Every exporter running Google Search Ads eventually says some version of the same sentence: we spent four thousand dollars last month, got twenty inquiries, half of them recruiters and competitors, so Google Ads doesn't work in our industry.
Sometimes that's true. Usually it isn't. What actually happened is that the account was built in a way that quietly hands your budget to people who will never buy, and structural problems never throw an error. They just spend the money.
Know your cost per inquiry before you optimise anything
Optimisation without a benchmark is just expensive guessing. Here are the ranges we see across export categories, English-language search, B2B:
| Industry | Average CPC | Landing page conversion | Cost per inquiry |
|---|---|---|---|
| Industrial machinery and equipment | $1.20–$3.50 | 1.5%–3% | $60–$180 |
| Chemicals and raw materials | $0.90–$2.40 | 2%–4% | $35–$110 |
| Consumer electronics and accessories | $0.60–$1.80 | 1.2%–2.5% | $40–$120 |
| Packaging and printing | $0.80–$2.00 | 2.5%–5% | $25–$80 |
| Building materials and hardware | $1.00–$2.80 | 1.8%–3.5% | $45–$130 |
Read the table before you read your invoice. A $150 inquiry is respectable for a machinery exporter quoting six-figure orders. The same $150 is a disaster for a packaging supplier whose average order is $3,000. The number only means something against your order value and your close rate — which means the first question isn't "how do we cut spend," it's "which of the three stages is broken: the click, the conversion, or the qualification."
Mistake one: counting form fills as conversions
Google's Smart Bidding learns exclusively from the conversion data you feed it. If every form submission counts as a conversion, and 40% of those submissions are price shoppers, competitors doing research and students collecting spec sheets, the system will work extremely hard to find you more of exactly that. Six months later you conclude the traffic has gone bad. It hasn't. You trained it.
Fix it in three layers, cheapest first:
- Qualify at the form. Make two of these required: company website, annual purchase volume, target market. Only a completed, qualified form counts. This alone removes a meaningful share of noise.
- Feed back the inquiries sales actually accepts. Push the "qualified" status from your CRM into Google — offline conversion imports, or Enhanced Conversions for Leads. Bidding only starts optimising toward buyers after this is live.
- Use mid-funnel conversions when the cycle is long. Sample requests, video factory audits, RFQ downloads. They give the algorithm a positive signal months before the first order lands.
[object Object]
Mistake two: one campaign per country
Smart Bidding needs roughly 15–20 conversions per campaign per week to leave the learning phase. Plenty of export accounts are split into thirty campaigns — country times product — each with a $50 daily budget. At that volume a campaign accumulates maybe five conversions a month. It never stops learning, never stabilises, and you read the volatility as "Google is unreliable."
A structure that actually functions:
- Split the top level by product line. Three to five campaigns for the things you genuinely want to sell, not a mirror of every SKU on the website.
- Split the second level by language region, not country. English-speaking markets, Spanish-speaking markets, the Gulf. Countries go inside the campaign as geo targeting.
- Set each campaign's daily budget at no less than three times its target cost per inquiry. That is the exploration budget the algorithm needs. If you can't fund it, consolidate campaigns instead of spreading thin.
- Separate Search from Display and never leave "search partners" and "display expansion" on by default. That traffic behaves nothing like search traffic and it will contaminate your conversion data.
Mistake three: broad match with no negative keyword discipline
Broad match plus Smart Bidding is Google's default recommendation. It works well for consumer e-commerce. For B2B export it is often a slow bleed. An industrial pump advertiser's search terms report will contain "pump price in pakistan," "how to repair water pump," and "pump job vacancy dubai" — three completely different people, none of them buying.
Thirty minutes a week, every week:
- Pull the search terms report, sort by cost, read the top fifty. For each one ask: would the person who typed this place an order?
- Add negatives for job, salary, repair, DIY, PDF, free, second hand, near me, Amazon, and the names of marketplaces.
- Exclude geographies you don't serve. If you sell into Europe, an Indian click isn't a bargain at a lower bid — it's a click you paid for.
- Pull the long-tail terms that did produce inquiries into a separate exact-match campaign with a higher bid and its own budget.
One caveat on copying negative lists: "distributor," "supplier" and "wholesale" are precision terms for a manufacturer selling to resellers and pure waste for one selling direct. Build your own list against your own sales model.
The eight seconds after the click
[object Object]
What that means in practice:
- One ad group, one landing page. "Stainless steel butterfly valve manufacturer" should land on the butterfly valve page, not the homepage. Homepage traffic from specific queries typically converts at a third of the rate.
- Specs above the fold, adjectives below it. Size range, pressure rating, material, certifications — CE, API, ISO. A procurement engineer is scanning for those.
- Four form fields. Name, email, country, requirement. Going from four fields to seven roughly halves completion.
- A persistent WhatsApp button. Buyers in the Gulf, Latin America and Southeast Asia will message before they'll fill in a form.
Three things to do this week
- Pull ninety days of data. Calculate real cost per inquiry and the share of inquiries sales considers valid. Compare both against the table above and identify which stage is broken.
- Audit conversion tracking. If a form fill is your only conversion, and qualified status never returns to Google, everything else you fix is a half-measure.
- Clean the search terms report and add negatives. Most accounts cut 15–30% of wasted spend immediately — faster than any copywriting change.
There's a quieter leak underneath all of this. Inquiries arrive through the website form, WhatsApp, Facebook, Instagram, the site chat widget and email, and each one lands in a different inbox. Nobody can say which ad produced which conversation, so attribution defaults to guesswork and "ads don't work" becomes the easiest conclusion available.
This is the part we care about most. Inquirly is a Google Premier Partner and a qualified Facebook partner in China, running paid search and social alongside the websites and the CRM, so the account structure, conversion feedback, landing pages and follow-up sit with one team instead of three vendors blaming each other. Fulongma Group cut acquisition cost tenfold in 100 days. Another client passed ten million RMB in revenue within two months while marketing cost fell to one sixth. Shandong Xinhai Mining grew revenue tenfold over five years with more than 95% of it overseas. IGET Vape automated WhatsApp lead capture and lifted conversion by 125% — same ad budget, better plumbing underneath it.
Next step
Send me your industry, your main export markets, and last month's ad spend and inquiry count. I'll benchmark your real cost per inquiry against the ranges above and tell you the one structural change to make first. No charge. Contact Ji Ping at +86 159 0147 6874, also on WhatsApp and WeChat. Start your message with "Google Ads" and I'll know what you need.
