Campaign rebuild
quality over volume- Every headline added a qualifier the sales team cared about.
- Landing pages filtered weak clicks before forms were submitted.
- Cost per good lead replaced cost per click.
Cost per lead is the wrong number to optimise. A fintech client we worked with was spending ₹4.2L per month on Google Ads, generating a healthy-looking 90+ leads per month, and still struggling to fill the sales team's calendar with real conversations. Inside 60 days we rebuilt the account around cost per good lead — and the same budget started producing more than 3x as many qualified calls.
The starting point
The client was a mid-stage fintech selling a working-capital product to SMB businesses in India. Average deal size ₹40,000–₹2,00,000 annually. Sales cycle two to six weeks. They had been running Google Ads for eighteen months, mostly on a "spend more, get more" assumption.
On paper the account looked fine: cost per lead was ₹4,600, well inside the industry benchmark their previous agency had quoted. But when the sales team logged outcomes back, the picture changed. Of every 100 leads, only 13 were the right size of business, only 6 booked a follow-up call, and only 1.4 turned into a paying customer. The true cost per good lead — the metric that actually maps to revenue — was around ₹70,000.
What we found in the audit
We ran the standard 7-day account audit. Three things stood out.
1. Buying-readiness was not segmented. Searches like "working capital loan for restaurant" (high intent, ready to apply) sat in the same campaign as "how does invoice financing work" (research stage, weeks away). The same bid logic was running on both. The same landing page was catching both. Predictably, the research traffic converted on the form at a high rate and then ghosted the sales team.
2. Ad copy was inviting the wrong clicks. Headlines read "Loans up to ₹5 Cr · Apply in 5 minutes · No collateral." Nothing in the copy filtered for company size, ticket value, or business type. Anyone who needed any kind of loan clicked.
3. Sales feedback was not flowing back to the platform. Google Ads was optimising for "lead form submitted" because that was the only conversion event configured. The platform had no way of knowing which of those leads were actually worth anything. So it kept finding more of the cheap, low-quality ones.
What we changed
The rebuild was a three-week sprint, structured around three principles: split by intent, filter on click, learn from sales.
Tiered campaigns by buying readiness
We split the account into three tiers.
- Tier 1 — Ready now (60% of budget): Buyer-ready queries. Searches with a clear product noun and an intent signal: "working capital loan apply", "invoice discounting SMB", "business loan ₹50 lakh". Bid aggressively, landing page pushes straight to a qualified application.
- Tier 2 — Still looking (35% of budget): Evaluation-stage queries. Comparison terms, alternatives, "vs" searches. Bid moderately, landing page is a comparison guide with a soft CTA to download a buyer's checklist. Lead capture, but tagged for nurture rather than direct sales handoff.
- Tier 3 — Just learning (5% of budget): Top-of-funnel research terms. Bid low, strict daily caps, landing page is a content article. Used as a learning signal for what's growing in interest, not as a primary lead source.
Qualifier copy that tells the wrong people not to click
Every Tier 1 ad got a qualifier baked into the headline or description. The phrasing varied by ad group, but the principle was consistent: name the customer we wanted, in language they would recognise, in a way that made non-customers self-deselect.
- Pricing context: "From ₹50,000 to ₹2 Cr · For ₹3 Cr+ annual turnover"
- Company size: "Built for 10–200 employee businesses"
- Outcome language: "Funds in 72 hours, not 3 weeks" (signals to fast-moving operators, signals away from price shoppers)
The click-through rate dropped. The conversion rate per click went up. Good calls per ₹1L of spend went up faster than either.
Offline conversion import to teach the platform
This was the change that moved the numbers the most, and it is the one most teams skip because it requires CRM cooperation. We wired Google Ads's offline conversion import to two new sales events: "qualified call held" (a real conversation with the right kind of business) and "opportunity created" (sales-qualified, in pipeline). Every lead carried its GCLID through the CRM, and the sales status was synced back to Google nightly.
The effect compounds. After three weeks of feedback, Google's smart bidding stopped optimising for "any form fill" and started optimising for "form fills that look like ones that later become qualified calls." It found different audiences and different placements than it had been finding on its own.
What happened
The first two weeks looked worse on every dashboard the client was used to looking at. Lead volume dropped from ~90/month to ~38/month. Cost per lead climbed from ₹4,600 to ₹11,000. The CMO got nervous; we had agreed in advance that this was the expected dip.
By week six, the numbers that actually mattered had moved:
- Good calls per month: 6 → 19 (same total spend)
- Cost per good lead: ₹70,000 → ₹43,200 (a 38% reduction)
- Sales-qualified opportunities per month: 4 → 11
- Sales team time spent on disqualification calls: down roughly half, by the head of sales's count
Vanity volume metrics looked worse. Revenue-tied metrics looked considerably better. That trade-off is almost always the right one for a B2B fintech, and almost always the one a generalist agency will not make on the client's behalf.
What did not work
Two things to flag honestly:
Tier 3 (top-of-funnel) was harder to justify than expected. The plan was to use Tier 3 as a brand and intent-learning layer. In practice, attribution back to revenue was murky enough that we cut the Tier 3 budget by half in month three. Top-of-funnel paid search likely works better for higher-ticket products with longer sales cycles than this client had.
The qualifier copy occasionally over-filtered. One ad group ran headlines so specific to manufacturing SMBs that we lost qualified services businesses for a fortnight before we caught it. Lesson: monitor industry-vertical share of leads weekly during the first 30 days, not just cost and volume.
What to take from this if you are running Google Ads
The principles travel even if the product does not.
- Compute cost per good lead, not cost per lead. Define "good" with the sales team — usually a combination of company size, role, and a real conversation having happened.
- Split campaigns by buying readiness, not by keyword theme. Two queries about the same product can have wildly different intent. Bid logic and landing pages should reflect that.
- Put qualifiers in the ad copy. The right click is more valuable than the cheap click. A well-placed price floor or company-size qualifier saves more in wasted sales time than any bidding optimisation.
- Wire offline conversion import. If Google Ads only sees form fills, it will keep finding people who fill forms. Teach it what actually matters.
- Expect the dashboards to get worse before they get better. Agree the dip with the executive team in advance, in writing, with the metric you are trading up to.
About this case
This was a 60-day engagement scoped as a paid-search rebuild. flowX worked directly with the client's growth lead and head of sales; budget and offer were unchanged from the prior agency. Numbers shown above are real for this engagement. Outcomes vary by product, market, sales motion, and execution — we share specifics so you can judge applicability, not as a promise of similar results.
Real work. Results vary by product, market, and execution.