specific to the cited experiments. Not every test wins, and projected contribution depends on the measurement window and store economics.
Six of the nine leaks VASTA inspects do not yet have a published result behind them. Those are described as process, not proof, until a proof pack exists. How we measure