A Series A B2B SaaS company in India
Landing pages and conversion
2.4x the qualified demos on the same budget
2.4x
qualified demos a month
From 41 to 97 qualified demos a month, between Q1 and Q3 2025, the company's CRM
Q1 2025
41
qualified demos a month
Q3 2025
97
qualified demos a month
They were spending more each quarter and booking the same number of demos. We stopped buying traffic for a month, fixed what happened after the click, and rebuilt how a demo was qualified. The spend didn't change. The number of real conversations did.
The problem
Paid spend had grown for three quarters and demo volume was flat. Sales said half the demos weren't worth taking. Nobody could say which campaigns produced the good ones, because the CRM recorded a lead source and nothing after it.
That last sentence is the one that mattered. A lead source field tells you where a form fill came from, not what happened to it, so every conversation about the budget was a conversation about opinions. Marketing could prove the campaigns were producing demos. Sales could prove the demos weren't worth taking. Both were right, and the two facts lived in different reports.
There was also no agreed meaning for the word qualified. Sales used it to mean a company of the right size with a problem they could solve and a person who could buy. Marketing used it to mean a demo that got booked. Those two definitions were 56 demos apart in a month, and the gap was being paid for out of the ad budget every week.
What we did
Three things, in order, and the order was the whole method.
We bought no new traffic for the first month. No new campaigns, no budget increases, nothing added. The existing campaigns kept running at the same spend while we worked on what happened after the click. This was the hardest part to get agreement on, because the instinct when demo volume is flat is to add a channel.
We rebuilt the demo page around one question. The old page was a product tour: six feature blocks, a video, logos, and a form at the bottom. The buyer's actual question, which we got from eleven recorded sales calls, was whether the product would work with the two systems they already ran. So that became the page. The headline named the two systems. The proof sat directly under it. The form moved above the fold. Everything that answered a question nobody was asking came off. The method is the one in the landing page teardown checklist.
We let sales define qualified, then put the definition in the form. The SDR team wrote the rules: company size, the systems in use, and whether the person had a budget line or sat next to someone who did. Two of those became fields on the form. The third became the first question on the call. Adding fields to a form reduces submissions, and it did. Submissions fell and qualified demos rose, which is the trade that was worth making and the one that needs explaining in advance, before the weekly numbers show a drop.
Then we rebuilt the reporting, so every demo traced back to the campaign and the page that produced it, and so that trace survived into the stage the deal reached. That is what let us cut three campaigns. They had been filling the calendar for months with companies that could never have bought, and until the report existed there was no argument that could have ended them. The reasoning behind the metric is in cost per qualified lead.
Who did what
I led the work, wrote the page and the ad copy, and built the reporting. The company's designer built the page. Their SDR team set the qualifying rules, and their RevOps lead wired the CRM fields. The decision to hold spend flat for a month was the founder's, and it was the decision that made everything else possible.
What didn’t work
Two things, and the second one cost more than the first.
The chat widget. Before any of the above, we put a chat widget on the pricing page, because it took a day and the case was obvious: people are on the pricing page, they have questions, let them ask. It produced far more conversations and almost none of them were buyers. They were existing customers with support questions, students, and people comparing tools for a report. Qualified demos did not move. We turned it off after three weeks.
What I got wrong was the stage. A chat widget harvests intent that already exists, and the people on that page who wanted a demo were already clicking the button. The problem was never that they couldn't ask. It was that the page they landed on first had answered the wrong question.
The ad creative test. This is the one I would warn anyone off. We ran four creatives for two weeks and judged them on qualified demos. The test never reached significance, and it was never going to, because the budget per variant was too small for the number we had chosen to measure.
The arithmetic takes two minutes and we did not do it first. The standard approximation for sample size at 80 percent power and 95 percent confidence is 16 times p times 1 minus p, divided by the square of the absolute lift you want to detect. At a page-to-demo rate of 2.1 percent, looking for a 20 percent relative improvement, that is 16 times 0.021 times 0.979, divided by 0.0042 squared, which is about 18,000 visits per variant. The page was receiving under 4,000 visits a month in total. Split four ways, each creative was getting under 1,000 a month. We needed about 18 months and we had allocated two weeks.
The fix is not a bigger budget. It is to judge creative on something that happens often enough to measure, like the cost of a landing page visit, then validate only the winner on demos, which is what the creative testing loop is for. We lost two weeks of a quarter learning that, and the only reason it wasn't worse is that nothing was decided on the basis of a result that didn't exist.
The other numbers
Cost per qualified demo
$1,840 to $1,140, down 38 percent
Landing page to demo
2.1 percent to 5.3 percent
CAC payback
14 months to 9 months
From the library
Stage of the funnel
Questions
How long before the numbers moved?
Six weeks for the page, eleven for the reporting. The page was faster because a page can ship in a fortnight and start accumulating data immediately. The reporting took longer because it needed the CRM fields, the SDR team's habits and a month of clean data before anyone would trust it enough to cut a campaign on it.
Did the budget go up?
What would you do first in a similar company?