How a leading online auto parts retailer multiplied its qualified leads by 2.7.

01 · What they thought they knew

"There are about 25,000 repair shops out there, that is our market. We know them, we have been in this business a long time." Reworded from a typical brief, anonymized

That figure of 25,000 is not made up. It comes from the reference industry studies, the panels that the whole trade uses as a benchmark.

The problem is not that the number is wrong. It is that it describes a research panel, built to measure market trends, and not a list of reachable prospects. A company that uses it as its market size gives itself a target it can neither call, nor verify, nor grow.

02 · What the business registry says

The other reflex is to go to the official registry. Filtering on the activity code for the maintenance and repair of light motor vehicles returns on the order of 60,000 businesses. That figure is more than double the previous one, and it looks more solid because it comes from a legal registry.

It is just as misleading, in the other direction. Those 60,000 rows contain, all mixed together:

• Closed businesses, still on the registry because the deregistration never followed.
• Heavy truck shops and agricultural workshops, whose parts needs have nothing in common.
• Businesses that are not repair shops at all but share the activity code.
• Chains and owned networks, whose purchasing is centralized at head office: calling them location by location achieves nothing.
• Sole traders with no premises, unreachable for a field rep.

An activity code says what a company declared it would do when it registered. It says nothing about what it does today, with what equipment, or whether it decides its own purchasing.

03 · The real scope: 15,000

After reading business by business, around 15,000 repair shops are left that meet all four conditions at once: passenger cars, open, independent of any network with centralized purchasing, and equipped with a real mechanical bay.

That is the first gain, and it is purely commercial: three quarters of the file disappears before the first call. A team that was working 60,000 rows works 15,000, with the certainty that each one is reachable and makes its own decisions.

04 · The market registered somewhere else

Narrowing the scope was only half the work. A parts distributor does not only sell to general repair shops: it also sells to tire shops, body shops, auto glass specialists, roadside assistance workshops, and small-town franchised agents, the shops carrying a manufacturer's sign but run entirely independently.

All of those businesses are properly registered. They are simply not under the repair shop activity code, but under the codes of their own trade. An export of that one code, however complete, will never surface them, because they are not in it.

By widening the reading of the registry to those adjacent codes, we identified 47,000 additional businesses. They then went through exactly the same screening as the first 60,000, and the result is the same: a quarter of them are inside the buying scope, about 12,000 businesses, each labeled by trade so the filtering stays in their hands.

05 · Rebuilding the number nobody had

Put end to end, the two scopes give the genuinely addressable total.

06 · Where the ×2.7 comes from

Before any commitment, a sample was delivered to their reps with one simple instruction: match it against your database, row by row, by phone.

The result is the ratio that gives this page its title. For every row they already had, they received 2.7 usable ones. The gap is not a projection: it is a count done by their own teams, on their own database, with their own matching key.

Six rows in ten were completely unknown to them. At a long-established company, certain it knew every player in its sector.

The remaining rows were not wasted either. A database of repair shops ages fast: relocations, number changes, takeovers, rebrands, closures. Those came back corrected, which was not the ask and turned out to be the second gain.

07 · The four effects, in the order they happen

• Knowing your real market size. 27,000 addressable businesses, neither the study's theoretical 25,000 nor the registry's 60,000. A number that is built rather than inherited, and checkable by phone.
• Multiplying usable leads by 2.7. Matching against their database measured it row by row, with no extrapolation.
• Halving prospecting time. Three quarters of the registry ruled out before the first call, and every remaining row arrives with its opening line already written.
• Bringing the existing database up to date. A side effect, never asked for, observed every time.

08 · The reading grid

Every business is described by 48 columns with a closed vocabulary. No free text, no interpretation: values picked from a list, and for each one a justification citing its source. Here are the axes that actually decide a parts sale.

09 · Every business, placed against its market

Knowing a repair shop exists does not tell you whether it is busy. So we count the public reviews received over the last six months and compare every business against the whole automotive market measurable over the same window, 29,506 businesses.

• Very high activity, top 10% of the market: 27 recent reviews or more.
• High, top 25%: 8 reviews or more.
• Average: 2 reviews, the market median.
• Low: below the median. No recent reviews: zero over the period.

A rep no longer prioritizes at random: they know who is in the top decile of their sector.

10 · What we admit we do not know

A data vendor that states its error margins before being asked is a vendor you can believe on everything else.

• Accuracy measured at around 90% on a sample checked by independent reviews. 25 records were individually corrected following those reviews, and carry the note in their justification.
• Closures: a stated deduction of 4 to 8%. Some recent closures are not yet recorded anywhere in public sources.
• About 13% of records do not allow manufacturer approval to be settled from public sources alone. Those cases stay undecided and are flagged as such.
• Listing duplicates are merged. The same shop can appear several times depending on the source. We detect them and state volumes in unique businesses, never in raw rows.
• Scope limited to the mainland market surveyed. Outlying territories need a dedicated coverage effort.
• Never a false zero. When an indicator cannot be measured, the cell stays empty rather than showing a zero that would read as information.

Coverage does not rest on a single pass: the territory is swept by two independent methods, one systematic mapping sweep and one locality-by-locality search, then cross-checked against the official business registry.

11 · Matching against your CRM

The recommended matching key is the phone number, not the name, because a legal name gets rewritten and a number can be compared. Deduplication is done against your database: you receive the net delta, not a file that overlaps what you already have. That is exactly what made it possible to separate genuinely new accounts from records needing an update, and therefore to measure the ×2.7 rather than assert it.

On top of that comes the legal identification where available: registration number, owner name from the registries, years in business, headcount, and revenue where it is published.