Google Ads alerts for freelancers: know what needs a closer look

Define the freelancer's monitoring job

An independent Google Ads consultant usually divides the day between account work, client communication and the business of being self-employed. There is rarely a separate person watching every campaign while the consultant writes a proposal or joins a meeting.

Alerts can make that arrangement more manageable by surfacing an explicitly configured condition. Their purpose is to tell you that a selected observation deserves attention. They do not establish its cause, decide the client's commercial priorities or provide a substitute for a review routine.

Start by identifying what you could do differently if a signal arrived. Would you inspect spend, confirm that the selected conversion definition still applies, ask the client about a stock change or check a campaign's current status? An alert without a plausible next step becomes another interruption.

The practical aim is selective visibility. A freelancer does not need a notification for every movement that a full report could show. Choose the observations that belong in the agreed service and that you can investigate with the available account context.

Keep monitoring separate from optimisation. A triggered condition can lead to an account read, a client question or a documented decision to wait for more information. It should not force a live edit simply because the account differs from yesterday.

Agree what the service includes

Clients can hear “monitoring” as a promise that someone will notice and solve every problem immediately. State what the arrangement actually covers: the configured signals, the times you review them and the process for decisions needing client input.

A scheduled evaluation is not continuous human attendance. If you inspect notifications during working hours, describe that routine accurately. Faster checks do not create availability during meetings, travel or a planned absence.

Clarify what facts the client must provide. A consultant may not know that a product sold out, a phone number changed or a form was deliberately removed. Account data can show a movement without explaining the business event behind it.

Agree the authority required for changes. A monitoring arrangement does not necessarily authorise a budget increase or a campaign pause. ppctor's conversational edits require a separate proposal and confirmation by the authorised connected owner in the relevant context.

For retained records, write the selected condition and intended response in everyday language. “Review yesterday's account cost when it is at or above the agreed amount” is an understandable service task. “Ensure the account never overspends” would describe a different capability.

Revisit this agreement when the account's objectives change. A new product line or lead definition can make an old threshold irrelevant even though the saved alert continues to evaluate exactly as configured.

Choose a signal with a next step

A useful signal starts with a question you would otherwise ask manually. Perhaps a client wants you to notice unusually high recorded cost, a low conversion count over a completed period or the absence of enabled campaigns. The next step differs in each case.

Choose a supported metric and target rather than compressing several business concerns into one vague label. “Poor performance” has no single account value. Cost per conversion, total cost and conversion value describe different observations.

Do not use a threshold to imply causation. A higher cost per conversion might reflect lower conversion counts, higher costs or both. The initial notification should lead to examining the relevant inputs, not to an automatic diagnosis that bidding is wrong.

Write the follow-up beside the alert decision. For a cost signal, inspect the campaigns contributing to the period's cost. For a conversion signal, inspect the selected conversion population and the relevant business context before assuming that demand stopped.

Keep the initial configuration focused. Several alerts for nearly the same measure can produce multiple notifications about one underlying observation. Review the intended decisions before adding another overlapping condition.

Use a scheduled report for a routine overview that does not need a threshold. Use an alert when the change of condition is what should interrupt your normal work. Those tools serve different review habits.

Understand the measured outcome

An alert's metric must be understood before its threshold can be meaningful. Cost per conversion depends on the conversion count selected by the configured measure. A client may instead care about a named lead action or completed purchases.

The numeric Ads alert catalogue uses its declared provider metric definitions. A conversational follow-up that selects one named conversion action can use a narrower population. Do not describe that follow-up as reproducing the original alert when the definition changed.

For business reporting, ppctor can distinguish primary conversions, primary purchase conversions, primary leads and named actions. Secondary conversions are excluded from the default primary aggregates. Those distinctions do not mean every saved alert silently adopts a newly discussed business definition.

Ask what the client actually regards as success. A submitted form can differ from a sale, and a lead event name does not establish whether it represents the right stage for that site. Keep the client's chosen interpretation with the reporting question.

If a follow-up compares Ads leads with GA4, the corresponding GA4 event must be selected or confirmed. A plausible event name can help a conversation, but it is not evidence that the event measures the client's intended lead outcome.

Review value definitions as carefully as counts. Purchase value and all-primary conversion value can differ when non-purchase actions carry values. Label the measure used rather than present any value-based ratio as if it were automatically sales revenue.

Write the threshold in plain language

The comparison is part of the alert, not a minor display detail. Above fifty and at or above fifty give different results when the value is exactly fifty. Use the supported comparison that expresses the intended boundary.

The shared numeric comparisons include above, at or above, below and at or below. Choose from the configured catalogue. Do not assume an arbitrary formula or an exact-equality condition is available merely because it can be expressed in a chat message.

Keep the unit beside the threshold. A cost amount needs its account currency; a percentage and a multiple need their own meaning. An isolated number is a poor basis for a client agreement or a later investigation.

Choose the threshold from the account's context and the decision you intend to make. A figure appropriate for a high-volume retailer may produce constant interruptions for a small service business. There is no universal freelancer threshold that settles every client's tolerance.

A threshold change should have a reason. If the client's acceptable acquisition cost changes, record that decision and review the configured condition. Quietly raising a boundary because notifications feel inconvenient can hide a problem rather than improve the monitoring process.

Distinguish a decision boundary from a statistical model. A fixed numerical alert asks whether the configured comparison is true. It does not by itself estimate the probability that the account has an unusual underlying event.

Choose the observation window

The window defines what activity the value covers. The numeric Ads catalogue includes today so far, yesterday, the last seven days and the last thirty days. These answer different questions and should not be used interchangeably.

Today so far is incomplete. A low conversion count early in the day can be normal for the account. Yesterday concerns a completed calendar day, but some outcomes may still arrive later. Longer windows retain older activity that can continue influencing the value.

For a consultant working in a different time zone from the client account, preserve the actual date boundaries. Your own afternoon may be early morning in the account's reporting day. Comparing mismatched days can create an apparent contradiction.

Match the investigation to the configured window before adding new views. A seven-day condition and a yesterday-only report can both be correct. The latter is a diagnostic slice, not proof that the earlier signal was false.

Consider the account's normal rhythm when choosing the window. If leads occur mainly on working days, a daily threshold can behave differently at weekends. That context belongs in the review process rather than being silently inferred by the alert.

Keep a written question that includes the window. It will make later client explanations easier and help you reproduce the relevant observation without relying on the loose meaning of words such as “recent.”

Match checks to your response routine

Check cadence tells the system when to evaluate; it does not change the measurement window. An hourly evaluation can inspect a thirty-day value. A daily evaluation can inspect yesterday. Choose the two settings independently.

The current numeric alert options are hourly, every six hours, every twelve hours and daily. They provide scheduled checks rather than a continuously updated live feed. Choose an interval that fits the signal and your actual response arrangement.

If the next useful review happens during a daily account session, the fastest cadence may add little value. If a client has a time-sensitive promotion and a defined response process, a more frequent supported check can fit that context.

Do not equate frequent checks with frequent fresh provider information. Several evaluations may see the same available result. Notification behaviour also depends on whether the configured condition changed state.

Consider your interruption budget alongside account coverage. A solo consultant needs enough uninterrupted time to investigate and implement work. The monitoring plan should help choose that work rather than repeatedly displace it.

Review cadence after the temporary reason ends. A promotion-specific configuration should not become the permanent default through neglect. Keep the reason for a faster interval with the client record so the later adjustment is deliberate.

Account for delayed and revised data

Google's data freshness guidance explains that account statistics are not immediately complete, processing varies by measure and later adjustments can change reported values. A check observes the data available at its evaluation time.

This matters especially when the freelancer is tempted to react quickly. A low recent conversion count can justify an investigation, but it should not immediately become a client statement that sales stopped or tracking failed.

Preserve the observed time and window. If a later report differs, inspect whether the period, conversion population and attribution basis match before diagnosing an error. Later-arriving information can change an otherwise comparable observation.

A provider read failure is different from delayed results that were successfully returned. A failed evaluation must not be described as a normal healthy check. Equally, a successfully read recent total should not be treated as a final financial reconciliation.

Use completion and timing language appropriate to the evidence. “The available value crossed the configured boundary” is supportable. “The final cost is settled” or “every conversion has arrived” would require a different basis.

Choose a follow-up interval appropriate to the business question rather than refreshing repeatedly until the number feels reassuring. Additional reads are useful when they can resolve a defined uncertainty, not simply because the first result was unwelcome.

Keep missing values distinct from zero

A ratio can be undefined even when some account data exists. Cost per conversion cannot be meaningfully calculated when there are no conversions in the selected window. Treating that situation as zero would create a misleading low-cost signal.

The current numeric decision logic checks whether the metric's required inputs exist. When they do not, the evaluation is skipped and the previous state is retained. It does not announce recovery based on a value that was never measured.

That behaviour is important for small client accounts. A period without conversions can happen during normal low-volume operation. It also can deserve investigation. Neither interpretation is improved by reporting an undefined acquisition cost as a measured zero.

Use a count-based condition when the specific question is whether the recorded conversion count is low. That asks a different question from a ratio boundary. Select the supported measure that matches the concern.

A failed account read also must not be converted into zero. The evaluation records failure separately from a valid check. A consultant should inspect the reported status when the expected result is unavailable rather than assume no activity occurred.

In client communication, state the limitation plainly: the ratio could not be measured for that window, or the provider read did not complete. Those explanations are more accurate than saying the account is within range.

Understand firing and recovery

The numeric alert design notifies on a state transition. When the condition first becomes true, it fires. If it remains true at later checks, the system does not create a fresh firing message at every interval.

When a valid later value moves back within the configured range, a recovery can be reported. The notification describes that condition's return to range. It does not prove that the cause was fixed or that every related business issue disappeared.

This behaviour protects a freelancer's attention. An unresolved high cost per conversion can persist for several checks without flooding the channel with the same notification. The condition still needs an owner in your review routine.

Do not interpret silence as recovery. The alert may still be firing, may be paused, may have insufficient data or may have encountered an evaluation error. Inspect its current configuration and check status when the distinction matters.

A skipped ratio evaluation retains the previous state. If the condition had fired, the absence of required inputs does not produce a fabricated recovery. A later valid observation is needed to determine the appropriate state.

Keep operational incident reporting distinct from a client's performance alert. An application error and a measured threshold transition are separate kinds of events. They should not be explained to the client as if both meant campaign performance crossed a boundary.

Keep each client's scope separate

A freelancer can manage several clients without a large agency structure. The same need for clear account separation still applies. A useful notification identifies the connected account context and the target that was evaluated.

An account-level condition aggregates its defined account scope. A campaign-level condition concerns the chosen campaign. The two can differ even during the same period, and a campaign observation should not be presented as the entire client's result.

Use the named target in your notes and follow-up requests. A phrase such as “that campaign” is helpful only when the conversation's context resolves it confidently. If several candidates fit, establish the actual one before preparing a change.

Do not reuse one client's numerical boundary merely because another account has a similar name or industry. Offers, volume, currency and conversion definitions can differ. A consistent review method should preserve those differences.

Keep receiving destinations appropriate to the intended audience. A client channel may be suitable for a settled performance signal, while the freelancer's internal analysis requires additional context. Delivery choices do not change the metric's definition.

When an account connection changes, revisit its alerts and interpretation. Do not assume an old campaign selection or client agreement remains applicable to a newly connected account.

Plan an invented client alert

The following example is invented. An independent consultant works with Meadow Service, a fictional repair business. The client wants a review when yesterday's account cost is at or above 120 in the account's currency.

The consultant writes the condition as cost, at or above 120, over yesterday, evaluated daily. The boundary includes exactly 120. The planned response is to inspect campaign contributions and confirm whether the client had a planned promotion.

A valid observation of 128 first moves the condition into firing state. The consultant checks the selected date and finds that the increase is concentrated in the promotion campaign. This observation does not independently establish whether the spend was commercially appropriate.

The client confirms that the promotion was planned. The consultant documents the context and leaves the account unchanged while the agreed review continues. A triggered alert does not require a corrective edit when the observed activity fits the client's plan.

If the next valid observation is 130, the condition remains firing without a new firing notification. If a later valid value is 115, it returns within range. Neither state transition proves the promotion succeeded or failed.

This example separates the measured condition, the investigation and the commercial decision. It also demonstrates why the freelancer needs a response routine even when repeat notifications are intentionally suppressed.

Investigate before choosing an action

Begin with the alert's own definition. Identify the account, target, metric, comparison, threshold and observation window. Confirm the current check status before deciding that the message describes a fresh performance change.

Then read the relevant inputs. A cost-per-conversion signal benefits from understanding both cost and the counted conversions. A high total cost signal benefits from campaign contributions. The diagnostic question should follow the actual condition.

Look for known business context before changing settings. A sale, service interruption or product availability change can explain a movement that account data alone cannot. Ask the client the focused question that could affect the decision.

Keep scope changes explicit during investigation. A narrower campaign view or named-action report can be useful, but it no longer reproduces an account-level provider metric exactly. Label that view as additional evidence.

Decide whether the next step is further reading, a client clarification, a supported edit or a documented review later. These are all valid outcomes. An alert is not an instruction to manufacture an intervention.

Record what remains unresolved. A solo consultant returning after another meeting should see the pending question, not just an earlier notification. This makes the workflow resilient to normal interruptions in freelance work.

Read cost signals carefully

Recorded account cost can show where attention is needed, but it is not an agreed-budget ledger by itself. The alert compares its selected value with a fixed boundary. It does not project the whole period or enforce a spending ceiling.

A cost condition for yesterday may identify a day worth reviewing. A budget-pacing question asks how the current period's spending and recent pace relate to a configured target. Use the appropriate tool rather than make one signal answer both questions.

The budget-pacing guide explains that separate calculation and its assumptions. Its projection is conditional on the recent observed pace. It is not a guarantee of future spending or the final bill.

When investigating a cost alert, first identify the campaigns contributing to the selected period. Check whether the client approved a promotion or amended its plan. A higher total can be intentional and still require clear communication.

Do not equate an account target with every campaign's live budget setting. Several campaign settings can contribute to the same account cost. Any proposed budget edit needs its own supported target and clearly expressed amount or percentage.

If the client asks for a spending change, preserve the instruction and confirm the resulting proposal. The alert did not itself authorise that adjustment. The final receipt should say what was checked, while later pacing remains a separate observation.

Avoid using one busy day as a complete diagnosis

A single day can deserve immediate review without establishing a trend. Compare suitable periods if the decision requires a pattern, and retain the original alert window in the explanation. A later longer-period report should not erase why the initial signal appeared.

If the client supplied an account-wide commercial ceiling, keep it as the agreed reference. Do not imply that setting an alert at that amount prevents spending beyond it. Monitoring and enforcement have different behaviour.

Investigate conversion signals

For a conversion signal, start by confirming the counted outcome. A total provider conversion count and a named contact-form action may move differently. They can both be correct because they measure different populations.

Separate counts from value. A primary call conversion can carry value; a purchase count and its purchase value are distinct inputs. An account's all-primary value may include non-purchase actions and should not automatically be labelled product sales.

Check whether the chosen event or conversion action changed during the relevant period. Configuration metadata can show the current role or name, but that does not independently prove the client's implementation operated correctly throughout the observation window.

A GA4 comparison needs an explicit event when the lead outcome is site-specific. Ask the client to confirm the corresponding event. Similar-looking names can suggest candidates but do not settle the business meaning.

For very low-volume accounts, inspect the count and timing before overinterpreting a ratio. A meaningful commercial concern can exist even when cost per conversion cannot be measured. Describe the available evidence instead of filling the missing ratio with zero.

Do not promise that a reporting tool has tested the form, phone route or checkout. Those functions require appropriate operational checks. Account statistics can inform the question while the client or relevant specialist verifies the actual customer path.

Separate account and connected-service checks

Not every supported alert concerns a numeric Ads performance value. Account-status and enabled-campaign checks address current account conditions. Merchant Center and Analytics checks use their own defined source and target context.

An absence-of-enabled-campaigns signal is not the same as a zero-spend signal. Enabled status does not guarantee delivery, and a campaign may be deliberately paused. Preserve the precise observation rather than describe both as an account outage.

For Merchant Center, choose the supported health check with its intended country and channel scope. That condition concerns the returned Merchant evidence. It should not be presented as a universal audit of every policy or product issue.

An Analytics alert depends on the connected property and selected segment or event context. It cannot supply another site's data merely because the freelancer asks for a similar measure. Establish the appropriate connection before interpreting its result.

Read each tool's declared window and condition. A current-state check differs from a recent performance aggregate. Their notification text can look equally urgent while representing different kinds of evidence.

For a freelancer, these distinctions help assign the next task. A Merchant issue may need the feed owner, an Analytics concern may need the site's measurement specialist, and an Ads budget question may be within your direct account-management remit.

Reduce interruptions without hiding problems

Alert quality depends on the decision it supports. Adding more conditions does not necessarily improve coverage. Review whether each configured signal has a distinct concern, relevant threshold and practical response.

Remove redundancy by understanding overlap. An account cost signal and several campaign cost signals can all refer to the same spending movement. Keep the separate ones when their targets lead to separate decisions, not merely because they are available.

Use transition behaviour as intended. A condition that remains firing needs an unresolved-item review routine, not repeat notifications on every check. Keep your pending investigations visible in the way you manage work.

Do not solve inconvenience by setting a boundary so distant that it no longer represents the client's concern. If the existing signal is unsuitable, revise its measure, scope or window with a documented reason.

Pausing an alert stops its active monitoring arrangement; it does not fix the account condition. Record why it was paused and whether a replacement check is needed. The client should not confuse a quieter channel with improved account health.

Review notifications alongside actual response capacity. A good freelance routine protects time for investigation. Selective signals, suitable cadence and clearly assigned next steps make that possible without pretending that the account is continuously supervised.

Configure the alert through Slack

Use the alert workflow in a channel connected to the intended Google Ads account. The current Slack command family includes /ppctor alert, /ppctor alert list and /ppctor alert create. Creation uses the configuration modal.

The modal keeps the selected metric, comparison, threshold, target, window and cadence within the supported configuration. A conversational request can help prepare that configuration, but it should not become a separate improvised writer with different validation rules.

Before saving, read the completed condition as a sentence. Confirm the account and target, the boundary's inclusion or exclusion of equality, and the intended observation period. This review is easier than discovering a wrong unit after the first notification.

Set the receiving choice appropriate to the agreed workflow. Delivery configuration determines who sees the observation; it does not change how the value is calculated. Consider whether the client needs the raw signal or a later explanation from you.

If Slack cannot open the configuration view, use the provided application path rather than assume creation happened. A submitted configuration and an unavailable modal are different outcomes. Check the saved alert list when confirmation is needed.

Use the list to review enabled and paused alerts for the connected account. A saved alert's state should be understood alongside its settings and check history, particularly when a freelancer returns to an account after several days of other work.

Use conversation for the follow-up

After a notification, ask a focused question tied to its observation. “Show campaign cost for yesterday for this account” is a suitable starting point for an account-cost signal. Add the selected metric definition when outcomes are involved.

The assistant can discover available reporting fields and supported settings operations from local catalogues. Actual account data still needs the relevant read. Capability discovery alone is not evidence of the client's current numbers.

Ask for a table when campaign contributions are easier to inspect in rows. Ask for a chart when a trend is useful. The framework formats returned data; it should not manufacture values or treat an incomplete preview as the entire account.

If you change the question during the follow-up, state the changed scope or population. “Now show only primary purchase conversions” narrows the outcome. It should not silently rewrite what the earlier alert measured.

Creative help can remain conversational. A draft explanation to the client or a clearer wording for a review note does not require another provider query unless it introduces new factual account claims.

Keep results and interpretations separate. The assistant can explain the returned observation and suggest relevant investigative questions. Your client-specific facts and decision authority remain necessary before a supported live change is prepared.

Preserve the useful evidence when changing format

Requesting a clearer table for the same returned result should use that result rather than repeat an identical account read. A new period or metric may require additional data. The distinction matters when you want a quick explanation rather than another investigation.

If the result cannot answer the question, name what is missing. Do not repeatedly ask the same unchanged request after a non-retryable limitation. Resolve the required connection, target or supported field first.

Confirm a separately justified edit

An investigation may establish that a live change is appropriate. Ask for the supported action with a clear target and desired value. The proposal should name the entity and its parents where those are needed to identify the effect.

Review each logical step separately. Adding keywords and enabling their parent ad group are different actions. A readable proposal should disclose both when both were requested, rather than burying the second inside a technical summary.

Only the authorised connected owner in the proposal context can confirm the conversational edit. A freelancer who can discuss an account is not automatically authorised to execute every proposal. Arrange the appropriate handoff for the client connection.

A pending proposal expires after fifteen minutes. The relevant settings are checked again before execution. If another person changed them, the process can stop rather than apply the older instruction against a conflicting current state.

Read the final status carefully. Verified execution confirms the checked result, not later business performance. If execution is uncertain, the change may already exist; inspect the account before creating another potentially duplicating request.

For the full conversational sequence, see confirming and executing Google Ads edits. Alerts provide an observation; that separate workflow handles the proposed mutation and its result.

Explain the observation to the client

A useful client note says what was observed, the period it covers and what you checked next. Avoid presenting an alert as proof of a cause. The client needs the significance of the observation and the decision still required.

For example, a fictional explanation might say that yesterday's cost crossed the agreed review boundary, the increase was concentrated in the promotion campaign and the client confirmed the planned offer. It should not invent a return on the promotion.

When data is unavailable, say what could not be measured. A missing ratio and a failed provider read are different limitations. Do not reassure the client that the account is healthy merely because no new firing message appeared.

For a proposed change, separate the recommendation from its current state. “Prepared for confirmation” means nothing has been applied yet. “Executed and verified” concerns the checked action. Later results require a suitable performance review.

Use human-readable targets and relevant numbers, not internal account identifiers or debugging detail. The client should understand which campaign or asset is involved without decoding a technical resource name.

End with a concrete next step when one exists: confirm the business fact, review the pending proposal or inspect the later defined period. A concise note is useful because it preserves the decision, not because it hides uncertainty.

Plan absence and service handover

Freelancers need a continuity plan that reflects real availability. Before a planned absence, explain who will review important signals and what authority that person has. Saved alerts do not provide a substitute for that arrangement.

Keep the relevant account connection and proposal approval controls in mind. A substitute consultant may be able to review client evidence without being the authorised connected actor for execution. Do not promise that any colleague can use the same confirmation button.

Provide the configured conditions, unresolved investigations and client context in the handover. A list of alert names alone is insufficient when the recipient needs to know why a boundary exists and what response was agreed.

If monitoring is paused as part of a service change, state that decision explicitly. It should not be described as resolving a firing condition. Confirm which review responsibilities continue elsewhere.

On return, review current state rather than infer the account's history from notification volume. Transition-based delivery can remain quiet while a condition stays firing. Check the relevant saved configuration and unresolved work.

When a client engagement ends, review account and receiving arrangements through the appropriate management workflow. Do not leave the client assuming an independent consultant continues a monitoring service that is no longer agreed.

Review the configuration deliberately

Alert maintenance is part of the service because the business context changes. Review whether each condition still represents a concern the client wants you to investigate, and whether its metric and selected target remain appropriate.

Revisit thresholds after a confirmed change in acceptable acquisition cost, account budget or offer. Preserve the reason for the change. The updated condition should reflect a new decision rather than an attempt to make the channel quieter.

Check windows and cadence after temporary campaigns end. A daily observation suitable during a promotion can be unsuitable for the client's normal low-volume cycle. Neither configuration is universally correct.

Review overlap before adding a new signal. Ask whether it introduces a distinct decision or merely duplicates an existing account concern under another name. A focused set is easier to manage and explain.

Keep unresolved provider or connection problems visible. A condition that could not be evaluated is not the same as a valid within-range result. Resolve the cause before relying on that check as part of the agreed coverage.

Measure the workflow by useful decisions and clear communication, not by the number of notifications. A well-understood signal can support a disciplined freelance service even when no edit follows.

Freelancer alert questions

Does an hourly alert notify every hour?

Not for an unchanged numeric condition. It evaluates at the configured cadence and notifies when the condition moves into firing or returns within range. Keep unresolved firing conditions in your own review routine.

Does silence mean the account recovered?

No. The condition may remain firing, be paused, lack required inputs or encounter an evaluation failure. Review current settings and check status when that distinction affects the next decision.

Can an alert automatically pause the campaign?

The alert itself reports its configured observation. A conversational live edit requires a separate supported proposal and confirmation. Configured Rules are another workflow with their own actions, mode and safeguards.

What happens to cost per conversion when there are no conversions?

The ratio is not measured as zero. The numeric decision logic skips a metric without its required inputs and retains the previous state. A separate count-based condition can address a low recorded count.

Why can a later report show a different value?

First compare the account, dates, population and measure. Provider processing and later adjustments can also change comparable recent data. The notification records an observation at its evaluation time.

Can the same thresholds be copied to every client?

A consistent review method is useful, but the boundaries need each client's context. Currency, outcomes, volume and response agreements can differ. Keep the reason for each saved condition with that client's work.