Buying Leads for an Accounting Firm? Measure Signed Mandates, Not CPL

Accounting Firm Lead Generation: Run a Cohort Pilot Before You Buy More Leads

If you run an accounting practice and are considering purchased leads, a directory, a referral platform, paid search, or a specialist agency, do not decide from the quoted cost per lead. Decide from a traceable cohort that ends in an accepted mandate, includes the time your staff spent qualifying and correcting records, and separates sourced revenue from revenue that merely happened after a marketing touch.

The practical test is a capped, time-bounded pilot with four controls: define what a qualified opportunity means before the first record arrives; retain the source and collection context for each record; record every stage and the work it consumes; and compare mature cohorts on accepted mandates and contribution, not on raw contacts or clicks. If a vendor cannot export those fields or explain where a contact came from, the offer is not measurable enough to scale. If a record fails the professional-fit, data-provenance, or contact-policy gate, do not send it to an associate merely to improve a funnel statistic.

This is an operating protocol, not a promise that one channel will outperform another. The worked figures below are synthetic examples to show the calculation. They are not results from a live campaign, a benchmark, or advice about a particular firm's legal obligations.

What this adds to the existing advice

The source page behind this article explains how an accounting firm can define an ideal client profile, treat public events as hypotheses rather than purchase intent, keep a contact-level compliance record, and distinguish fit, context, and compliance. It also recommends measuring accepted accounts, useful replies, held meetings, opportunities, and loss reasons. Those are useful foundations, but they do not by themselves answer the buyer's next question: when two lead sources produce different records, staff workload, sales cycles, and mandate values, which one deserves the next euro and the next hour?

The reusable asset here is a mandate-realization cohort ledger: a minimum data schema, a set of stage definitions, a cost formula, and a pilot decision table. It forces the comparison to stay within the same service, lead age, and observation window. It also records the reasons a lead was rejected, so that a low cost per contact cannot hide unsuitable work or staff time spent repairing a vendor's data.

The search context is narrow. In the 14–21 October 2026 Search Console window, the source site's query “generation de leads expert comptable” recorded 9 impressions at an average position of 13.1. That is a small observed query sample for topic selection, not a ranking promise, evidence of demand volume, or a conversion result. The primary page is the French guide to lead generation for accounting firms. Its visible approach covers targeting and compliance; this article concentrates on economic measurement after a lead enters a firm's process.

A record, a response, and a mandate are different outcomes

“Lead” is often used for at least four things: a row in a file, a person who made an enquiry, an account a partner agrees is worth pursuing, or a prospective client with a defined need and a plausible next step. A price quote attached to the first definition cannot be compared fairly with a fee attached to the last.

Create stage labels that match decisions your firm actually makes. A workable chain is: received; identity and source verified; eligible for the firm's service; accepted by a responsible partner; contacted lawfully and appropriately; substantive reply; meeting booked; meeting held; opportunity accepted; engagement letter issued; mandate signed; service started; first fees collected; and retained or closed. Keep “not eligible,” “duplicate,” “unreachable,” “no response,” “not now,” “out of scope,” “capacity unavailable,” and “opposed” as separate outcomes. These labels explain what failed. “Lost” alone does not.

For each stage, name the person who owns the transition and the evidence needed to move forward. A sales assistant may verify a company number and assign a source. A partner may decide whether the work fits the cabinet's expertise and capacity. A signed engagement letter is a stronger proof of acquisition than an appointment booked, while the first collected invoice is a stronger proof of realized value than the letter alone. Do not merge these outcomes into a single conversion rate.

Why cost per lead can select the wrong channel

Suppose a platform charges for each record and another source produces fewer enquiries but requires more partner time. The first can appear cheaper when the denominator is “records delivered” and more expensive when the denominator is “held meetings,” “signed mandates,” or “net contribution after delivery effort.” The answer depends on the firm's definitions, not on a universal benchmark.

Include all costs that change with the channel or cohort: vendor fee, media spend, data or enrichment costs, setup charges allocated across the pilot, contractor fees, staff time spent cleaning records, qualification time, follow-up time, meeting time before an engagement is signed, and avoidable rework caused by inaccurate information. Decide in advance whether fixed salaries and overhead are included. There is no one universally correct allocation; there is a correct requirement to state the allocation and use the same rule for each compared source.

Use a small, explicit metric set:

Metric	Calculation	Decision it informs	What it can conceal
Verified eligible-contact cost	Total pilot cost ÷ unique records that pass the agreed eligibility gate	Whether the delivered list contains usable accounts	Difference in account value or maturity
Cost per held meeting	Total pilot cost ÷ meetings actually held	Whether the source creates a real conversation	Partner time and meeting quality
Cost per signed mandate	Total pilot cost ÷ signed mandates in the cohort	Acquisition cost at the contract stage	Fee level, delivery margin, and long-cycle cases
Follow-up load per mandate	Total qualification and follow-up hours ÷ signed mandates	Whether the channel fits team capacity	Opportunity cost of partner attention
Contribution at a fixed horizon	Fees collected by the horizon − direct delivery cost − acquisition cost	Whether the acquired work has paid back under stated assumptions	Future retained fees outside the horizon
Source-to-mandate rate	Signed mandates ÷ unique source records in the cohort	A funnel description for that source and period	Attribution, small samples, and cohort maturity

State the numerator, denominator, inclusion rules, and date range next to every chart. For example, exclude duplicate records from the count of unique leads, but retain their cleanup time in total cost if that time was actually incurred. Otherwise the metric rewards a supplier for duplicate delivery.

A synthetic example: how a low quote can become an expensive mandate

The following arithmetic is deliberately fictional. It uses a single illustrative month and 12 fictional records solely to demonstrate the workbook. No firm, provider, or campaign produced these results, and the figures must not be reused as a market rate or forecast.

Assume a firm pays €3,000 for a small lead batch. Its employees spend 7 hours validating, routing, and following up; for the illustration only, the firm values this time at €60 per hour. Total measured acquisition cost is therefore €3,000 + (7 × €60) = €3,420. Of 12 fictional records, 4 pass the pre-agreed service-fit and contact gate, 2 meetings are held, and 1 mandate is signed.

The resulting synthetic ratios are €855 per eligible record (€3,420 ÷ 4), €1,710 per held meeting (€3,420 ÷ 2), and €3,420 per signed mandate (€3,420 ÷ 1). These are not observed performance claims. They show why quoting a €250 “lead price” would omit labor and would describe a different denominator.

Now suppose, still only as a fictional illustration, that the signed mandate produces €6,000 of fees collected within the chosen review horizon and €3,200 of direct delivery cost. Its contribution before acquisition is €2,800. After the €3,420 acquisition cost, the cohort's realized contribution is negative €620 at that horizon. A later retained engagement might change lifetime economics, but it must be measured later rather than silently included today.

The example does not mean a one-mandate cohort should automatically be rejected. The small fictional denominator cannot establish a stable conversion rate. It means the firm must either extend observation, test a comparable cohort, or decide that the economics are not yet established. A partner should not convert one signed case into a universal prediction.

The mandate-realization ledger

Put one row per unique prospective account in the ledger and store stage changes as dated events rather than overwriting the previous state. A compact spreadsheet is sufficient for a pilot if access is controlled and changes are auditable. A CRM is better when it can preserve the same fields and export them without vendor mediation.

Field	Required value	Why it matters
cohort_id	Campaign, source, and intake-period identifier	Keeps the comparison reproducible
lead_id	Stable internal key, not a public identifier	Joins stage events without duplicating a person
source_name and source_url	Vendor, referral, inbound page, register, or campaign plus exact record origin	Separates “who supplied it” from “what was observed”
collected_at and verified_at	Timestamps with timezone	Exposes stale records and response lag
service_offer	Narrow service the firm is evaluating	Prevents unlike mandates being blended
fit_decision	Pass, fail, or review, with reason	Makes “qualified” auditable
signal_fact	Observable fact, its source, date, and confidence	Prevents an assumption from becoming a claim
contact_policy_check	Channel, information notice, exclusion check, and decision	Makes the permission gate visible to an authorized reviewer
stage_events	Event name, timestamp, owner, and supporting evidence	Preserves the path through the funnel
minutes_by_role	Partner, assistant, and specialist minutes by activity	Captures the scarce capacity consumed
direct_costs	Vendor, media, data, and variable operating costs	Provides the cost numerator
mandate_value	Fee schedule, signed value, invoiced and collected amounts	Keeps expected and realized value separate
delivery_cost	Direct delivery effort using the firm's declared rule	Prevents revenue alone from standing in for margin
outcome_reason	Signed, deferred, declined, unsuitable, no capacity, or other coded reason	Turns losses into actionable feedback
objection_state	Restricted suppression status available to the authorized contact team	Prevents a future import from reactivating an objection

Minimize personal data. The ledger is a measurement tool, not a reason to keep every field a supplier can provide. Keep access limited, set retention and deletion rules, and have the firm's privacy lead assess the applicable requirements for the source and channel. The relevant professional conduct rule in France is not a substitute for a complete privacy analysis. The European Data Protection Board's published guidance and resources include its legitimate-interest guidelines: the Board cautions that direct marketing is not automatically a legitimate interest in every case, and that necessity and the balancing of rights still matter. The exact channel, recipient, and applicable electronic-communications rules require a case-specific review; this article does not determine a lawful basis.

Freeze the cohort before looking at the result

Choose the cohort entry rule before launch: for example, every record received from a named source during a documented date interval, after duplicate normalization, with a specific service offer assigned. Do not remove a record later just because it did not reply. If a record is invalid, excluded, or a duplicate, record the reason and keep the work spent on it in the cost ledger.

Compare cohorts with the same service offer and similar age. A payroll outsourcing enquiry and an annual bookkeeping enquiry have different complexity and fee patterns. A lead received yesterday has had less time to mature than one received several months ago. If sources have different maturation windows, show the same “days since intake” cuts rather than treating the younger cohort as a final result.

Keep at least three views: source cohort, service segment, and lead age. If volume permits, add geography, firm size band, and existing relationship status. Do not create so many slices that every chart describes one record. When a cell is small, show its count, describe it as exploratory, and avoid ranking channels on that cell.

Record the attribution rule too. “Sourced by” should mean the first verifiable acquisition origin under a written rule. “Influenced by” can include later interactions but should not replace the source field. If a referral, website visit, and paid search all touched an account, retain the sequence rather than assigning the entire mandate to whichever field the CRM happened to populate last. Multi-touch attribution is a model, not a fact about causality.

A capped pilot in six steps

1. Define a service and an exclusion boundary

Pick one service that the firm can deliver consistently. Specify the target legal form, size range or complexity, location, systems supported, expected onboarding effort, and cases that must be declined or referred. Get a partner to approve the definition. “Any company that needs accounting” is not a testable segment.

2. Write the outcome contract with the supplier

Before payment, request field definitions, sample records, collection dates, source descriptions, deduplication logic, contact or channel provenance, delivery format, replacement policy, and the exact event that triggers a charge. Ask what the supplier knows about exclusivity and whether the same account is sold to other firms. If the supplier cannot answer, record the unknown and lower the maximum exposure; do not use an assumed answer in the ROI model.

3. Set a loss cap and stop rules

Set a maximum spend and staff-hour cap that the partners can afford even if the pilot yields no mandates. The cap is a governance choice, not an industry benchmark. Stop delivery immediately if provenance is missing, a suppression request fails, duplicate levels exceed the firm's written tolerance, or the source's offer does not match the signed agreement. Never increase the cap because early records “feel promising.”

4. Instrument intake and follow-up

Create the cohort key and stage-event fields before the first record arrives. Assign an owner for each stage and write a feasible follow-up schedule. Record unsuccessful attempts, time spent, and the reason for a delay. The objective is not to maximize speed at any cost; it is to learn whether the team can respond consistently without bypassing professional review.

5. Reconcile commercial outcomes to source records

At each review, match the CRM to the supplier invoice, the calendar to meetings held, engagement letters to signed mandates, invoices to fees collected, and time records to staff effort. Reconcile duplicates and refunds. Leave missing values marked unknown instead of imputing success. This is where a nominal “lead accepted” count often diverges from what partners remember informally.

6. Decide, document, and change one variable

At the review date, choose one of four actions: stop, continue the same test to a maturity date, renegotiate the data or fee contract, or run a controlled comparison against another source. Record the evidence and uncertainty. If you change offer, segment, response process, and supplier at once, the next cohort will not explain which change mattered.

The break-even calculation without a fake benchmark

For a cohort, use:

Acquisition cost = supplier and media charges + data and setup allocation + measured qualification and follow-up labor

Contribution by horizon H = fees collected by H − direct service delivery cost by H − acquisition cost

Cost per signed mandate = acquisition cost ÷ signed mandates in the cohort

If there are no signed mandates, cost per signed mandate is undefined, not zero and not infinity presented as a market fact. Show the numerator, the zero denominator, and the observation period. You may still calculate cost per eligible account or meeting, but label those as intermediate measures.

For forward planning, the firm can estimate expected value using its own historical data: probability of signing within a defined window, expected collected fees, direct delivery cost, and retained value under a stated retention assumption. Keep this model separate from realized contribution. Stress it with a downside case: lower fit, longer time to sign, more partner minutes, or a mandate that ends early. An estimate should not be described as actual ROI.

Do not import a conversion-rate benchmark from another country, service line, firm size, or provider and call it a target. Public benchmarks often use different meanings of “lead,” different sales cycles, or self-selected samples. A benchmark can suggest a question; it cannot set a credible acceptance threshold for your cabinet without a close match.

Capacity is part of the economics

A channel can produce signed mandates and still be a poor choice if those mandates arrive when the relevant team is full, require unplanned onboarding, or displace higher-fit client work. Record available capacity by the roles that actually perform the service. Count partner review, file opening, data remediation, and recurring exceptions, not just the routine delivery estimate.

Treat “no capacity” as a distinct outcome. It is not the supplier's failure if the firm accepted more opportunities than it could serve, but it is a real constraint on the channel's value to the firm. A queue of signed work that cannot start may damage both client trust and staff sustainability. Conversely, an underutilized specialist team can make a source with a moderate acquisition cost attractive for a defined period. Context changes the decision.

Write a capacity ceiling alongside the pilot spend ceiling. When the next eligible meeting would exceed the ceiling, pause new intake or route the account to a transparent waitlist. Avoid continuing to buy records that the firm cannot respond to. This simple pause rule protects service quality and makes the experiment interpretable.

Professional conduct and data checks are gates, not scoring bonuses

For French experts-comptables, Article 152 of Decree No. 2012-432 says promotion must provide useful information; proposals of services to people who did not request them must remain compatible with professional rules; and communications must be restrained, accurate, not misleading, and free of comparative elements. The rule also refers to independence, dignity, professional secrecy, and loyalty. Read the current Légifrance text of Article 152 and consult the firm's professional or legal adviser for the particular activity and channel. This summary is not legal advice.

Do not infer permission from the fact that a company or contact appears in a public register. Ask the supplier to document its source, collection date, notice, channel, and any restrictions; have the firm's authorized privacy reviewer determine what is lawful before outreach. If the source cannot document the origin or the contact's applicable preference state, hold the record rather than testing it on a prospect.

Keep contact permission and business-fit separate. A highly suitable company can still be excluded from a channel because the necessary provenance or permission is absent. A legally contactable person can still be a poor fit for the firm's capabilities. Neither gate can be “offset” by a high numerical lead score.

The buyer also needs an operational suppression process. A current objection should remain effective if a vendor resends the same person in a later file. Store only what the responsible team needs to prevent re-contact, limit access, and ensure imports are checked before activation. The exact design and retention period depend on applicable law and the firm's advice; the generic ledger above does not decide them.

What public evidence does—and does not—tell a buyer

The primary Lead-Gene guide presents a source-traceable, professional-fit approach and distinguishes a public event from buying intent. It is a commercial guide from a supplier, not an independent comparison of channel economics. It does not publish a controlled trial that would show which channel produces the best signed-mandate contribution for a particular firm. Use its checklist as a process baseline; do not treat it as proof of its own service's performance.

The CPA.com and Association for Accounting Marketing brief on CAS lead sources reports that existing clients and referrals rank strongly for CAS practice development in its survey. That is a useful counterweight to a story that every firm needs to buy more cold contacts. Its one-page brief does not provide enough visible survey-method detail to transfer those rankings as a French market estimate, and CAS in the United States is not identical to every service sold by a French cabinet. It is direction, not a conversion benchmark.

The BCG analysis of B2B marketing and sales argues for connecting lead generation, enrichment, qualification, and sales handoff rather than treating them as disconnected departmental goals. That supports the cohort design: a contact source cannot be evaluated fairly if the team receiving the record does not feed back whether it was usable. BCG's corporate client experience is not a predicted result for a small accounting firm.

An academic study of opportunity management in B2B service organizations offers a useful methodological parallel, not an accounting-firm benchmark. Its decision-support framework analyzes archival opportunities from one global on-site services provider and distinguishes the decision to pursue an opportunity from the probability of winning after a bid. The authors explicitly discuss selection limitations and omitted variables such as bidding costs and competition. That distinction is why this ledger counts both all eligible records and later-stage outcomes, preserves rejected and unpursued rows, and does not treat a conversion rate among contacted leads as the whole channel result. The study's setting, data, and models do not establish what a French accounting practice should expect.

The U.S. Small Business Administration's market-research guide recommends testing demand, market size, pricing, and competition with information suited to the target market. That is a useful checklist for framing a pilot, but its U.S. small-business context is not evidence about French accounting firms. Use it to formulate questions; answer those questions with the firm's own eligible prospects and reconciled outcomes.

The Insee 2025 business-creation report records 1,165,800 business creations in France in 2025, a national administrative-statistics total covering multiple legal forms and sectors. This is market context, not a count of firms seeking a new accountant. The population includes micro-entrepreneurs and many businesses may not be eligible for the service under test. Do not multiply that total by a guessed conversion rate to manufacture a total addressable market.

A public forum discussion from an accounting-firm founder describes purchased databases that did not produce results and were perceived as outdated. Treat it as one dated user report, not prevalence evidence. It does, however, make record freshness, verification date, and duplicate handling sensible questions in a supplier contract.

We also reviewed publicly indexed video and webinar listings about marketing for accounting firms and B2B measurement. For example, the description of the “Stop Chasing Clients” accounting-firm marketing webinar frames the discussion around positioning and client fit. The public listing exposed a description rather than a transcript or underlying dataset, so it was treated as a practitioner-question signal, not as performance evidence. A training agenda can reveal common questions; it cannot substantiate a conversion claim. The pilot therefore relies on source documents and on the firm's own reconciled data, not on inaccessible video claims.

Decision matrix: what should happen after the pilot?

Evidence at review	Appropriate next action	Avoid
Provenance or contact gate fails	Stop activation; request records and remedies; involve the responsible reviewer	Sending a “small test” to see whether anyone complains
Records are valid but service fit is poor	Stop or redefine the segment and service offer	Raising volume while leaving the fit rule unchanged
Fit is good but response and handoff are inconsistent	Fix ownership and workflow, then observe a new cohort	Blaming the source before the firm's process is instrumented
Meetings occur but mandates do not	Review need, authority, scope, fee clarity, timing, and meeting quality	Optimizing for booked meetings alone
Mandates sign but delivery work overwhelms capacity	Cap intake, adjust service boundaries, or pause the channel	Counting signed value without start capacity or direct cost
Mature cohort has positive contribution under the declared horizon	Consider a limited next cohort while preserving the same ledger	Extrapolating one small cohort into a guaranteed return
Results remain mixed or immature	Extend the observation date or run a controlled comparison	Calling inconclusive data a win or failure

“Positive contribution” in this table means positive under the firm's defined horizon and cost allocation, not proof of long-term profitability. The partner group should decide what level of evidence and risk is acceptable for its own budget before reviewing the result.

Common failure modes

Optimizing the supplier's definition of a lead

If the contract charges for an email address, the supplier can meet its delivery target even when the address is stale or the account is unsuitable. Agree to an eligibility and duplicate rule before signing. Keep chargeable delivery, verified account, qualified conversation, and mandate as distinct units.

Quietly changing exclusions after seeing the data

Removing inconvenient records from the denominator makes the conversion rate look better. Preserve every original row and append an exclusion event with reason, date, and owner. Report both delivered count and eligible count. Exclusions are decisions to explain, not hidden data cleaning.

Comparing young and mature cohorts

If one source is evaluated after a short interval and another after a much longer interval, the reported meeting and signing rates are not comparable. Use a shared age window and display immature outcomes separately. When the firm's buying cycle is longer than the pilot window, the correct status is “not yet mature.”

Confusing association with incrementality

An account may already have been referred or searching for the firm before the paid lead vendor's touch. A source field cannot prove the channel caused the mandate. Preserve the recorded sequence and avoid claiming incremental revenue unless the design actually supports that conclusion. A simple comparison group may help, but matching and randomization have their own limitations and should be planned with someone who understands the data.

Counting staff time as free

Partner time is often the scarcest acquisition input. If the spreadsheet records vendor charges but omits repeated calls, data repair, or onboarding, it understates the pilot cost. Use a consistent time valuation and show a sensitivity range if the true opportunity cost is uncertain. Do not present the chosen hourly value as an objective market wage.

Stopping too early or continuing indefinitely

A first response can arrive before a mandate is signed, and a mature opportunity can take longer than the initial review. Set an end date, a later maturation check, and a loss cap before launch. A vendor should not be allowed to turn every inconclusive result into an open-ended extension with fresh fees.

Mixing different offers in one funnel

Payroll setup, annual accounts, tax planning, and outsourced finance may have different qualification questions, fee structures, and staffing constraints. Either separate cohorts or report them separately. A blended average can hide an offer that consistently attracts unsuitable enquiries.

Treating the law as a marketing checkbox

Passing an internal compliance field does not make an inaccurate or intrusive message appropriate. Review the actual source, offer, recipient function, channel, and message. Article 152 is a professional-conduct rule; privacy and electronic-marketing requirements may add separate conditions. Ask a qualified adviser when uncertain.

The review meeting that turns a pilot into learning

Invite the partner who accepted the opportunities, the staff member who worked the records, the person who reconciles the CRM, and the supplier owner if needed. Review a sample of individual records before reviewing aggregate rates. For each record, compare the source fact, the initial hypothesis, the message or enquiry, the partner's fit decision, time consumed, stage reached, and final reason.

Ask in this order: Were the records what the contract said they would be? Were they relevant to the approved service? Could the team legally and professionally handle them? Did the team follow the process it agreed to? Did the source produce activity that progressed? Did the resulting signed work create contribution under the selected horizon? Each answer points to a different owner. This order prevents a conversation about “lead quality” from becoming a blame exchange between marketing and partners.

Change one main variable for the next cohort. If the records are fresh but too broad, tighten the segment. If the segment is right but nobody follows up, fix ownership. If meetings occur but mandates do not, review qualification and proposal fit. If mandates sign but margins are weak, revisit service scope and pricing. A single targeted adjustment makes the next comparison more informative than replacing the entire funnel at once.

flowchart TD
  A[Supplier or channel record arrives] --> B{Source, identity, and contact gate pass?}
  B -->|No| X[Hold or suppress; record reason]
  B -->|Yes| C{Service fit and capacity pass?}
  C -->|No| Y[Reject or refer; keep reason and effort]
  C -->|Yes| D[Create cohort row and assign owner]
  D --> E[Record dated contact and stage events]
  E --> F{Meeting held?}
  F -->|No| G[Record delay, no reply, or objection]
  F -->|Yes| H{Opportunity accepted?}
  H -->|No| I[Record fit, scope, timing, or fee reason]
  H -->|Yes| J[Track signed mandate and service start]
  J --> K[Reconcile fees, delivery cost, and staff time]
  G --> L[Review cohort at the agreed maturity date]
  I --> L
  K --> L
  L --> M{Evidence supports next step?}
  M -->|No or immature| N[Stop, wait, or improve measurement]
  M -->|Yes, within risk cap| O[Authorize one bounded next cohort]

A compact executable ledger template

This JSON is a schema example, not a customer record or software that contacts prospects. Store real identifiers only in the approved CRM; use synthetic records in demonstrations. Dates should be ISO-8601 values with an explicit timezone. Keep a dated event list so an audit can reconstruct how the account moved through the process.

{
  "cohort_id": "2026Q4-source-offer-intake-window",
  "lead_id": "internal-random-key",
  "source": {
    "name": "supplier-or-channel",
    "record_url": "internal-record-reference",
    "collected_at": "2026-10-24T09:00:00+02:00",
    "verified_at": null
  },
  "offer": "one-defined-accounting-service",
  "fit": { "status": "review", "reason": null, "owner": null },
  "contact_gate": {
    "channel": null,
    "provenance_reviewed": false,
    "exclusion_checked": false,
    "decision": "hold"
  },
  "stage_events": [],
  "minutes_by_role": { "partner": 0, "assistant": 0, "specialist": 0 },
  "direct_cost_eur": 0,
  "mandate": { "signed_at": null, "fees_collected_eur": 0 },
  "outcome_reason": "not-yet-reviewed"
}

The example.invalid URL is intentionally non-routable. Replace it only in a controlled internal system with a source URL the firm's policy permits storing. The schema's boolean fields are not a legal conclusion; they are reminders to route decisions to the authorized person.

Practical checklist before authorizing the next euro

• [ ] The firm has approved one service, one segment, and a written exclusion boundary.
• [ ] The supplier has documented origin, freshness, exclusivity, duplicate handling, charge rules, and export fields.
• [ ] The firm's privacy or compliance owner has reviewed provenance and intended contact channels before activation.
• [ ] Every record has a cohort ID, source, intake date, offer, owner, and stage history.
• [ ] Delivered, unique, eligible, accepted, held-meeting, signed, started, and collected outcomes are separate counts.
• [ ] Qualification, cleanup, follow-up, partner meeting, and rework time are included under the same costing rule.
• [ ] Cohorts are compared at a shared age and within comparable services; immature outcomes are labeled.
• [ ] The fee horizon and direct-service-cost method are stated beside any contribution calculation.
• [ ] No external benchmark is treated as a target unless the definitions, service, market, and sample match.
• [ ] The pilot has a spend cap, capacity cap, stop rules, review date, and owner for the final decision.
• [ ] The review records what was observed, what is inferred, what remains unknown, and the one variable to change next.

Bottom line

A lead-generation purchase is not a decision about how many names to acquire. It is a decision about whether a source can produce relevant, lawfully and professionally reachable opportunities that the firm can serve at a contribution and workload it accepts. Price per contact is one input in that decision, not its conclusion.

Start with the source's evidence, define the cohort before intake, reconcile the CRM to real meetings and signed mandates, and count the effort needed to create and deliver the work. If the data is too immature or the sample is too small, say so and choose whether the next step is to wait, improve measurement, or stop. That answer may feel less decisive than a vendor's headline conversion rate; it is also much easier to defend to partners, staff, and the prospects whose information the firm handles.

Sources and method limits

Research cutoff: 24 October 2026. The analysis read the Lead-Gene source page, the current text of Article 152 on Légifrance, an Insee publication, BCG's B2B sales-and-marketing discussion, a B2B service-opportunity research article, the SBA market-research guide, EDPB legitimate-interest guidance, CPA.com's CAS lead-source brief, France Num CRM guidance, one public forum discussion, and publicly indexed video/webinar descriptions. Search Console selection used the exact query “generation de leads expert comptable,” 9 impressions, average position 13.1, in the 14–21 October 2026 window. No campaign, CRM, vendor dataset, or accounting-firm interviews were available for this article. The forum evidence is one anecdotal post; the CPA.com summary does not expose enough survey-method detail to transfer its rankings; BCG's examples concern broader B2B organisations; the Insee business-creation total is not a measure of accounting demand. The academic study concerns one global services firm, its historical opportunities, and its own model; it is not an accounting-firm conversion estimate. SBA guidance is US-oriented and used only to frame research questions. EDPB guidance is not individualized legal advice. Video descriptions without accessible transcripts were excluded from factual claims. The cohort schema and decision method are original design recommendations, not validated benchmark results.

For the original synthetic calculation above, the method was arithmetic applied to a constructed example of 12 fictional records, 4 fictional eligible records, 2 fictional held meetings, and 1 fictional signed mandate. The values were created solely to illustrate formulas; the sample contains no real businesses, was not collected on a date range, and cannot support an estimate, benchmark, or conclusion about any channel. All figures in that example are excluded from empirical findings.